Back on the longevity thread, I proposed (without really knowing if or how it was possible) how we could just replace organs with young cells piecemeal.
Does anyone know what happens to the barrels of tanks (armored vehicles, not containers) after they wear out? Can old barrels be restored and reused, or do they have to go to the junkyard?
Let's say I get a positive covid test today. (I don't, but I want to pre-plan.)
What's the current best practice of what to take? Paxlovid is still illegal, but what could I ask my doctor to prescribe off-label, or what could I take OTC, to give me the best outcomes?
The author's Twitter handle is @plain_fiction, which is telling, but do we have any physicists here that want to explain what a "warp bubble" is supposed to be, and why this thing featured on Slashdot is surely nonsense? Anyway, in addition to the usual prior against such trings, Harold White and his team previously failed to notice that the physics-defying EM drive they tested didn't actually work, suggesting, I suppose, a lack of rigor on their part (or worse).
Yeah, this isn't the place I'd go for reliable information on "warp bubbles", but it is an interesting read.
Miguel Alcubierre's theoretical "warp drive" is fairly well regarded by the physics community, at least to the extent of not being obvious nonsense and worth talking about. The math is pretty hairy, and I've never tried to sit down and work my way through it, but:
If the math is right, the general relativity theoretically allows for the creation of a "bubble" of warped space-time with a volume of normal space in the center but the space ahead of the bubble being compressed and the space behind the bubble being expanded - and this is a dynamic effect, so the bubble progressively "moves" forward despite not having a velocity. If this can be done in a stable and controlllable matter, voila, space travel without pesky fuel requirements and speed limits. And maybe you can build a time machine as well, or maybe you'll blow yourself up or erase yourself from history if you do because the theory and math of closed timelike loops gets really messy.
BUT: to do this, you absolutely need for part of your warped space-time to be made out of a structure with a negative energy density, and negative energy densities don't seem to be a thing (no, antimatter doesn't count). So, nice theory, impossible practice.
BUT BUT, there is a quantum-mechanical phenomenon called the Casimir Effect which creates what appears to be a negative energy density in very small volumes by suppressing quantum vacuum fluctuations. Maybe we can build a "warp drive" that way?
BUT BUT BUT, we don't know whether "negative energy density" in quantum mechanics means the same thing as "negative energy density" in general relativity. We'd need a theory of quantum gravity for that, and we haven't got one.
So now White thinks he's done the math and figured out how to build a microscopic testbed for this. As you note, he's not always 100% on this sort of thing, so it's quite possible he got the math wrong. And the necessary microscale engineering might be impractical as well. And if it *does* work, it's probably not going to scale very well.
Even so, a demonstration of an impractically microscopic warp drive would be theoretically valuable, so I hope more credible researchers communicating in more credible forums check White's math and figure out whether this is worth trying.
If I'm reading that article correctly, they didn't create a physical object. Rather, a scientist described a nanoscale structure which, if built, should create a tiny but detectable warp bubble.
>Specifically,” said White during the AIAA presentation, “a toy model consisting of a 1-micron diameter sphere centrally located in a 4-micron diameter cylinder was analyzed to show a three-dimensional Casimir energy density that correlates well with the Alcubierre warp metric requirements.”
>“This qualitative correlation,” he adds, “would suggest that chip-scale experiments might be explored to attempt to measure tiny signatures illustrative of the presence of the conjectured phenomenon: a real, albeit humble, warp bubble.”
So, no warp drive, but maybe an actual physical experiment we can run that might teach us something interesting about quantum physics.
And I'd say it's still a pretty big step forward. All the theories about FTL travel rely on things like "negative energy" or "exotic matter", stuff which isn't *impossible* according to the laws of physics but we've never actually seen it in real life. So if we can actually make a physical object with those impossible-sounding properties, that's a pretty big deal. Assuming the theory pans out.
Small correction: It looks like there is an experimental part to the paper. They were studying some other nanoscale structure and got measurements that looked sort of warp-drive-ish, and used that to come up with another structure they want to test. Again, assuming I'm reading this correctly.
There's an experimental part, but what looked warp-drive-ish was their calculation of what their proposed-to-actually-build structure would do, not their measurements of the actual structure (which as of the paper they seemingly haven't/hadn't built yet).
For start, is it reported somewhere reputable? Given that someone claims that NASA invented FTL (or suggests it) I am going to assume that it is pile of blatant lies.
Note that "Related posts" has "Meet the Man Building an Anti-Gravity Device, and the Alien God That Inspired Him" and link to itself. Not really trustworthy site.
Not sure is it parody or really inept attempt at science popularization.
To be clear some of facts mentioned there are true (NASA exists for example), but I suspect that more interesting ones are misleadingly twisted or blatant lies.
DataSecretsLox user WeDoTheodicyInThisHovse has started a December 2021 Welcome thread for new users who wish to introduce themselves. Having met her, I think she's a nice person and a great entry point for those who are interested.
Sure but it is the other active posters you have to worry about. IIRC Theodicy is one of the posters on the "People I'd love to interact with if only I could do it somewhere besides DSL" list.
Anyone one standard deviation to the right of center would probably really enjoy the welcome thread, though.
Well, perhaps they do. I think it's worth noting all the libertarians that enjoy it there as well. As for me, I see nothing intrinsic to someone on the left that would keep them from spending time on DSL and thinking they got something worthwhile from it.
I'm curious how people interpret the behavior of a business that's sending me daily emails, usually with one limited-time offer after another, including reminders of the amount of time left on the latest special.
Note that IIRC, this was not their practice until perhaps last year, and I've been a happy customer of theirs for at least a decade, and subscribed to their mailing list. Again IIRC, emails were often announcements of new products, and specials weren't continuous.
I'm especially interested in any insights from people working retail, particularly a smallish online business which also has a few storefronts. Products are physical, and consumable - but slowly, in case that matters.
Well, they've finally got the response from me anyone who knew me would have predicted; I've unsubscribed from their mailing list.
I'll still buy their products; they are my go to place for herbs and spices. And I hope the constant specials are not a sign that they are in financial trouble, desperate for more business, as I'd rather not go back to getting stale overpriced herbs and spices from my local grocery store.
But I need to stop waking up to more emails in my personal mailbox than fit on a single screen.
None currently, unless you count my local newspaper - which I mostly subscribed to to keep my housemate from risking covid buying it retail, at a point pre-vaccination when we were buying groceries only once a month, and soon started getting even those delivered.
I'm frustrated with the substack model. I want to interact with bloggers, and substack's commenting facilities are substandard; LiveJournal was better than this, 20 years ago.
As I see it, the problem substack solves is "how to get certain well known journalists more money, while taking a rake-off for the founders". It does little or nothing for the less well known, and I strongly suspect the sum of the individual subscriptions is commonly way more than they are in fact worth. I don't want to contribute to yet another winner-take-all / star system payment model.
Mostly, though, I'm frugal, and especially disinclined to take on recurring costs, such as subscriptions of any kind.
At any rate, if I do take on one more subscription, I'll probably go with the New York Times, in spite of excessive politicization and what appear to me to be declining standards. Even at its prices, it's more cost-effective than substack. Though when it comes to political posturing, I can and do get more than I want, for free, as email and paper mail, in spite of aggressively unsubscribing whenever possible.
So far only ACX and Freddie deBoer. The former for the excellent community and deep-dives into topics I don't have the skills to research for myself. The latter because he has repeatedly put into words the ideas that I have felt but couldn't articulate.
I'm open to subscribing to more, but none have the post output per month plus content I care for in the same way as ACX and FdB. There's one more that I'm on the fence about, but I haven't found myself reading it enough lately.
I think some of these newsletters have potential value that could be unlocked by paying for a subscription, but I have a hard time convincing myself to do that. At the end of the day, I've chosen to get a paid subscription for my top-1 newsletter; when something else replaces it and becomes rank no. 1 for me, I'll get a paid subscription to that.
I wish Substack had a bundle-subscription option like NetFlix so I could pay for more than one newsletter (versus the current model that requires N x ~5USD per-month in costs).
I think there are a large number of content adjacent newsletters that would do well financially from joint substack. There are Substack tools for joint blogs although you can't bundle multiple separate blogs. So you'd give someone Contributor access or w/e permission and then you and your buddy or buddies could post content as a group. The cost-benefit would probably be complex when you also run an individual substack, though.
I’d probably spend around $500 on substack alone of I paid for all that I read. Compare it to Apple News* which is $120/year for family access of a host of traditional newspapers and magazines. Newsletters are shockingly expensive right now unless one’s interest is very narrow and limited to 1 or 2 subscriptions.
Thanks @Gunflint! I guess I wasn't clear - by 'newsletters' I really meant Substack style newsletters, and not traditional newspaper/magazine media. That said, it's certainly valuable to hear about the relative tradeoff here!
Way back at the beginning of the pandemic, people were publishing lots of articles and videos explaining simple models of how disease spreads, showing for example what happens in the Susceptible-Infectious-Recovered (SIR) model in which a population is uniformly mixed and randomly spreading a virus around, producing a neat little exponential curve that turns into sort of a bell shape.
I always assumed that these toy models were only just for explanatory purposes, not for any serious epidemiological research. The following article is interesting because it explores a non-simple model of disease spread, but it also makes a shocking claim that models "like" SIR are "the most commonly used".
Surely the average epidemiological researcher is not that stupid? So, is this article is being misleading/disingenuous, or are trivial models like SIR/SEIR *actually* widely and directly used by epidemiologists to make predictions, make recommendations and evaluate policy effects?
Aside from that one stunning allegation, the model presented in the article is interesting (but as the author notes, not necessarily accurate) because it produces vaguely similar hard-to-explain patterns like those we see in the real pandemic. Basically the conclusion this type of model suggests is that the course of the pandemic in a region *cannot* be predicted in detail without a lot of good data we don't have about how specific people are physically interconnected with other people in the real world. A corollary: attempting to evaluate the effects of public policy based on things like "cases before and after policy was put in place" can give more noise than signal.
Arguably the most interesting result from this model is that a variant can come to dominate in a population *without* being more infectious/contagious, which has caused me to wonder if delta/omicron are less infectious/contagious than they appear to be. But remember, there are *thousands* of catalogued variants and it's very rare that one of them "takes over the world", as delta did, which suggests that delta really is more infectious.
I like the fact that Lemoine's model can explain the phenomenon that I've been noticing — which is a clear downward trend in case numbers as certain variants become more prevalent. For instance Alpha didn't come to dominate either the US or UK's viral landscape until after the 20-21 winter surges. In fact, despite the popular delusion among epidemiologists that Alpha kicked off case surges in various countries, I can't find it dominating any surges until after the peak had passed. Likewise we see Delta only coming to dominate the viral landscape in Brazil after the Gamma surge. (OTOH, unlike Alpha, outside South America, Delta definitely creating surges all over the world PLUS it came to dominate the viral landscape worldwide.)
The best reason for using a mind-numbingly simple model is if you don't have enough data to fit a better one. We all know that an agents-in-a-network based model would be more realistic, but it's got a zillion extra parameters and you don't necessarily have enough data to fit them sensibly.
OK, but why don't we have enough data? This isn't a COVID-specific thing; this is something the epidemiological community has needed since the 1980s at least. If you're trying to model an epidemic, whether HIV or COVID or anything in between, you need data on how people actually interact with one another, not spherical disease vectors randomly colliding in zero-dimensional space, but actual human behavior patterns. And it is data that would have been easy to get to at least a first order through surveys and the like; possibly something that could have been piggybacked onto existing surveys like the GSS. And then validated by comparing model to reality in e.g. annual flu outbreaks.
So why, after forty years, are epidemiologists still working with zeroth-order models?
First off, how accurate are your R0 values? That's the heaviest weighted variable in most models, and hardest one to determine. I'm still seeing wide ranging R0 values of SARS-CoV-2. In the early stages of the pandemic, I saw values as low as 1.5 and as high as 5.0, but it was population and nation dependent. B.1 barely touched SE Asian and African populations, but it ripped through European populations. Some populations seemed to have a higher natural immunity against B.1, and there were some cool studies correlating HLA frequencies and IFRs. But all those became moot when B.1.617.2 (Delta Classic) hit the scene. Previously "resistant" populations were getting hit hard by Delta, but the R0 for Delta was still showing up as between 2.5 and 5 depending on the country (and depending on the research team performing the "calculation" — R0 = wild ass guess?). And worse yet, I never saw *any* epidemiologists publish their reasoning for how they calculating their R0 values. Nowadays the generally accept R0 for Delta is 3.0. But is that accurate?
Anyway, right off the bat you've got an uncertain variable underpinning your model. You might say — well, run your model with a range of R0 models and see which fits the actual data best. But, it's pretty clear that the reproduction value of COVID-19 change over the course of an outbreak. For instance, with a R0 of 3.0 the virus should keep spreading until roughly 70% of a population has been infected. But none of the surges have infected more than 10% of the population before burning out. Why? One of the theories goes that once the density of people with convalescent immunity gets beyond a certain level, an R0 of 3.0 can't keep propagating. Which has raised all sorts of questions about the connectedness of different groups of people. For example, how do you calculate say the connectedness between Hassidic Jews and Blacks in NYC? Or even at a more basic level, the differing connectedness between family members, friends, coworkers, and fellow community members.
An alternative theory is that once a surge starts happening, people start taking more precautions (masks, social distancing, etc.). But how do you quantify that variable? Oy, my brain hurts!
Anyway, modelers seem to be very invested in their models. But none of the modeling predictions that the CDC publishes every month have been accurate for more than 2 weeks out. That makes sense to me. We can't predict the weather more than 2 weeks out, why should we be able to predict the course of an epidemic?
Blegging for a recommendation for a Silicon Valley GP that's rationalist-adjacent(-ish?). I don't expect a full-send, but anyone even marginally closer to the community or that others get along with would be super appreciated.
My therapist recommended that I try to meet women at Asperger's/HFA support groups. Has anyone here done this? No offense, but I feel like this would be a good place to ask.
There seem to be a lot of programmers here. And it's often said that there a lot of autistic people here. Why doesn't someone program something like Bumble, a dating app where only women send messages, specifically for autistic women who are open to dating autistic men?
Because isn't something called a support group meant to be a low-stress place for attendees?
Both me and my girlfriend are autists. We didn't try to find autistic partners on purpose, just happened to have similar interests. Actually, we didn't even know that we are on the spectrum until we started dating. It was funny, really. At one moment she started investigating whether she is autistic, found lots of evidience in favour and I was like: "Pff, this doesn't have to mean anything, I mean I fit like 80% of this as well, so what, does it make me autistic now?" Took me some time to consider the possibility that the answer is actually "Yes".
We have pretty good and deep relationship. We explicitly talk about our feelings, do not need to mask and in general have short inferential distances and understand each other really well. For her it's the best relationship she has ever had. For me its at least top 2: more challenging but also more fulfilling. I think it has something to do with the fact that I'm more high functional than she is, and nearly every of her previous relationships was a mess which gave her couple of traumas.
I would not recommend this, I think people overvalue similarity in relationships. 2 aspergers types will frequently butt heads in my experience and you don't exactly want someone who shares your weaknesses in a relationship. Instead I would recommend pursuing women who grew up with autistic siblings. I say this because a while ago I looked back at the set of women I had gotten along well with and noticed that an absolutely statistically improbable number of them had autistic siblings. I suspect that growing up around someone helps them not be so off put by certain behaviors (in women unfamiliar with neurodivergent types, you can easily trigger Anamoly detection filters and just get thrown in the "creepy" bucket)
Given the heritability of autism, isn't there a big overlap between "is autistic" and "has autistic siblings" anyway? This sounds to me like the best partner would be someone who is on the autistic spectrum, but has little or no visible problems.
I am not opposing your advice, though. People who are on the spectrum but have little or no problems will probably not join a support group. But you could find them by metting the people in the support group... and then trying to meet their siblings.
Oh boy. This sounds like it could go horribly wrong, then again a lot of my paternal family are 'on the spectrum' yet they managed to get married, so it could work!
I think there might be trouble communicating romantic interest with autistic women, but what do I know, it may well be many of them are open to finding love and partnership.
> I think there might be trouble communicating romantic interest with autistic women
I guess a proper pickup autist might start like this: "Hello! I am romantically interested in you. If you are not interested in me, that's okay, no pressure. However, if you decide that you would like to explore this topic further, here is my card, send me an e-mail, and we can have a dinner together."
I'm kind of on the spectrum myself, and back in the days of yore when anyone might have been interested in approaching me, you really would have needed to go the "here is a written sign saying I want to ask you out", as subtle hints and non-verbal signals would go right over my head.
(Not that I would ever have said 'yes', but part of ignoring any possible approaches was simply not twigging that they were possible approaches).
In my RSS reader, whenever I scroll past an item, it's faded out to mark it as read. Does anyone know of a similar Chrome extension for Twitter (ideally the read tweets would be synced between computers)?
Fun fact. The Spanish flu - which was a novel virus at the time - never went away. Modern seasonal flus are direct descendants of that. What happened is that more transmittable strains were less fatal. Could be good news for Omicron.
My understanding is that influenzas descended from the Spanish Flu were the dominant seasonal flu strains for several decades after the 1918-1920 pandemic, but they largely died off in general circulation in the 1950s. The pattern being that pandemic flus tend to be followed by endemic seasonal flus descended from the original pandemic strain.
The 1957 H2N2 pandemic was the dominant seasonal flu until the 1968 H3N2 pandemic. The latter's descendants are still in widespread seasonal circulation (or were, until the Covid lockdowns cancelled the last couple flu seasons), but were joined by a different H1N1 strain (same general variety as Spanish Flu, but not necessarily directly descended from it) from the 1977 flu pandemic, and then by yet another H1N1 strain from the 2009 flu pandemic.
There were also two families of Influenza B among the seasonal flu, although I don't think either of them have been traced back to any known flu pandemics; instead, they seem to have been endemic seasonal viruses since time immemorial (in the colloquial sense, not the specific legal sense of having been demonstrably so since the ascension of Richard the Lionheart as King of England). One of these two families, the Yamagata lineage, looks like it may have died out due to the Covid lockdowns.
The pattern of Influenza A pandemics developing into less-deadly seasonal/endemic flu strains is partially hopeful and partially dreadful sign for Covid. Hopeful because less deadly is obviously a better outcome from our perspective than more deadly, if we must have an endemic virus. Dreadful because "less deadly" isn't the same as "harmless" or even "mostly harmless". Tens of thousands of people in the US alone die of seasonal influenza each year, which isn't an ideal circumstance.
And it's only a weak signal for what to expect from Covid, since while SARS-COV-2 and Influenza A are both viruses that cause upper respiratory infections in humans and can cause deadly pandemics, they're not at all closely related viruses (different phyla: SARS-COV-2 is a positive-sense RNA virus, while Influenza A is a negative-sense RNA virus) so we can only infer so much from one's evolution as an endemic human virus about what to expect from the other. Endemic/Seasonal Covid might be a moderately deadly threat like seasonal flu, picking off tens of thousands of people mostly from among the old, sick, and immunocompromised. Or it might truly become mostly harmless, like the other endemic human coronoviruses that are among the causes of the common cold. Or it might remain deadly at some significant fraction of the deadliness of current Covid strains among the subset of the population that's vulnerable to infection.
My understanding is that flu reproduces sexually. The H and N antigens are modular. When a new strain is introduced from pigs or birds, I think what happens is that it keeps most of the virus specialized for humans, but swaps in new antigens, which humans haven't seen for a while.
I didn't know about the recombination part, so thank you for pointing that out.
After digging a bit, I see that Influenza A has eight RNA strands that seem to work kinda like mini-chromosomes. Two of them contain the genes for the HA and NA proteins respectively (variants of which give the H and N numbers by which the strains are traditionally categorized). The strands do indeed mix and match when the same host has concurrent infections by different strains.
From what I gather, three of the eight strands differ radically between common seasonal Influenza A varieties (the HA and NA strands, plus one more) while the other five are much more closely related. The 1918 pandemic strain's genome, at least the parts that we've recovered of it, seems to be near (but not quite at) the base of the philogenic tree for both the HA and NA strands relative to modern seasonal H1N1 and for at least one of the other strands common to all modern major seasonal Influenza A varieties.
The standard interpretation seems to be that the five common strands and the H1 and N1 strands are descendants of a recent ancestor of the 1918 pandemic strain. The 2009 H1N1 strain is believed to have come from a descendant of the 1918 strain that had become endemic in North American domestic pigs during the original pandemic, and the 1977 strain is suspected to have come from a stored sample of 1950s-era seasonal H1N1 that got loose while being studied in a Soviet lab.
If the same host is infected with two different strains of influenza at once, you can end up with genes from each one, and potentially get HA/NA genes that weren't circulating in humans before and so nobody has any immunity to them.
Yeah, I shouldn't have used the word reproduction. Sex is never for reproduction, but only for exchanging genetic material. Some species have mechanisms to force sex before reproduction. Maybe other species shouldn't be said to reproduce sexually, but that really draws the line in the wrong place.
Bacteria have a lot of adaptations for sex, but viruses only exchange genetic material incidentally, when two viruses infect the same cell at the same time. But some viruses have some adaptations to make the mixture more useful. In particular, flu has a modular genome, to make it easy to swap one H for another and separately N. While the host-specific adaptations are separate
I had a long talk with a biologist about this, and I think it's somewhat wrong.
Flu viruses are constantly mutating. They can't not mutate. Their genetic code is unstable. Only the evolutionarily imperative to be good at what they do (infect people) keeps them even mildly similar to each other.
The Spanish flu was very, very good at being a flu virus. But when it infected everyone in 1918, everyone got immunity to it. So continuing to be the Spanish flu was no longer an option. It mutated into other flu strains, including the ones we have today. So it's true that all modern flus are descended from the Spanish flu.
But that doesn't mean they're "better than" the Spanish flu from a flu-fitness perspective. It's entirely possible that if we dug up the old Spanish flu from some laboratory and set it loose, it would give us another 1918-level pandemic event, much more severely than any existing flu, because it really is just an incredibly successful and effective flu virus, and nobody has immunity to it anymore.
Existing flu strains can't mutate "back" into the Spanish flu because it's too small a target. It took evolution however many thousands of years to invent Spanish flu, and it will take it another thousand years to reinvent it, even though it still has some of the pieces left over.
Not to argue against you or a biologist who knows more about viruses than either of us...
But one factor that did not exist for most of the thousands of years pre-Spanish-Flu was Industrial-Age warfare. (The first death was a cook at a US Army training facility in Kansas. Not sure whether he was Patient Zero or not. The spread of infection from that point depended on the size of the Army training camp, and large numbers of people traveling via railroad/steamship to many other places in the United States, and the rest of the world. Cases were seen on the East Coast within a week or two, and cases were seen near the Western Front in Europe within eight weeks.)
There may have been earlier versions of various viruses which could have done a Spanish-Flu style outbreak, but weren't able to. After all, if the nations of the world are not moving large armies across the globe using Industrial-Age transportation tools, the outbreak might be local and take many months to move across a single nation.
It's very hard to tell what various Influenza viruses were capable of before the Industrial Age transportation networks allowed carriers to circle the globe within a few weeks.
That shakoist substack post linked was especially good. I am not sure he explicitly pointed it out but it seems like the central limit theorem strongly argues against the methodology used by ivmmeta.com people. If all of the studies (they include) were equally valid then they should nicely form a normal distribution about the actual value of ivermectin's effectiveness. That they do not, ought to invalidate the methodology of...pretending that they do.
Question for the resident psychiatrist: How does fluvoxamine look side effectwise?
(Background: triple vaccinated but zero antibodies due to immunesuppression, contemplating it as preventive measure, prior experience with SSRI and SNRI almost a decade ago)
I just learned that large passenger planes have their own radars that can be switched between weather mode and terrain mapping mode (this explains where the pilots get their information from when they announce over the intercom that turbulence is expected in X minutes). I assume the pilot flips a switch to toggle between modes.
I found videos showing what the radar screen looks like in weather mode, but nothing for terrain mode. Can anyone find one for me?
Also, how often do commercial pilots use radar terrain mode, and under what circumstances do they generally do it (e.g. - when lost, when approaching for landing)?
Do the radar computers come pre-loaded with detailed topographical maps of the whole planet, allowing them to automatically deduce where the plane is based on what the radar sees in terrain mode, or are pilots left to look at the screen and figure it out by comparing the image to paper maps in the cockpit?
I can't imagine how a large passenger plane could get lost and for landings the amount of ground clutter reflections might make it more trouble than it's worth given that altitude radar plus orientation instruments exist (and glide path beams now that I think of it).
Guess before reading further: how many (% of all) commercial airplane crashes between 1993 and 2002 were caused by fully controlled airplane crashing because pilots failed to notice they are flying into ground or water? Where it would be completely avoided as plane was steerable?
Which % was between 2008 and 2017?
1) when you land you want to end just at ground level - not higher, not lower and airports have various ground level
1a) In normal operation only during landing/take-off plane is low but not all operation is normal
2) some airports are near mountains. Following are extreme cases
I doubt GPS is fast and precise enough to pilot an airplane in rough condition or for an emergency landing. And I'm not sure GPS coverage over polar region or large oceans is top notch either.
GPS is fast and precise enough to avoid mountains, at least mountains that haven't moved since the last time the terrain database was updated. They are also precise enough for instrument landings in adverse weather. Source: own one, use it for avoiding mountains and landing in adverse weather. GPS also works over oceans and polar regions, not that it matters for this purpose due to the lack of mountains to dodge or runways to land on.
Nobody uses on-board radar to directly land airplanes, in emergency or otherwise. The military does use it for terrain avoidance; I haven't heard of civilian aircraft doing that but it's not absurd and might be a reasonable backup for GPS in remote mountainous areas.
Is worldwide terrain database good enough to use just GPS? I assumed that in some areas or during emergency on-board radar may be more likely to be used.
GPS coverage over the polar region and large oceans is top notch; it's better than GPS coverage anywhere else, because you have respectively more satellites and less interference. GPS also works much better in airplanes than, for example, in cars, because there's less crap above you to block it. Its precision is a few meters, which is plenty fine for an emergency landing if you have some way to tell how far away the ground is, such as a radar or visibility over 10 meters.
Given that the US government made GPS freely available for use worldwide after a Korean Air passenger flight was shot down over USSR prohibited airspace in 1983, precisely to prevent such events in the future, I'm pretty sure it's fast and precise enough to do all of those things.
[Accidentally navigating hundreds of kilometers out of course] and [making an emergency landing] are not similar events. GPS prevents the former well but don't help that much with the later.
High school senior here, planning to attend college. A few wonderings I'd love to have some perspective on:
1) Is a Computer Science degree as could as people say it is? I've tentatively decided on CS, but I've heard alternatively that there's either a crippling shortage of CS professionals in the US, or you can spend months on the jobsearch and the interview process could be a new circle of hell. I'm also specifically interested in cybersecurity; how does that compare with CS?
(To be quite honest, I'm also worried that if I go into CS I'll never get a girlfriend. Sue me, I'm a teen.)
2) What should I direct my effort toward in college? I've heard that college isn't useful for the education, but for the signalling power a degree holds. So rather than academics, what should I funnel my effort toward? "Enjoying my youth"? Building ties with competent like-minded peers in preparation for the future? Building work experience with internships? Working on personal projects?
3) How can I up my conscientiousness/build self discipline? My test scores are always excellent, but my gpa is abysmal in comparison. Part of that is being hit by depression and later a behavioural addiction to reading, but I'm also just lazy and weak-willed. 'Executive Dysfunction' was suggested back when I had a therapist.
4) General college/job/life advice? I can't see myself sticking with one career for the rest of my life; I'd rather switch it up every few years. I'd also like to travel, and am thinking of attending a year of college abroad, perhaps in Germany. These may just be juvenile sentiments that will pass, however, and in the case that I'm romanticizing something that sucks in actuality I'd like to know.
Can confirm that buggering off to Germany in the middle of a degree is a good idea. Though it was before Brexit so I had the opportunity to work there.
I'm responding primarily to question 1. For context, I'm a software engineer on the cusp of retirement, and in the prosperous/high status end of the income distribution, though not at the very top.
1) A computer science degree from a good tech college, or a very well regarded generalist school, will almost certainly get you employed after college, with decent or better pay. It will also put you on a path with potential for very good pay indeed, either as a senior high end technical person, or on the management track.
A coding course from an also-ran outfit will probably keep you employed, but the first job will be harder, and you won't be a candidate for the high end.
Be aware that there's a huge range of salaries in this area. The funnel for the best US jobs takes most of its input from MIT, Stanford, etc. Waterloo if you are Canadian. This funnel starts with internships during college, which are vitally important. Also, the best career paths now generally require a masters degree.
Essentially all the male software engineers I know are married; most of the females are too. A good salary can be a potent aphrodisiac, at least for those seeking a long term relationship.
Some of them wind up repeatedly divorced, due to putting their careers ahead of their relationships. Be prepared to work long hours, especially if you are ambitious.
I know nothing specific about cybersecurity.
2) I learned a lot in college, and am glad I paid attention to my courses, but I'd probably have earned more over the course of my life if I'd paid more attention to building long term relationships with people outside of my field. Contacts turn out to be vitally important.
Internships and contacts made on them will, among other things, quite likel get you your first job in CS.
4) I suspect CS will be a decreasingly good career as time passes. I certainly liked it better 20 or 30 years ago, but part of that may be having been promoted to a level where I have to deal with a lot more things I consider BS, rather than the code, which I love.
It's probably good that you see yourself as moving to new things later in your life.
I'm a big fan of learning languages. If your German is remotely good enough to manage, go to Germany and make it a whole lot better. But if you do, don't hang out mostly in an American enclave, speaking English, even though that will be very tempting.
> Be aware that there's a huge range of salaries in this area. The funnel for the best US jobs takes most of its input from MIT, Stanford, etc. Waterloo if you are Canadian. This funnel starts with internships during college, which are vitally important. Also, the best career paths now generally require a masters degree.
FWIW, I got a FAANG job right out of college with just a CS bachelors at a state school. Internships are important though.
Try out advent of code (https://adventofcode.com/) and see if you like it. Married CS person here, seems to work out as long as you know, you talk to girls sometimes and don't just stay in front of the computer.
If you're really concerned, make sure not to skip out on the gym. Strength training can be fun too. Or dance, as someone else said.
Cybersec is cool if you can get into it, and if you are interested in it, don't wait till third year for the school to teach it to you. CS is a really really wide field, and school will try and build a foundation and then show you some stuff (e.g. one graphics class, one database class, one security class, one UI class, one AI class), but none of the classes will really go as deep as you can go on your own if you put the work into it. And honestly, the lectures aren't always better than youtube tutorials and trying stuff yourself.
However it is important to understand enough about how computers work, what memory is, why does this algo take O(n) vs that one taking O(n^2) time. And the curriculum should make sure you understand all that by the time you graduate.
I think my Math and Chinese language classes were "harder" in the sense that I don't think I could have learned the stuff from them from the internet / textbook. So sometimes you might decide to just study CS stuff for fun and take the classes which are harder to learn on your own. (For me, I also failed to learn graphics and OpenGL on my own, so I gladly took the class.)
Re:3
One piece of advice I heard from someone about motivation and such goes as follows.
"Even if I don't feel like it, I always at least get ready at the gym and tie up my shoes. If after that I still don't want to do it, I have my own permission to switch right back. But often, once I am fully prepared to do it, it doesn't seem so bad, and I can go and complete my workout"
I think that approach works quite well, where you make some time, say "I will at least get everything set up here, maybe set up my editor, turn off the internet (Download the docs ahead of time) and set a timer for 15 minutes"
If after all that, you still don't get any work down, well, go for a walk or something and try again in a bit. But often I have trouble with the putting down the distractions and less with the picking up with the work.
You're getting lots of great replies already, but I'll add a few thoughts:
1) CS is good, and learning to program IMHO is one of the most valuable skills you can have. Basically every career involves computers already, and it's safe to say computers will be used even more in the future.
2) Do all that stuff you mentioned! Include a mix of potentially useful stuff (e.g. internships) and just-for-fun stuff (e.g. anime club). Some stuff will be awesome, some won't work out, and college is the time to experiment and find out.
3) Learning about yourself is just as important as academics, if not more. I'd highlight mental health as being something particularly important to work on, since nothing else really matters if your brain isn't in good shape. Campuses usually have mental health resources available, don't be shy to try them. That said, for self-help you can also check out https://www.lesswrong.com/posts/JgBBuDf5uZHmpEMDs/how-you-can-gain-self-control-without-self-control.
4) It's hard to give much general advice because everyone has different values; there are a lot of perfectly valid approaches to living life! I'd say first, take advantage of what college offers you. Go to office hours, go to career fairs and that sort of stuff, socialize, take elective classes that sound cool even if it's way out of your usual field, and try some clubs/activities. Second, keep an open mind about careers. It's better to think of college as a chance to collect information on potential careers, rather than training for a career. A lot of people end up happy in a career very different from what they could have imagined during college.
Married to a software engineer. We have a nice life. It helps that he’s good at what he does, and he’s good at what he does because he likes it. I can tell when he’s been writing code because he has a spring in his step.
As for girlfriends, my standard advice is to hang out around the library science program. It’s 95% female, and I know a lot of librarian-programmer pairs. I think it has something to do with CS people liking communication to be specific, and librarians being able to rise to the occasion.
Strike a balance between doing things you enjoy and that are useful to you (e.g., making and saving money, maintaining good health, finding and keeping friends you truly like). Ideally, often these two areas overlap, to people's points about figuring out whether you enjoy CS. Not everything needs to be enjoyable, and not everything needs to be useful.
Just because some people don't learn a lot in college doesn't mean you shouldn't. Try to learn a lot, whether it's the material itself or the meta-skills (how to learn hard things, how to write and communicate well, etc.) This is also a good way to build up your discipline, along with an exercise habit.
Find people you really like and keep them in your life. Old friends who have known you for decades (when you're older) are among the best things in life, so find those people now. That may mean you need to meet a lot of people in college. Don't assume the first people you meet are the ones you'll click with the most. In both college and grad school, I felt a little out of place for a year or more and then found great friends that I've stayed in touch with ever since. Try out different activities and groups and see what and who sticks.
Travel, a year abroad, and whatever enthusiasms you have now are definitely worth pursuing. You only get one life, as far as we know, so make sure you have some fun stories to tell when you're older. Even many experiences that suck in the moment will seem interesting and enjoyable as memories. So don't try to be *too* mature and practical. There's a place for that, but make sure to really live life and do some weird things.
You definitely don't need to stick with one career if you're reasonably smart, likable, and a good communicator. In fact, having multiple skills sets that overlap in interesting ways often make you more valuable on the job market.
> there's either a crippling shortage of CS professionals in the US, or you can spend months on the jobsearch and the interview process could be a new circle of hell.
Both could be true at the same time.
> I've tentatively decided on CS
Have you tried going through a programming tutorial?
If you can do this then it is a strong hint that it is a good field for you.
If not - can you do this with a tutor? If not, then it is a strong hint that it is not a good field.
Have you programmed for fun? (another strong hint).
> Building ties with competent like-minded peers in preparation for the future? Building work experience with internships? Working on personal projects?
Good ideas. Note existence of projects such as Google Summer of Code or Github Student pack. Look for more like that.
I would add "join interesting open source projects"
If you're good at it, CS is the best thing ever. It's the closest real life equivalent to being a Wizard developing magic spells. You should probably start learning to program before you get to college so you can get a feel for it and decide whether it is something you want to pursue or not.
As for "getting a feel for it" and "being good at it", don't be alarmed if it's hard and you can't figure things out. It's hard for everyone and it's always hard to figure things out, but especially when you start; nobody is born good at it.
1) I don't know if you enjoy computer programming. CS will open the door to making lots of money, but if you intrinsically enjoy programming, most of the process will be at least somewhat enjoyable; if you hate programming, it will be torture. Cybersecurity seems like a specialization that you can choose later. (But I am just guessing here.)
To find a girlfriend, learn to *dance*. It is a great balance to the sedentary CS lifestyle, and not only you will meet lots of girls, you are allowed (and socially expected) to touch them and do something that seems like a vertical simulation of sex (somewhat exaggerating here, but not too much, depends on the specific dance). So you essentially skip the first few steps of dating.
2) Yeah, it sucks that the part of your life when you have most free time is also the part when you least know how to use it well. Try a bit of all, I guess. Explore. At this phase of life, learning is more valuable than doing (you will have the following decades to profit from what you are learning now)... with the exception that sometimes doing things is also a source of learning. Just don't spend *all* your time doing *one* thing; that is, if you e.g. start an internship, don't let it consume all your time.
4) Learn German e.g. here: https://deutsch.info/ and get good at skills that are useful across a large range of professions, such as communication and math. If you know German and programming, you can easily find a well-paying job in Europe. (English and programming is already quite enough, but if you add German to the mix, it is even better.)
Seconding the dance recommendation--swing dance clubs are incredible for building social confidence and spending a lot of time interacting closely with the opposite sex.
So the optimal strategy is to be born in Europe, study in Europe, move to USA and make money until you burn out, return to Europe, retire early in your 30s, start a family.
If you are already a teenager in USA and feel adventurous... I am not sure whether it would make sense economically to learn German, move to Europe and study there, then return home. As a foreigner, you would probably have to pay something for the school. Also, travel and accomodation are not free. But given the astronomical costs of some American universities, is there a chance that it might still be cheaper?
Travel and accomodation are not free, but college education is free or costs little in a lot of European countries, including Germany (except Baden-Württemberg and parts of Sachsen, where it costs ~3000€ per year for non-EU students).
Accomodation cost can vary extremely, depending on the city. If you don't have excessive tuition fees like US or UK, this is by far the most expensive part. But you have accomodation cost regardless of whether you study in the US or abroad.
Do you enjoy problem solving and do you not enjoy people? Then major in CS like me, because every other job is either dirty (I would prefer to sit in an office), about people, and/or has way less problem solving.
It's not the most fun thing in the world, but I actually kind of get a dopamine hit doing leetcode (and DSA type problems are one of my least-liked areas of CS). That doesn't happen for me when I'm memorizing biochemical pathways. I was going to go to medical school but I realized that it would be hellish, because I hate memorizing factoids. I like understanding complex concepts that you can't just google and I like solving problems. The existence of programming is pretty much a miracle with regards to trying to find a problem solving job.
And to comment on some other comments nay-saying CS. Yes, CS is not perfect. Your job might feel like a job. That's sadly normal. But what's the alternative? Other forms of engineering? Most of them end up writing code. I was doing premed + an engineering major, and I realized that medicine is very boring, and that engineering means you will probably end up coding, and if you don't, your job will be no more interesting than programming, and will probably be dirty, i.e. in a plant or a construction site or something gross.
So yes, there are some people below basically talking about how corporate meetings and making webshit and Javashit is boring. I agree. The thing is that there are no alternatives I can think of that CS doesn't provide a pathway to. For me the alternatives would be self-employment, teaching/research, or some sort of employment that is low on meetings and high on interesting work.
I can't think of a field that qualifies for the last thing more than CS. There is a lot of webshit being produced these days. I don't find it very exciting to write corporate applications in high level languages compared to doing harder, more cutting edge stuff. But the only better option outside of CS I can think of would be researching another STEM subject. Guess what? If you want to be a professor, you can be a CS professor. It's probably easier to get a CS professorship or private research position because the field is expanding more than the old sciences.
I can't think of a job that I would find more interesting than coding other than other STEM research, and since CS gives me the opportunity to do such research in the future anyway, it's pretty much a no-brainer for me to do CS.
As for degree difficulty: if you are good at math, CS is the easiest STEM degree available, period. Have you programmed before? If not try building a calculator in Python or something. If you make As in math, and Python kind of fun, you will find CS super easy. If you are not good at math, however, a memorization based "STEM" degree might be easier. I think there is only one of these, depending on your definition of STEM. Maybe two. Biology has the absolute least math, then Chemistry. For Chemistry, you will take a truncated version of chemical physics. CS probably comes after Chemistry in terms of the math involved.I would say if you aren't good at math, CS is probably the third easiest STEM degree after Biology and Chemistry.
And I don't think it's necessary to compare to humanities degrees. Frankly, those are pointless and you shouldn't have to waste your time with them. Although in the job market right now they serve as a ticket to random office jobs that need a certain IQ level but not any skills. You could easily do them in 8th grade, but you need to get a business degree or whatever to show that you can read and right at at least a slightly above average level. If you're really smart though, you need to be majoring in STEM, and if you don't want to be a professor, it needs to be engineering or CS, or maybe math or statistics.
>2) What should I direct my effort toward in college? I've heard that college isn't useful for the education, but for the signalling power a degree holds. So rather than academics, what should I funnel my effort toward? "Enjoying my youth"? Building ties with competent like-minded peers in preparation for the future? Building work experience with internships? Working on personal projects?
Maintain a high GPA (aim for a 4.0 and if that fails aim for Summa Cum Laude). Education, for most majors, is a hollow signal. For CS, you will actually learn a lot of useful things. For STEM in general, if you get a job in your field, you will actually learn useful things. Business/humanities degrees are hollow signals unless you become a professor of those things.
Education being a signal means your GPA matters a lot. Education being about skill means it GPA matters less. In my experience, your GPA matters some amount which is proportional to how much of a hollow signal your degree is. The floor, however, is still really high. So GPA still matters a lot for CS and engineering.
I have this rich uncle who's a dentist, he told me the number one thing that matters in college is your GPA. So far my experience has proven him correct. You can recover from a low GPA to some extent proportional to how useful your degree actually is, but it's still a recovery. Don't make yourself have to recover.
As for what to do besides your GPA: research and internships. It's that simple. Nothing else will go on your resume. This isn't high school, you're not applying to college, you don't put bowling club on your resume. Do it if you like it, but keep in mind it's all about research and internships when it comes to building your resume. For CS, you also have a portfolio you can build from your basement, called github. Isn't that cool? You should put that on your resume too.
>I'm also specifically interested in cybersecurity; how does that compare with CS?
AFAIK there are no "cybersecurity" bachelor degrees and if there are you probably don't want one. I go to a good public school and they don't have one. The cybersecurity classes are CS electives. I am taking them because I also like cybersecurity. However, the field is somewhat hard to break into. You want to get a CS degree with cybersecurity classes on the side, and then potentially follow up with a masters if you feel like you need it. The reason why is that the jobs are very rare and you box yourself in with a "cybersecurity" degree.
>I've heard alternatively that there's either a crippling shortage of CS professionals in the US, or you can spend months on the jobsearch and the interview process could be a new circle of hell.
Same here. I think it comes down to competency. The job market looks good statistically speaking. Sometimes I hear people who start coding and a year later they're making $100k doing it full time, sometimes I hear about people who hardly get interviews and then bomb them when they do. I've seen resumes people post and there is a definite correlation with competence. https://old.reddit.com/r/cscareerquestions/comments/3e55c8/interviewers_can_people_really_not_pass_fizz_buzz/ctbowp4/
For whatever reason, maybe because there's no licensing, maybe because they hear CS is friendly to people without CS degrees, there's a lot of people who literally cannot do the most basic tasks, and who have no capacity to dynamically learn on their own, who apply for jobs. I think if you're above average it will be the most ripe, 1360s esque job market you will ever see. Signals like GPA and portfolio will help you get interviews and competence will help you pass them. People who can't fizzbuzz with no experience apply with a 2.8 GPA and the wrong skillset to 200 FAANG jobs and then flunk the interviews and then they whine that they can't find the job.
1) Plenty of cybersecurity professionals have a CS degree; think of it more as a specialization than a separate career path. Regarding the job market, there are lots of jobs available, but lots of people looking for jobs suck. I wouldn't hire at least half of my graduating CS class for any coding job. Additionally, the interview process at most companies is tuned to try and make the false positive rate low at the expense of the false negative rate, and it's not difficult to fall into a "would be good at the job, but is rejected at some stage in every interview process due to some random signal" trap. This is where having personal connections really helps.
If you want to meet women, choose a woman-dominated area that you're interested in, and minor in it. Or join clubs that aren't CS related. Also, you'd be surprised how much Electrical Engineering and Math have better gender ratios than CS—not 50%, but in my experience it was better.
2) College gives you a great opportunity to educate yourself; choose the classes that interest you, and that will provide useful skills in your job, and throw yourself into them. Additionally, if you do decide to go into CS, the more experience you have programming, the better. Do hackathons, personal projects, try to contribute to open source projects. Get as much time coding as you can. I imagine there are similar skills in other disciplines, though I don't really know. Beyond that, try to make friends who will work in the same industry as you! It's way easier to get a job somewhere if you have a friend on the inside who can chat with the hiring manager.
3) I also struggled with this. One of my major issues, which may or may not apply to you, is that I would sometimes not turn in half-finished homework—get comfortable turning in things halfway done. 50% is way better than 0%. (If you're having difficulty with this, try writing "Sorry, didn't have time to finish this" at the top of the homework. Not sure why that helped me, but it did.)
4) This is much easier to do if you go make 100k in CS while spending 20k for a few years after college.
Along the lines of turning in halfway done homework, I am a IC (Chip) designer. I did this type of thing even on tests. Often in physics and occasionally in math (calc, etc). I would hit a point where I lacked an equation and say, " Not sure what this equation is but if I was, this is what I would do next....." You can get a lot of points for that. Even in math, showing you know the process is far better than an incorrect guess or just unfinished work. (I would don't know how to differentiate this, but if I did, I would and then I would set it to zero to find the max/min of the original function)
1) Yes, it is an excellent choice for job prospects. Cybersecurity is fine, but you have 2-3 more years before you make that decision. No need to worry about it now.
For finding a girlfriend, yes definitely, CS has very few female students. Obviously, there are other ways to find a girlfriend, but it is also true that you will spend a lot of time with your fellow students. If you are really concerned about it, you *could* consider starting with a related subject, and switch to CS later. For example, math has much more balanced gender ratio than CS. But this only makes sense if you already have a high interest in math. I would only recommend this if the male/female ratio in CS really troubles you a lot.
2) I agree with the others. Take your studies serious. You will learn a lot of useful and a lot of (for you) useless stuff, but it's hard to predict in advance what will become useful for you.
3) Building social ties helps. Build a study group, solve exercises together with others. If you spend time with them, it will be much harder to get distracted by other things. This doesn't solve everything, so *also* try out good ideas from other people.
4) I would very much recommend spending a year abroad. It's not necessarily the fastest way to start earning money, but most people who did this wouldn't consider it as wasted time. (Except for the last year. So sad for exchange students to have online teaching and distance rules. :-( ) Germany is a nice choice, it has good universities, and (almost) no tuition fees. I think there are many good choices, and few ways to screw this decision up too badly.
Programming jobs are great, but a CS degree is neither necessary nor sufficient to get one. It's one of the fields where the signaling power of a degree is the smallest. Software is eating the world so any job turns into a programming job sooner or later.
Security is a desperate waste of time: companies and countries that don't have security will be replaced by companies and countries that do, but currently the decision-makers in most companies can't tell security from a hole in the ground, so they buy security products based on the opinions of shill operations like the Gartner Group. This works about as well as getting a company to profitability by buying new accounting software chosen by people who don't know anything about accounting.
Don't worry about the girlfriend thing. Being an adult, in both the literal sense and the metaphorical sense, is a much bigger factor here than choice of career, unless your alternatives to CS include becoming a famous actor, playing professional sports, or some other avenue to being a celebrity. Traveling helps too. You will be amazed how much more attractive you are to women at 25 than you are at 18.
3) is almost certainly the most critical thing to get right. No matter what you end up doing, it'll matter hugely. My advice is to listen to the huberman lab podcast, particularly the episodes on sleep, dopamine and addiction. They've helped me get a handle on very similar issues after nearly 3 decades of struggling with them.
I was also pulled in a million different directions as an eighteen-year-old and ultimately decided on CS. There are going to be a lot of people in these comments who say that CS is good, so I'll provide a counterpoint and say that I wish I hadn't done it.
The big question you'll have to answer fairly quickly is "Do I enjoy commercial software programming?"
None of your classes are going to give you the answer to this; the only way to really tell is to get an internship that lets you simulate the experience of being a software engineer (i.e. spending about fifty hours a week doing some combination of writing code and sitting in meetings discussing code).
I hated it, and I realized that I hated it during my first internship, but unfortunately my first internship wasn't until the summer between junior and senior year, so I was kind of stuck.
I thought, "Oh well, no big deal; I can just freelance when I need the money and travel the world, la dee da."
It's much harder than it sounds. Skillset gets stale very quickly in software, so it's hard to jump out and jump back in. Also, you're SO MUCH MORE VALUABLE to a company if they "own your time," so to speak, i.e. if they can call on you whenever they want without paying you extra money, i.e. if you're a salaried employee. If you are a competent programmer with a degree from a name-brand school, you'll probably have people lining up to hire you as a salaried employee but still face difficulty finding decent-paying freelance work.
I also though, "Oh well, with my new CS skillset I can just found a startup and be my own boss, la dee da!"
And yeah, I really did try it! And then the startup didn't work out (happens to about 75% of them), and I was broke and I needed a job, so I ended up back at an office desk for a few unhappy years before I finally got out for good.
What I'm saying is this: If "sitting in an office and writing code with a bunch of people doing the same thing" sounds very bad to you, I would seriously question CS as a career choice, because you run the risk of just kinda washing out like I did. If the collaborative coding project paradigm actually sounds okay to you, then it could be a great fit. There were a ton of people at my company who liked (or at least didn't mind) that paradigm, and for them it was great: They got paid well and were reasonably happy. I just hated every minute of it and had to spend about five years doing it before I finally realized I absolutely could not anymore.
So just...make sure being a software engineer is actually something you'd be okay with, lol.
Other Things You Ought To Know About Computer Science:
1. Unless you are a person who is super mathematically and logically inclined, you're going to have to work a lot harder in college than you would while majoring in, say, English. I had some great experiences in college, of course, but could most commonly be found in the basement of the CS building running on very little sleep and on hour 12 of trying to get some assignment to compile. This was not the experience for my friends who were humanities or social sciences majors.
2. CS might fuck up your GPA, which will make it difficult to get into a good grad school. You'd think they'd care and take into account that you had a fairly difficult major, but they don't, really; it's kind of a pure numbers game (at least for the professional schools).
3. You mention never getting a girlfriend. I mean...CS isn't going to single-handedly stop you from getting a girlfriend, but it does mean that there will naturally be fewer girls floating around in your social circles. It's not a totally dumb thing to consider, although if you're reasonable attractive/social it's not going to matter very much (although I have heard horror stories about the dating scene from my male friends who live in Silicon Valley, lol)
In retrospect, since I ended up going back to grad school anyway, I would probably have just majored in something I enjoyed (something that involved a lot of reading instead of a lot of coding), gotten a high GPA, spent a couple years figuring out exactly what I wanted to do, and then done grad school (it's easy enough to change your trajectory with a masters or postbacc program).
Your description of yourself does not exactly scream "happy as a CS major/software engineer," so I would tread carefully.
The scarcity/glut issue of programmers is this: there are a large number of people with some form of programming credential (like a CS degree) who can't program themselves out of a paper bag. The interview process may be stupid, but so are a huge number of applicants.
A great way to get a leg up on your job application is to do some software development work in college and post it on github or whatever. Find a tool you need, work on Open Source stuff, whatever. Show that you can produce something that works.
Happy and sad fact on corona: by the end of this year, 12.5 billion vaccination doses for corona will have been produced. Monthly production is 1.5 billion.
Given that there are 5.8 billion people of age 15+ on earth, that's easily enough to give every adult two shots, with almost a billion doses left for children, booster shots, etc. (Currently about 250 million booster shots have been given world-wide, so there is still margin.) By end of March, we could give three shots to all adults on earth.
It would even be enough if every single adult would take the shots. We know that some people decline the offer to get vaccinated.
Meanwhile, Africa has given out only 18 doses of vaccine per 100 people. This is not a malfunction of vaccine production, it's a malfunction of distribution.
I am moderately optimistic that this will change soonish, and that the excess doses are finally reaching Africa. I wish I would be sure about it.
Update: the largest vaccine manufacturer, the Indian Serum Institute, has announced (or threatened) to cut vaccine production (AstraZeneca) by half because they don't find buyers. Apparently they are sitting on a stockpile of half a billion doses that nobody wants.
I am always in favor of more support to poor countries, and outraged by the lack of funding for things like malaria and general economic support. This holds for the vaccine issue, but it's worth adding that sub-Saharan Africa remains among the least-hit regions in the world by Covid, probably because of low urbanization rates (https://ourworldindata.org/explorers/coronavirus-data-explorer?tab=map&facet=none&Metric=Confirmed+deaths&Interval=Cumulative&Relative+to+Population=true&Align+outbreaks=false&country=USA~AUS~ITA~CAN~DEU~GBR~FRA). I believe this remains true when accounting for under-reporting, but don't have a source on hand for this. The fact that the same pattern is seen not only in countries with questionable statistics and health systems (say, Tanzania and Zimbabwe) but also those with robust systems and some levels of transparency (Rwanda, Ghana, Botswana) suggests to me that this isn't just an artifact of misreporting. And this fits a more general worldwide pattern of urbanization predicting infection rates (with the striking exception of China and East Asia more generally).
I think even with this considered, Africa is being under-vaccinated, but the problem is not quite as stark as the raw numbers might suggest.
> a more general worldwide pattern of urbanization predicting infection rates
Do you have a source on that? When I look at US states, the top ten states for total infection rates are North Dakota, Alaska, Wyoming, South Dakota, Tennessee, Utah, Rhode Island, Montana, Kentucky, South Carolina.
Rhode Island and Utah are in the top ten most urban states, but of the other eight states on that list, four are in the ten *least* urban states, and the other four are also in the bottom half of urbanization.
I am also a bit doubtful about the urbanization hypothesis. Looking at the list of excess mortalities here, https://www.economist.com/graphic-detail/coronavirus-excess-deaths-tracker, there are a lot of countries in the top part which sound rather rural, e.g. Bulgaria, Peru, Albania, Kazakhstan, Romania.
South Africa is also in the top part. There are very few African countries in the list, though. But I think Africa was hit really hard by the last wave in July/August, much harder than other regions were hit by third/fourth waves.
Does someone know about India? It might be a good test. was hit hard in general, but were rural Indian states hit less hard than rural ones?
There's definitely been a lot of discussion of how boosters for us are getting in the way of first doses for others. But it has sounded to me for a while like most countries are not dose constrained, but are instead either logistics constrained or demand constrained. I'm still hesitant to state that this is a fact, because I just haven't seen any systematic discussion of it, but if anyone has such information that would be helpful.
The link you post suggests that in a few months, the world will have enough doses to get everyone two or three doses, which supports what I'm feeling right now. But I'm a bit skeptical, because this comes in the form of a press release from the International Federation of Pharmaceutical Manufacturers, telling everyone "look, there's no need to weaken our intellectual property protections". It seems to me that even if there are lots of countries that are dose-constrained, they have an incentive to say they've manufactured enough for everyone, and keep selling extra doses to rich countries that are letting them expire while locals refuse their doses, rather than doing the difficult thing of selling to poor countries.
That's a good catch, thanks for pointing this out! I still find the numbers plausible, it's pretty consistent with the prognoses from spring and from summer (which came from an independent analyst company, as far as I could tell). But yes, the source is not neutral.
Also, the shortage is very specific African. There doesn't seem to be a super-severe shortage in India, or Latin America, or Indonesia, or Mongolia, going by vaccination rates.
Except that I just noticed Haiti, which has ~2 doses per 100 people, compared to 125 doses in neighbouring Dominican Republic. Sad.
I would want to dig a bit further into data on some poorer countries in other regions, like Paraguay and Laos. But I am convinced by the broader point - it's mainly vaccination capacity that is missing, not doses.
Back on the longevity thread, I proposed (without really knowing if or how it was possible) how we could just replace organs with young cells piecemeal.
Looks like it's working well for skin:
https://www.statnews.com/2021/12/08/last-gasp-gene-therapy-saved-syrian-refugee-clinical-trial-starting/
Scott: Do you have a citation for the claim about Ritalin neurotoxicity that appears in this post? https://slatestarcodex.com/2017/12/28/adderall-risks-much-more-than-you-wanted-to-know/
You write: “Never mind, recent studies suggest Ritalin is just as likely to cause this problem.”
However, a study out this year seems to indicate no evidence of neurotoxicity from long-term Ritalin use at above-therapeutic doses in primates: https://www.sciencedirect.com/science/article/abs/pii/S0892036221000714?via%3Dihub
I’d like to compare this recent study to whatever sources you had in mind as evidence of Ritalin neurotoxicity.
(Bonus question: do the conclusions of this recent article look credible to you?)
Does anyone know what happens to the barrels of tanks (armored vehicles, not containers) after they wear out? Can old barrels be restored and reused, or do they have to go to the junkyard?
Let's say I get a positive covid test today. (I don't, but I want to pre-plan.)
What's the current best practice of what to take? Paxlovid is still illegal, but what could I ask my doctor to prescribe off-label, or what could I take OTC, to give me the best outcomes?
Did you find out? Great question.
Pretty sure fluvoxamine's on the list.
That looks like it's from January 2021. Is there anything more recent?
The author's Twitter handle is @plain_fiction, which is telling, but do we have any physicists here that want to explain what a "warp bubble" is supposed to be, and why this thing featured on Slashdot is surely nonsense? Anyway, in addition to the usual prior against such trings, Harold White and his team previously failed to notice that the physics-defying EM drive they tested didn't actually work, suggesting, I suppose, a lack of rigor on their part (or worse).
https://thedebrief.org/darpa-funded-researchers-accidentally-create-the-worlds-first-warp-bubble/
Yeah, this isn't the place I'd go for reliable information on "warp bubbles", but it is an interesting read.
Miguel Alcubierre's theoretical "warp drive" is fairly well regarded by the physics community, at least to the extent of not being obvious nonsense and worth talking about. The math is pretty hairy, and I've never tried to sit down and work my way through it, but:
If the math is right, the general relativity theoretically allows for the creation of a "bubble" of warped space-time with a volume of normal space in the center but the space ahead of the bubble being compressed and the space behind the bubble being expanded - and this is a dynamic effect, so the bubble progressively "moves" forward despite not having a velocity. If this can be done in a stable and controlllable matter, voila, space travel without pesky fuel requirements and speed limits. And maybe you can build a time machine as well, or maybe you'll blow yourself up or erase yourself from history if you do because the theory and math of closed timelike loops gets really messy.
BUT: to do this, you absolutely need for part of your warped space-time to be made out of a structure with a negative energy density, and negative energy densities don't seem to be a thing (no, antimatter doesn't count). So, nice theory, impossible practice.
BUT BUT, there is a quantum-mechanical phenomenon called the Casimir Effect which creates what appears to be a negative energy density in very small volumes by suppressing quantum vacuum fluctuations. Maybe we can build a "warp drive" that way?
BUT BUT BUT, we don't know whether "negative energy density" in quantum mechanics means the same thing as "negative energy density" in general relativity. We'd need a theory of quantum gravity for that, and we haven't got one.
So now White thinks he's done the math and figured out how to build a microscopic testbed for this. As you note, he's not always 100% on this sort of thing, so it's quite possible he got the math wrong. And the necessary microscale engineering might be impractical as well. And if it *does* work, it's probably not going to scale very well.
Even so, a demonstration of an impractically microscopic warp drive would be theoretically valuable, so I hope more credible researchers communicating in more credible forums check White's math and figure out whether this is worth trying.
If I'm reading that article correctly, they didn't create a physical object. Rather, a scientist described a nanoscale structure which, if built, should create a tiny but detectable warp bubble.
>Specifically,” said White during the AIAA presentation, “a toy model consisting of a 1-micron diameter sphere centrally located in a 4-micron diameter cylinder was analyzed to show a three-dimensional Casimir energy density that correlates well with the Alcubierre warp metric requirements.”
>“This qualitative correlation,” he adds, “would suggest that chip-scale experiments might be explored to attempt to measure tiny signatures illustrative of the presence of the conjectured phenomenon: a real, albeit humble, warp bubble.”
So, no warp drive, but maybe an actual physical experiment we can run that might teach us something interesting about quantum physics.
And I'd say it's still a pretty big step forward. All the theories about FTL travel rely on things like "negative energy" or "exotic matter", stuff which isn't *impossible* according to the laws of physics but we've never actually seen it in real life. So if we can actually make a physical object with those impossible-sounding properties, that's a pretty big deal. Assuming the theory pans out.
If someone speaks physicist, here's the actual paper they're reporting on: https://epjc.epj.org/articles/epjc/abs/2021/07/10052_2021_Article_9484/10052_2021_Article_9484.html
Small correction: It looks like there is an experimental part to the paper. They were studying some other nanoscale structure and got measurements that looked sort of warp-drive-ish, and used that to come up with another structure they want to test. Again, assuming I'm reading this correctly.
There's an experimental part, but what looked warp-drive-ish was their calculation of what their proposed-to-actually-build structure would do, not their measurements of the actual structure (which as of the paper they seemingly haven't/hadn't built yet).
For start, is it reported somewhere reputable? Given that someone claims that NASA invented FTL (or suggests it) I am going to assume that it is pile of blatant lies.
Note that "Related posts" has "Meet the Man Building an Anti-Gravity Device, and the Alien God That Inspired Him" and link to itself. Not really trustworthy site.
Not sure is it parody or really inept attempt at science popularization.
To be clear some of facts mentioned there are true (NASA exists for example), but I suspect that more interesting ones are misleadingly twisted or blatant lies.
DataSecretsLox user WeDoTheodicyInThisHovse has started a December 2021 Welcome thread for new users who wish to introduce themselves. Having met her, I think she's a nice person and a great entry point for those who are interested.
https://www.datasecretslox.com/index.php/topic,5274.msg190011.html
Sure but it is the other active posters you have to worry about. IIRC Theodicy is one of the posters on the "People I'd love to interact with if only I could do it somewhere besides DSL" list.
Anyone one standard deviation to the right of center would probably really enjoy the welcome thread, though.
Well, perhaps they do. I think it's worth noting all the libertarians that enjoy it there as well. As for me, I see nothing intrinsic to someone on the left that would keep them from spending time on DSL and thinking they got something worthwhile from it.
I'm curious how people interpret the behavior of a business that's sending me daily emails, usually with one limited-time offer after another, including reminders of the amount of time left on the latest special.
Note that IIRC, this was not their practice until perhaps last year, and I've been a happy customer of theirs for at least a decade, and subscribed to their mailing list. Again IIRC, emails were often announcements of new products, and specials weren't continuous.
I'm especially interested in any insights from people working retail, particularly a smallish online business which also has a few storefronts. Products are physical, and consumable - but slowly, in case that matters.
Well, they've finally got the response from me anyone who knew me would have predicted; I've unsubscribed from their mailing list.
I'll still buy their products; they are my go to place for herbs and spices. And I hope the constant specials are not a sign that they are in financial trouble, desperate for more business, as I'd rather not go back to getting stale overpriced herbs and spices from my local grocery store.
But I need to stop waking up to more emails in my personal mailbox than fit on a single screen.
Is it a small business? Maybe hired a marketing person? Big place? I blame consultants.
What newsletters do you have a paid subscription for, and why?
None currently, unless you count my local newspaper - which I mostly subscribed to to keep my housemate from risking covid buying it retail, at a point pre-vaccination when we were buying groceries only once a month, and soon started getting even those delivered.
I'm frustrated with the substack model. I want to interact with bloggers, and substack's commenting facilities are substandard; LiveJournal was better than this, 20 years ago.
As I see it, the problem substack solves is "how to get certain well known journalists more money, while taking a rake-off for the founders". It does little or nothing for the less well known, and I strongly suspect the sum of the individual subscriptions is commonly way more than they are in fact worth. I don't want to contribute to yet another winner-take-all / star system payment model.
Mostly, though, I'm frugal, and especially disinclined to take on recurring costs, such as subscriptions of any kind.
At any rate, if I do take on one more subscription, I'll probably go with the New York Times, in spite of excessive politicization and what appear to me to be declining standards. Even at its prices, it's more cost-effective than substack. Though when it comes to political posturing, I can and do get more than I want, for free, as email and paper mail, in spite of aggressively unsubscribing whenever possible.
So far only ACX and Freddie deBoer. The former for the excellent community and deep-dives into topics I don't have the skills to research for myself. The latter because he has repeatedly put into words the ideas that I have felt but couldn't articulate.
I'm open to subscribing to more, but none have the post output per month plus content I care for in the same way as ACX and FdB. There's one more that I'm on the fence about, but I haven't found myself reading it enough lately.
I've been a regular reader of a few newsletters. I tried ranking the ones I read regularly by the value I get, and this is where I ended up with:
- yourlocalepidemiologist
- astralcodexten
- bariweiss
- https://stratechery.com/
- persuasion
- razib
- greenwald
- www.theinsight.org
- hardcoresoftware.learningbyshipping.com
- fx
- glennloury
I think some of these newsletters have potential value that could be unlocked by paying for a subscription, but I have a hard time convincing myself to do that. At the end of the day, I've chosen to get a paid subscription for my top-1 newsletter; when something else replaces it and becomes rank no. 1 for me, I'll get a paid subscription to that.
I wish Substack had a bundle-subscription option like NetFlix so I could pay for more than one newsletter (versus the current model that requires N x ~5USD per-month in costs).
Yeah, a bundle subscription would be great for those of us trying to stick to a budget 🙁
I think there are a large number of content adjacent newsletters that would do well financially from joint substack. There are Substack tools for joint blogs although you can't bundle multiple separate blogs. So you'd give someone Contributor access or w/e permission and then you and your buddy or buddies could post content as a group. The cost-benefit would probably be complex when you also run an individual substack, though.
I’d probably spend around $500 on substack alone of I paid for all that I read. Compare it to Apple News* which is $120/year for family access of a host of traditional newspapers and magazines. Newsletters are shockingly expensive right now unless one’s interest is very narrow and limited to 1 or 2 subscriptions.
Thanks @Gunflint! I guess I wasn't clear - by 'newsletters' I really meant Substack style newsletters, and not traditional newspaper/magazine media. That said, it's certainly valuable to hear about the relative tradeoff here!
Way back at the beginning of the pandemic, people were publishing lots of articles and videos explaining simple models of how disease spreads, showing for example what happens in the Susceptible-Infectious-Recovered (SIR) model in which a population is uniformly mixed and randomly spreading a virus around, producing a neat little exponential curve that turns into sort of a bell shape.
I always assumed that these toy models were only just for explanatory purposes, not for any serious epidemiological research. The following article is interesting because it explores a non-simple model of disease spread, but it also makes a shocking claim that models "like" SIR are "the most commonly used".
https://cspicenter.org/blog/waronscience/have-we-been-thinking-about-the-pandemic-wrong-the-effect-of-population-structure-on-transmission/
Surely the average epidemiological researcher is not that stupid? So, is this article is being misleading/disingenuous, or are trivial models like SIR/SEIR *actually* widely and directly used by epidemiologists to make predictions, make recommendations and evaluate policy effects?
Aside from that one stunning allegation, the model presented in the article is interesting (but as the author notes, not necessarily accurate) because it produces vaguely similar hard-to-explain patterns like those we see in the real pandemic. Basically the conclusion this type of model suggests is that the course of the pandemic in a region *cannot* be predicted in detail without a lot of good data we don't have about how specific people are physically interconnected with other people in the real world. A corollary: attempting to evaluate the effects of public policy based on things like "cases before and after policy was put in place" can give more noise than signal.
Arguably the most interesting result from this model is that a variant can come to dominate in a population *without* being more infectious/contagious, which has caused me to wonder if delta/omicron are less infectious/contagious than they appear to be. But remember, there are *thousands* of catalogued variants and it's very rare that one of them "takes over the world", as delta did, which suggests that delta really is more infectious.
Thank for that link! I'm bookmarking it.
I like the fact that Lemoine's model can explain the phenomenon that I've been noticing — which is a clear downward trend in case numbers as certain variants become more prevalent. For instance Alpha didn't come to dominate either the US or UK's viral landscape until after the 20-21 winter surges. In fact, despite the popular delusion among epidemiologists that Alpha kicked off case surges in various countries, I can't find it dominating any surges until after the peak had passed. Likewise we see Delta only coming to dominate the viral landscape in Brazil after the Gamma surge. (OTOH, unlike Alpha, outside South America, Delta definitely creating surges all over the world PLUS it came to dominate the viral landscape worldwide.)
The best reason for using a mind-numbingly simple model is if you don't have enough data to fit a better one. We all know that an agents-in-a-network based model would be more realistic, but it's got a zillion extra parameters and you don't necessarily have enough data to fit them sensibly.
OK, but why don't we have enough data? This isn't a COVID-specific thing; this is something the epidemiological community has needed since the 1980s at least. If you're trying to model an epidemic, whether HIV or COVID or anything in between, you need data on how people actually interact with one another, not spherical disease vectors randomly colliding in zero-dimensional space, but actual human behavior patterns. And it is data that would have been easy to get to at least a first order through surveys and the like; possibly something that could have been piggybacked onto existing surveys like the GSS. And then validated by comparing model to reality in e.g. annual flu outbreaks.
So why, after forty years, are epidemiologists still working with zeroth-order models?
Just a reminder, I did not accept the premise. Wondering if someone would want to confirm or (better yet) deny.
First off, how accurate are your R0 values? That's the heaviest weighted variable in most models, and hardest one to determine. I'm still seeing wide ranging R0 values of SARS-CoV-2. In the early stages of the pandemic, I saw values as low as 1.5 and as high as 5.0, but it was population and nation dependent. B.1 barely touched SE Asian and African populations, but it ripped through European populations. Some populations seemed to have a higher natural immunity against B.1, and there were some cool studies correlating HLA frequencies and IFRs. But all those became moot when B.1.617.2 (Delta Classic) hit the scene. Previously "resistant" populations were getting hit hard by Delta, but the R0 for Delta was still showing up as between 2.5 and 5 depending on the country (and depending on the research team performing the "calculation" — R0 = wild ass guess?). And worse yet, I never saw *any* epidemiologists publish their reasoning for how they calculating their R0 values. Nowadays the generally accept R0 for Delta is 3.0. But is that accurate?
Anyway, right off the bat you've got an uncertain variable underpinning your model. You might say — well, run your model with a range of R0 models and see which fits the actual data best. But, it's pretty clear that the reproduction value of COVID-19 change over the course of an outbreak. For instance, with a R0 of 3.0 the virus should keep spreading until roughly 70% of a population has been infected. But none of the surges have infected more than 10% of the population before burning out. Why? One of the theories goes that once the density of people with convalescent immunity gets beyond a certain level, an R0 of 3.0 can't keep propagating. Which has raised all sorts of questions about the connectedness of different groups of people. For example, how do you calculate say the connectedness between Hassidic Jews and Blacks in NYC? Or even at a more basic level, the differing connectedness between family members, friends, coworkers, and fellow community members.
An alternative theory is that once a surge starts happening, people start taking more precautions (masks, social distancing, etc.). But how do you quantify that variable? Oy, my brain hurts!
Anyway, modelers seem to be very invested in their models. But none of the modeling predictions that the CDC publishes every month have been accurate for more than 2 weeks out. That makes sense to me. We can't predict the weather more than 2 weeks out, why should we be able to predict the course of an epidemic?
Blegging for a recommendation for a Silicon Valley GP that's rationalist-adjacent(-ish?). I don't expect a full-send, but anyone even marginally closer to the community or that others get along with would be super appreciated.
My therapist recommended that I try to meet women at Asperger's/HFA support groups. Has anyone here done this? No offense, but I feel like this would be a good place to ask.
There seem to be a lot of programmers here. And it's often said that there a lot of autistic people here. Why doesn't someone program something like Bumble, a dating app where only women send messages, specifically for autistic women who are open to dating autistic men?
Because isn't something called a support group meant to be a low-stress place for attendees?
Both me and my girlfriend are autists. We didn't try to find autistic partners on purpose, just happened to have similar interests. Actually, we didn't even know that we are on the spectrum until we started dating. It was funny, really. At one moment she started investigating whether she is autistic, found lots of evidience in favour and I was like: "Pff, this doesn't have to mean anything, I mean I fit like 80% of this as well, so what, does it make me autistic now?" Took me some time to consider the possibility that the answer is actually "Yes".
We have pretty good and deep relationship. We explicitly talk about our feelings, do not need to mask and in general have short inferential distances and understand each other really well. For her it's the best relationship she has ever had. For me its at least top 2: more challenging but also more fulfilling. I think it has something to do with the fact that I'm more high functional than she is, and nearly every of her previous relationships was a mess which gave her couple of traumas.
This could make an interesting essay, if you wanted to write more. It reminds me of a New York Times story that I read a decade ago (https://www.nytimes.com/2011/12/26/us/navigating-love-and-autism.html).
Wouldn't those places be very dense with males already?
I would expect so, yes.
I would not recommend this, I think people overvalue similarity in relationships. 2 aspergers types will frequently butt heads in my experience and you don't exactly want someone who shares your weaknesses in a relationship. Instead I would recommend pursuing women who grew up with autistic siblings. I say this because a while ago I looked back at the set of women I had gotten along well with and noticed that an absolutely statistically improbable number of them had autistic siblings. I suspect that growing up around someone helps them not be so off put by certain behaviors (in women unfamiliar with neurodivergent types, you can easily trigger Anamoly detection filters and just get thrown in the "creepy" bucket)
Love the concept of "Anomaly detection filters"!
Given the heritability of autism, isn't there a big overlap between "is autistic" and "has autistic siblings" anyway? This sounds to me like the best partner would be someone who is on the autistic spectrum, but has little or no visible problems.
I am not opposing your advice, though. People who are on the spectrum but have little or no problems will probably not join a support group. But you could find them by metting the people in the support group... and then trying to meet their siblings.
Oh boy. This sounds like it could go horribly wrong, then again a lot of my paternal family are 'on the spectrum' yet they managed to get married, so it could work!
I think there might be trouble communicating romantic interest with autistic women, but what do I know, it may well be many of them are open to finding love and partnership.
> I think there might be trouble communicating romantic interest with autistic women
I guess a proper pickup autist might start like this: "Hello! I am romantically interested in you. If you are not interested in me, that's okay, no pressure. However, if you decide that you would like to explore this topic further, here is my card, send me an e-mail, and we can have a dinner together."
(Sorry, my game is a little rusty.)
hey baby, do you publish your API?
"Pickup autist"
I see what you did there!
I'm kind of on the spectrum myself, and back in the days of yore when anyone might have been interested in approaching me, you really would have needed to go the "here is a written sign saying I want to ask you out", as subtle hints and non-verbal signals would go right over my head.
(Not that I would ever have said 'yes', but part of ignoring any possible approaches was simply not twigging that they were possible approaches).
In my RSS reader, whenever I scroll past an item, it's faded out to mark it as read. Does anyone know of a similar Chrome extension for Twitter (ideally the read tweets would be synced between computers)?
The best workaround so far is to use Tweetdeck, in which you can clear the column containing your timeline after you've read it.
Fun fact. The Spanish flu - which was a novel virus at the time - never went away. Modern seasonal flus are direct descendants of that. What happened is that more transmittable strains were less fatal. Could be good news for Omicron.
My understanding is that influenzas descended from the Spanish Flu were the dominant seasonal flu strains for several decades after the 1918-1920 pandemic, but they largely died off in general circulation in the 1950s. The pattern being that pandemic flus tend to be followed by endemic seasonal flus descended from the original pandemic strain.
The 1957 H2N2 pandemic was the dominant seasonal flu until the 1968 H3N2 pandemic. The latter's descendants are still in widespread seasonal circulation (or were, until the Covid lockdowns cancelled the last couple flu seasons), but were joined by a different H1N1 strain (same general variety as Spanish Flu, but not necessarily directly descended from it) from the 1977 flu pandemic, and then by yet another H1N1 strain from the 2009 flu pandemic.
There were also two families of Influenza B among the seasonal flu, although I don't think either of them have been traced back to any known flu pandemics; instead, they seem to have been endemic seasonal viruses since time immemorial (in the colloquial sense, not the specific legal sense of having been demonstrably so since the ascension of Richard the Lionheart as King of England). One of these two families, the Yamagata lineage, looks like it may have died out due to the Covid lockdowns.
The pattern of Influenza A pandemics developing into less-deadly seasonal/endemic flu strains is partially hopeful and partially dreadful sign for Covid. Hopeful because less deadly is obviously a better outcome from our perspective than more deadly, if we must have an endemic virus. Dreadful because "less deadly" isn't the same as "harmless" or even "mostly harmless". Tens of thousands of people in the US alone die of seasonal influenza each year, which isn't an ideal circumstance.
And it's only a weak signal for what to expect from Covid, since while SARS-COV-2 and Influenza A are both viruses that cause upper respiratory infections in humans and can cause deadly pandemics, they're not at all closely related viruses (different phyla: SARS-COV-2 is a positive-sense RNA virus, while Influenza A is a negative-sense RNA virus) so we can only infer so much from one's evolution as an endemic human virus about what to expect from the other. Endemic/Seasonal Covid might be a moderately deadly threat like seasonal flu, picking off tens of thousands of people mostly from among the old, sick, and immunocompromised. Or it might truly become mostly harmless, like the other endemic human coronoviruses that are among the causes of the common cold. Or it might remain deadly at some significant fraction of the deadliness of current Covid strains among the subset of the population that's vulnerable to infection.
Great comment 👍
My understanding is that flu reproduces sexually. The H and N antigens are modular. When a new strain is introduced from pigs or birds, I think what happens is that it keeps most of the virus specialized for humans, but swaps in new antigens, which humans haven't seen for a while.
That's horrifying/fascinating!
I didn't know about the recombination part, so thank you for pointing that out.
After digging a bit, I see that Influenza A has eight RNA strands that seem to work kinda like mini-chromosomes. Two of them contain the genes for the HA and NA proteins respectively (variants of which give the H and N numbers by which the strains are traditionally categorized). The strands do indeed mix and match when the same host has concurrent infections by different strains.
From what I gather, three of the eight strands differ radically between common seasonal Influenza A varieties (the HA and NA strands, plus one more) while the other five are much more closely related. The 1918 pandemic strain's genome, at least the parts that we've recovered of it, seems to be near (but not quite at) the base of the philogenic tree for both the HA and NA strands relative to modern seasonal H1N1 and for at least one of the other strands common to all modern major seasonal Influenza A varieties.
The standard interpretation seems to be that the five common strands and the H1 and N1 strands are descendants of a recent ancestor of the 1918 pandemic strain. The 2009 H1N1 strain is believed to have come from a descendant of the 1918 strain that had become endemic in North American domestic pigs during the original pandemic, and the 1977 strain is suspected to have come from a stored sample of 1950s-era seasonal H1N1 that got loose while being studied in a Soviet lab.
What do you mean by reproducing sexually? Viruses reproduce by tricking cells into producing more viruses.
https://www.cdc.gov/flu/about/viruses/change.htm
If the same host is infected with two different strains of influenza at once, you can end up with genes from each one, and potentially get HA/NA genes that weren't circulating in humans before and so nobody has any immunity to them.
Yeah, I shouldn't have used the word reproduction. Sex is never for reproduction, but only for exchanging genetic material. Some species have mechanisms to force sex before reproduction. Maybe other species shouldn't be said to reproduce sexually, but that really draws the line in the wrong place.
Bacteria have a lot of adaptations for sex, but viruses only exchange genetic material incidentally, when two viruses infect the same cell at the same time. But some viruses have some adaptations to make the mixture more useful. In particular, flu has a modular genome, to make it easy to swap one H for another and separately N. While the host-specific adaptations are separate
I had a long talk with a biologist about this, and I think it's somewhat wrong.
Flu viruses are constantly mutating. They can't not mutate. Their genetic code is unstable. Only the evolutionarily imperative to be good at what they do (infect people) keeps them even mildly similar to each other.
The Spanish flu was very, very good at being a flu virus. But when it infected everyone in 1918, everyone got immunity to it. So continuing to be the Spanish flu was no longer an option. It mutated into other flu strains, including the ones we have today. So it's true that all modern flus are descended from the Spanish flu.
But that doesn't mean they're "better than" the Spanish flu from a flu-fitness perspective. It's entirely possible that if we dug up the old Spanish flu from some laboratory and set it loose, it would give us another 1918-level pandemic event, much more severely than any existing flu, because it really is just an incredibly successful and effective flu virus, and nobody has immunity to it anymore.
Existing flu strains can't mutate "back" into the Spanish flu because it's too small a target. It took evolution however many thousands of years to invent Spanish flu, and it will take it another thousand years to reinvent it, even though it still has some of the pieces left over.
Not to argue against you or a biologist who knows more about viruses than either of us...
But one factor that did not exist for most of the thousands of years pre-Spanish-Flu was Industrial-Age warfare. (The first death was a cook at a US Army training facility in Kansas. Not sure whether he was Patient Zero or not. The spread of infection from that point depended on the size of the Army training camp, and large numbers of people traveling via railroad/steamship to many other places in the United States, and the rest of the world. Cases were seen on the East Coast within a week or two, and cases were seen near the Western Front in Europe within eight weeks.)
There may have been earlier versions of various viruses which could have done a Spanish-Flu style outbreak, but weren't able to. After all, if the nations of the world are not moving large armies across the globe using Industrial-Age transportation tools, the outbreak might be local and take many months to move across a single nation.
It's very hard to tell what various Influenza viruses were capable of before the Industrial Age transportation networks allowed carriers to circle the globe within a few weeks.
That shakoist substack post linked was especially good. I am not sure he explicitly pointed it out but it seems like the central limit theorem strongly argues against the methodology used by ivmmeta.com people. If all of the studies (they include) were equally valid then they should nicely form a normal distribution about the actual value of ivermectin's effectiveness. That they do not, ought to invalidate the methodology of...pretending that they do.
Question for the resident psychiatrist: How does fluvoxamine look side effectwise?
(Background: triple vaccinated but zero antibodies due to immunesuppression, contemplating it as preventive measure, prior experience with SSRI and SNRI almost a decade ago)
It's a pretty standard SSRI and my section on SSRI side effects at https://lorienpsych.com/2020/10/25/ssris/#5_What_side_effects_might_I_get_on_an_SSRI should apply.
Thanks, it just does not seem to be used all that much so was assuming there was some reason for that.
I just learned that large passenger planes have their own radars that can be switched between weather mode and terrain mapping mode (this explains where the pilots get their information from when they announce over the intercom that turbulence is expected in X minutes). I assume the pilot flips a switch to toggle between modes.
I found videos showing what the radar screen looks like in weather mode, but nothing for terrain mode. Can anyone find one for me?
Also, how often do commercial pilots use radar terrain mode, and under what circumstances do they generally do it (e.g. - when lost, when approaching for landing)?
Do the radar computers come pre-loaded with detailed topographical maps of the whole planet, allowing them to automatically deduce where the plane is based on what the radar sees in terrain mode, or are pilots left to look at the screen and figure it out by comparing the image to paper maps in the cockpit?
I assume it is used to avoid mountains.
I can't imagine how a large passenger plane could get lost and for landings the amount of ground clutter reflections might make it more trouble than it's worth given that altitude radar plus orientation instruments exist (and glide path beams now that I think of it).
Why would they have to do any of that when GPS exists?
GPS gives you position, does not give you a mountain height.
I would be interested to see which passenger flight routes pass over mountains at a height where that would be a concern.
Guess before reading further: how many (% of all) commercial airplane crashes between 1993 and 2002 were caused by fully controlled airplane crashing because pilots failed to notice they are flying into ground or water? Where it would be completely avoided as plane was steerable?
Which % was between 2008 and 2017?
1) when you land you want to end just at ground level - not higher, not lower and airports have various ground level
1a) In normal operation only during landing/take-off plane is low but not all operation is normal
2) some airports are near mountains. Following are extreme cases
- Lukla Airport (LUA)
- Courchevel Airport (CVF)
- Toncontin International Airport (TGU)
See https://en.wikipedia.org/wiki/Controlled_flight_into_terrain in general - "the second-highest fatal accident category after Loss of Control Inflight (LOCI)."
"CFIT was identified as a cause of 25% of USAF Class A mishaps between 1993 and 2002."
And between 2008 and 2017 "CFITs accounted for six percent of all commercial aircraft accidents"
It absolutely is a concern at takeoff and landing near mountains, also depending on where ATC directs you at those times. Such airports exist.
Presumably your navigation system knows where the mountains are, so if it knows where you are it can tell you what mountains are nearby.
I doubt GPS is fast and precise enough to pilot an airplane in rough condition or for an emergency landing. And I'm not sure GPS coverage over polar region or large oceans is top notch either.
GPS is fast and precise enough to avoid mountains, at least mountains that haven't moved since the last time the terrain database was updated. They are also precise enough for instrument landings in adverse weather. Source: own one, use it for avoiding mountains and landing in adverse weather. GPS also works over oceans and polar regions, not that it matters for this purpose due to the lack of mountains to dodge or runways to land on.
Nobody uses on-board radar to directly land airplanes, in emergency or otherwise. The military does use it for terrain avoidance; I haven't heard of civilian aircraft doing that but it's not absurd and might be a reasonable backup for GPS in remote mountainous areas.
Is worldwide terrain database good enough to use just GPS? I assumed that in some areas or during emergency on-board radar may be more likely to be used.
You can download 30x30m terrain data for most of the world right now.
SRTM? Is it really good enough? Even for maps it is often lacking due to voids, gaps, flattened peaks and so on.
GPS coverage over the polar region and large oceans is top notch; it's better than GPS coverage anywhere else, because you have respectively more satellites and less interference. GPS also works much better in airplanes than, for example, in cars, because there's less crap above you to block it. Its precision is a few meters, which is plenty fine for an emergency landing if you have some way to tell how far away the ground is, such as a radar or visibility over 10 meters.
Given that the US government made GPS freely available for use worldwide after a Korean Air passenger flight was shot down over USSR prohibited airspace in 1983, precisely to prevent such events in the future, I'm pretty sure it's fast and precise enough to do all of those things.
[Accidentally navigating hundreds of kilometers out of course] and [making an emergency landing] are not similar events. GPS prevents the former well but don't help that much with the later.
Correction, the US government *would* make it freely available, it was still in development then.
Also in existence are ground-based navaids (VORTAC/DME, NDB). Some vehicles (unsure if any civilian) can also do star fixes in broad daylight.
High school senior here, planning to attend college. A few wonderings I'd love to have some perspective on:
1) Is a Computer Science degree as could as people say it is? I've tentatively decided on CS, but I've heard alternatively that there's either a crippling shortage of CS professionals in the US, or you can spend months on the jobsearch and the interview process could be a new circle of hell. I'm also specifically interested in cybersecurity; how does that compare with CS?
(To be quite honest, I'm also worried that if I go into CS I'll never get a girlfriend. Sue me, I'm a teen.)
2) What should I direct my effort toward in college? I've heard that college isn't useful for the education, but for the signalling power a degree holds. So rather than academics, what should I funnel my effort toward? "Enjoying my youth"? Building ties with competent like-minded peers in preparation for the future? Building work experience with internships? Working on personal projects?
3) How can I up my conscientiousness/build self discipline? My test scores are always excellent, but my gpa is abysmal in comparison. Part of that is being hit by depression and later a behavioural addiction to reading, but I'm also just lazy and weak-willed. 'Executive Dysfunction' was suggested back when I had a therapist.
4) General college/job/life advice? I can't see myself sticking with one career for the rest of my life; I'd rather switch it up every few years. I'd also like to travel, and am thinking of attending a year of college abroad, perhaps in Germany. These may just be juvenile sentiments that will pass, however, and in the case that I'm romanticizing something that sucks in actuality I'd like to know.
Can confirm that buggering off to Germany in the middle of a degree is a good idea. Though it was before Brexit so I had the opportunity to work there.
I'm responding primarily to question 1. For context, I'm a software engineer on the cusp of retirement, and in the prosperous/high status end of the income distribution, though not at the very top.
1) A computer science degree from a good tech college, or a very well regarded generalist school, will almost certainly get you employed after college, with decent or better pay. It will also put you on a path with potential for very good pay indeed, either as a senior high end technical person, or on the management track.
A coding course from an also-ran outfit will probably keep you employed, but the first job will be harder, and you won't be a candidate for the high end.
Be aware that there's a huge range of salaries in this area. The funnel for the best US jobs takes most of its input from MIT, Stanford, etc. Waterloo if you are Canadian. This funnel starts with internships during college, which are vitally important. Also, the best career paths now generally require a masters degree.
Essentially all the male software engineers I know are married; most of the females are too. A good salary can be a potent aphrodisiac, at least for those seeking a long term relationship.
Some of them wind up repeatedly divorced, due to putting their careers ahead of their relationships. Be prepared to work long hours, especially if you are ambitious.
I know nothing specific about cybersecurity.
2) I learned a lot in college, and am glad I paid attention to my courses, but I'd probably have earned more over the course of my life if I'd paid more attention to building long term relationships with people outside of my field. Contacts turn out to be vitally important.
Internships and contacts made on them will, among other things, quite likel get you your first job in CS.
4) I suspect CS will be a decreasingly good career as time passes. I certainly liked it better 20 or 30 years ago, but part of that may be having been promoted to a level where I have to deal with a lot more things I consider BS, rather than the code, which I love.
It's probably good that you see yourself as moving to new things later in your life.
I'm a big fan of learning languages. If your German is remotely good enough to manage, go to Germany and make it a whole lot better. But if you do, don't hang out mostly in an American enclave, speaking English, even though that will be very tempting.
> Be aware that there's a huge range of salaries in this area. The funnel for the best US jobs takes most of its input from MIT, Stanford, etc. Waterloo if you are Canadian. This funnel starts with internships during college, which are vitally important. Also, the best career paths now generally require a masters degree.
FWIW, I got a FAANG job right out of college with just a CS bachelors at a state school. Internships are important though.
Try out advent of code (https://adventofcode.com/) and see if you like it. Married CS person here, seems to work out as long as you know, you talk to girls sometimes and don't just stay in front of the computer.
If you're really concerned, make sure not to skip out on the gym. Strength training can be fun too. Or dance, as someone else said.
Cybersec is cool if you can get into it, and if you are interested in it, don't wait till third year for the school to teach it to you. CS is a really really wide field, and school will try and build a foundation and then show you some stuff (e.g. one graphics class, one database class, one security class, one UI class, one AI class), but none of the classes will really go as deep as you can go on your own if you put the work into it. And honestly, the lectures aren't always better than youtube tutorials and trying stuff yourself.
However it is important to understand enough about how computers work, what memory is, why does this algo take O(n) vs that one taking O(n^2) time. And the curriculum should make sure you understand all that by the time you graduate.
I think my Math and Chinese language classes were "harder" in the sense that I don't think I could have learned the stuff from them from the internet / textbook. So sometimes you might decide to just study CS stuff for fun and take the classes which are harder to learn on your own. (For me, I also failed to learn graphics and OpenGL on my own, so I gladly took the class.)
Re:3
One piece of advice I heard from someone about motivation and such goes as follows.
"Even if I don't feel like it, I always at least get ready at the gym and tie up my shoes. If after that I still don't want to do it, I have my own permission to switch right back. But often, once I am fully prepared to do it, it doesn't seem so bad, and I can go and complete my workout"
I think that approach works quite well, where you make some time, say "I will at least get everything set up here, maybe set up my editor, turn off the internet (Download the docs ahead of time) and set a timer for 15 minutes"
If after all that, you still don't get any work down, well, go for a walk or something and try again in a bit. But often I have trouble with the putting down the distractions and less with the picking up with the work.
You're getting lots of great replies already, but I'll add a few thoughts:
1) CS is good, and learning to program IMHO is one of the most valuable skills you can have. Basically every career involves computers already, and it's safe to say computers will be used even more in the future.
2) Do all that stuff you mentioned! Include a mix of potentially useful stuff (e.g. internships) and just-for-fun stuff (e.g. anime club). Some stuff will be awesome, some won't work out, and college is the time to experiment and find out.
3) Learning about yourself is just as important as academics, if not more. I'd highlight mental health as being something particularly important to work on, since nothing else really matters if your brain isn't in good shape. Campuses usually have mental health resources available, don't be shy to try them. That said, for self-help you can also check out https://www.lesswrong.com/posts/JgBBuDf5uZHmpEMDs/how-you-can-gain-self-control-without-self-control.
4) It's hard to give much general advice because everyone has different values; there are a lot of perfectly valid approaches to living life! I'd say first, take advantage of what college offers you. Go to office hours, go to career fairs and that sort of stuff, socialize, take elective classes that sound cool even if it's way out of your usual field, and try some clubs/activities. Second, keep an open mind about careers. It's better to think of college as a chance to collect information on potential careers, rather than training for a career. A lot of people end up happy in a career very different from what they could have imagined during college.
Married to a software engineer. We have a nice life. It helps that he’s good at what he does, and he’s good at what he does because he likes it. I can tell when he’s been writing code because he has a spring in his step.
As for girlfriends, my standard advice is to hang out around the library science program. It’s 95% female, and I know a lot of librarian-programmer pairs. I think it has something to do with CS people liking communication to be specific, and librarians being able to rise to the occasion.
Strike a balance between doing things you enjoy and that are useful to you (e.g., making and saving money, maintaining good health, finding and keeping friends you truly like). Ideally, often these two areas overlap, to people's points about figuring out whether you enjoy CS. Not everything needs to be enjoyable, and not everything needs to be useful.
Just because some people don't learn a lot in college doesn't mean you shouldn't. Try to learn a lot, whether it's the material itself or the meta-skills (how to learn hard things, how to write and communicate well, etc.) This is also a good way to build up your discipline, along with an exercise habit.
Find people you really like and keep them in your life. Old friends who have known you for decades (when you're older) are among the best things in life, so find those people now. That may mean you need to meet a lot of people in college. Don't assume the first people you meet are the ones you'll click with the most. In both college and grad school, I felt a little out of place for a year or more and then found great friends that I've stayed in touch with ever since. Try out different activities and groups and see what and who sticks.
Travel, a year abroad, and whatever enthusiasms you have now are definitely worth pursuing. You only get one life, as far as we know, so make sure you have some fun stories to tell when you're older. Even many experiences that suck in the moment will seem interesting and enjoyable as memories. So don't try to be *too* mature and practical. There's a place for that, but make sure to really live life and do some weird things.
You definitely don't need to stick with one career if you're reasonably smart, likable, and a good communicator. In fact, having multiple skills sets that overlap in interesting ways often make you more valuable on the job market.
> there's either a crippling shortage of CS professionals in the US, or you can spend months on the jobsearch and the interview process could be a new circle of hell.
Both could be true at the same time.
> I've tentatively decided on CS
Have you tried going through a programming tutorial?
If you can do this then it is a strong hint that it is a good field for you.
If not - can you do this with a tutor? If not, then it is a strong hint that it is not a good field.
Have you programmed for fun? (another strong hint).
> Building ties with competent like-minded peers in preparation for the future? Building work experience with internships? Working on personal projects?
Good ideas. Note existence of projects such as Google Summer of Code or Github Student pack. Look for more like that.
I would add "join interesting open source projects"
> 3
Please let me know if you have good hints here.
EDIT: obviously also enjoy life - as long as it is not involving life-destroying narcotics or harming others. Explore, travel, dance etc.
If you're good at it, CS is the best thing ever. It's the closest real life equivalent to being a Wizard developing magic spells. You should probably start learning to program before you get to college so you can get a feel for it and decide whether it is something you want to pursue or not.
As for "getting a feel for it" and "being good at it", don't be alarmed if it's hard and you can't figure things out. It's hard for everyone and it's always hard to figure things out, but especially when you start; nobody is born good at it.
1) I don't know if you enjoy computer programming. CS will open the door to making lots of money, but if you intrinsically enjoy programming, most of the process will be at least somewhat enjoyable; if you hate programming, it will be torture. Cybersecurity seems like a specialization that you can choose later. (But I am just guessing here.)
To find a girlfriend, learn to *dance*. It is a great balance to the sedentary CS lifestyle, and not only you will meet lots of girls, you are allowed (and socially expected) to touch them and do something that seems like a vertical simulation of sex (somewhat exaggerating here, but not too much, depends on the specific dance). So you essentially skip the first few steps of dating.
2) Yeah, it sucks that the part of your life when you have most free time is also the part when you least know how to use it well. Try a bit of all, I guess. Explore. At this phase of life, learning is more valuable than doing (you will have the following decades to profit from what you are learning now)... with the exception that sometimes doing things is also a source of learning. Just don't spend *all* your time doing *one* thing; that is, if you e.g. start an internship, don't let it consume all your time.
3) https://lorienpsych.com/2021/06/05/depression/
4) Learn German e.g. here: https://deutsch.info/ and get good at skills that are useful across a large range of professions, such as communication and math. If you know German and programming, you can easily find a well-paying job in Europe. (English and programming is already quite enough, but if you add German to the mix, it is even better.)
Seconding the dance recommendation--swing dance clubs are incredible for building social confidence and spending a lot of time interacting closely with the opposite sex.
Pay is generally a lot lower in Germany than the US though.
Speaking very generally...
In USA:
- huge salaries;
- more business opportunities.
In Europe:
- free-ish education;
- free-ish healthcare;
- actually 8-hour workdays;
- more vacations;
- more maternal/parental leave.
So the optimal strategy is to be born in Europe, study in Europe, move to USA and make money until you burn out, return to Europe, retire early in your 30s, start a family.
If you are already a teenager in USA and feel adventurous... I am not sure whether it would make sense economically to learn German, move to Europe and study there, then return home. As a foreigner, you would probably have to pay something for the school. Also, travel and accomodation are not free. But given the astronomical costs of some American universities, is there a chance that it might still be cheaper?
Travel and accomodation are not free, but college education is free or costs little in a lot of European countries, including Germany (except Baden-Württemberg and parts of Sachsen, where it costs ~3000€ per year for non-EU students).
Accomodation cost can vary extremely, depending on the city. If you don't have excessive tuition fees like US or UK, this is by far the most expensive part. But you have accomodation cost regardless of whether you study in the US or abroad.
Do you enjoy problem solving and do you not enjoy people? Then major in CS like me, because every other job is either dirty (I would prefer to sit in an office), about people, and/or has way less problem solving.
It's not the most fun thing in the world, but I actually kind of get a dopamine hit doing leetcode (and DSA type problems are one of my least-liked areas of CS). That doesn't happen for me when I'm memorizing biochemical pathways. I was going to go to medical school but I realized that it would be hellish, because I hate memorizing factoids. I like understanding complex concepts that you can't just google and I like solving problems. The existence of programming is pretty much a miracle with regards to trying to find a problem solving job.
And to comment on some other comments nay-saying CS. Yes, CS is not perfect. Your job might feel like a job. That's sadly normal. But what's the alternative? Other forms of engineering? Most of them end up writing code. I was doing premed + an engineering major, and I realized that medicine is very boring, and that engineering means you will probably end up coding, and if you don't, your job will be no more interesting than programming, and will probably be dirty, i.e. in a plant or a construction site or something gross.
So yes, there are some people below basically talking about how corporate meetings and making webshit and Javashit is boring. I agree. The thing is that there are no alternatives I can think of that CS doesn't provide a pathway to. For me the alternatives would be self-employment, teaching/research, or some sort of employment that is low on meetings and high on interesting work.
I can't think of a field that qualifies for the last thing more than CS. There is a lot of webshit being produced these days. I don't find it very exciting to write corporate applications in high level languages compared to doing harder, more cutting edge stuff. But the only better option outside of CS I can think of would be researching another STEM subject. Guess what? If you want to be a professor, you can be a CS professor. It's probably easier to get a CS professorship or private research position because the field is expanding more than the old sciences.
I can't think of a job that I would find more interesting than coding other than other STEM research, and since CS gives me the opportunity to do such research in the future anyway, it's pretty much a no-brainer for me to do CS.
As for degree difficulty: if you are good at math, CS is the easiest STEM degree available, period. Have you programmed before? If not try building a calculator in Python or something. If you make As in math, and Python kind of fun, you will find CS super easy. If you are not good at math, however, a memorization based "STEM" degree might be easier. I think there is only one of these, depending on your definition of STEM. Maybe two. Biology has the absolute least math, then Chemistry. For Chemistry, you will take a truncated version of chemical physics. CS probably comes after Chemistry in terms of the math involved.I would say if you aren't good at math, CS is probably the third easiest STEM degree after Biology and Chemistry.
And I don't think it's necessary to compare to humanities degrees. Frankly, those are pointless and you shouldn't have to waste your time with them. Although in the job market right now they serve as a ticket to random office jobs that need a certain IQ level but not any skills. You could easily do them in 8th grade, but you need to get a business degree or whatever to show that you can read and right at at least a slightly above average level. If you're really smart though, you need to be majoring in STEM, and if you don't want to be a professor, it needs to be engineering or CS, or maybe math or statistics.
>2) What should I direct my effort toward in college? I've heard that college isn't useful for the education, but for the signalling power a degree holds. So rather than academics, what should I funnel my effort toward? "Enjoying my youth"? Building ties with competent like-minded peers in preparation for the future? Building work experience with internships? Working on personal projects?
Maintain a high GPA (aim for a 4.0 and if that fails aim for Summa Cum Laude). Education, for most majors, is a hollow signal. For CS, you will actually learn a lot of useful things. For STEM in general, if you get a job in your field, you will actually learn useful things. Business/humanities degrees are hollow signals unless you become a professor of those things.
Education being a signal means your GPA matters a lot. Education being about skill means it GPA matters less. In my experience, your GPA matters some amount which is proportional to how much of a hollow signal your degree is. The floor, however, is still really high. So GPA still matters a lot for CS and engineering.
I have this rich uncle who's a dentist, he told me the number one thing that matters in college is your GPA. So far my experience has proven him correct. You can recover from a low GPA to some extent proportional to how useful your degree actually is, but it's still a recovery. Don't make yourself have to recover.
As for what to do besides your GPA: research and internships. It's that simple. Nothing else will go on your resume. This isn't high school, you're not applying to college, you don't put bowling club on your resume. Do it if you like it, but keep in mind it's all about research and internships when it comes to building your resume. For CS, you also have a portfolio you can build from your basement, called github. Isn't that cool? You should put that on your resume too.
>I'm also specifically interested in cybersecurity; how does that compare with CS?
AFAIK there are no "cybersecurity" bachelor degrees and if there are you probably don't want one. I go to a good public school and they don't have one. The cybersecurity classes are CS electives. I am taking them because I also like cybersecurity. However, the field is somewhat hard to break into. You want to get a CS degree with cybersecurity classes on the side, and then potentially follow up with a masters if you feel like you need it. The reason why is that the jobs are very rare and you box yourself in with a "cybersecurity" degree.
>I've heard alternatively that there's either a crippling shortage of CS professionals in the US, or you can spend months on the jobsearch and the interview process could be a new circle of hell.
Same here. I think it comes down to competency. The job market looks good statistically speaking. Sometimes I hear people who start coding and a year later they're making $100k doing it full time, sometimes I hear about people who hardly get interviews and then bomb them when they do. I've seen resumes people post and there is a definite correlation with competence. https://old.reddit.com/r/cscareerquestions/comments/3e55c8/interviewers_can_people_really_not_pass_fizz_buzz/ctbowp4/
For whatever reason, maybe because there's no licensing, maybe because they hear CS is friendly to people without CS degrees, there's a lot of people who literally cannot do the most basic tasks, and who have no capacity to dynamically learn on their own, who apply for jobs. I think if you're above average it will be the most ripe, 1360s esque job market you will ever see. Signals like GPA and portfolio will help you get interviews and competence will help you pass them. People who can't fizzbuzz with no experience apply with a 2.8 GPA and the wrong skillset to 200 FAANG jobs and then flunk the interviews and then they whine that they can't find the job.
> making webshit and Javashit is boring
It is not like there are no other jobs. It may be less paid or something but it is definitely there.
1) Plenty of cybersecurity professionals have a CS degree; think of it more as a specialization than a separate career path. Regarding the job market, there are lots of jobs available, but lots of people looking for jobs suck. I wouldn't hire at least half of my graduating CS class for any coding job. Additionally, the interview process at most companies is tuned to try and make the false positive rate low at the expense of the false negative rate, and it's not difficult to fall into a "would be good at the job, but is rejected at some stage in every interview process due to some random signal" trap. This is where having personal connections really helps.
If you want to meet women, choose a woman-dominated area that you're interested in, and minor in it. Or join clubs that aren't CS related. Also, you'd be surprised how much Electrical Engineering and Math have better gender ratios than CS—not 50%, but in my experience it was better.
2) College gives you a great opportunity to educate yourself; choose the classes that interest you, and that will provide useful skills in your job, and throw yourself into them. Additionally, if you do decide to go into CS, the more experience you have programming, the better. Do hackathons, personal projects, try to contribute to open source projects. Get as much time coding as you can. I imagine there are similar skills in other disciplines, though I don't really know. Beyond that, try to make friends who will work in the same industry as you! It's way easier to get a job somewhere if you have a friend on the inside who can chat with the hiring manager.
3) I also struggled with this. One of my major issues, which may or may not apply to you, is that I would sometimes not turn in half-finished homework—get comfortable turning in things halfway done. 50% is way better than 0%. (If you're having difficulty with this, try writing "Sorry, didn't have time to finish this" at the top of the homework. Not sure why that helped me, but it did.)
4) This is much easier to do if you go make 100k in CS while spending 20k for a few years after college.
Along the lines of turning in halfway done homework, I am a IC (Chip) designer. I did this type of thing even on tests. Often in physics and occasionally in math (calc, etc). I would hit a point where I lacked an equation and say, " Not sure what this equation is but if I was, this is what I would do next....." You can get a lot of points for that. Even in math, showing you know the process is far better than an incorrect guess or just unfinished work. (I would don't know how to differentiate this, but if I did, I would and then I would set it to zero to find the max/min of the original function)
1) Yes, it is an excellent choice for job prospects. Cybersecurity is fine, but you have 2-3 more years before you make that decision. No need to worry about it now.
For finding a girlfriend, yes definitely, CS has very few female students. Obviously, there are other ways to find a girlfriend, but it is also true that you will spend a lot of time with your fellow students. If you are really concerned about it, you *could* consider starting with a related subject, and switch to CS later. For example, math has much more balanced gender ratio than CS. But this only makes sense if you already have a high interest in math. I would only recommend this if the male/female ratio in CS really troubles you a lot.
2) I agree with the others. Take your studies serious. You will learn a lot of useful and a lot of (for you) useless stuff, but it's hard to predict in advance what will become useful for you.
3) Building social ties helps. Build a study group, solve exercises together with others. If you spend time with them, it will be much harder to get distracted by other things. This doesn't solve everything, so *also* try out good ideas from other people.
4) I would very much recommend spending a year abroad. It's not necessarily the fastest way to start earning money, but most people who did this wouldn't consider it as wasted time. (Except for the last year. So sad for exchange students to have online teaching and distance rules. :-( ) Germany is a nice choice, it has good universities, and (almost) no tuition fees. I think there are many good choices, and few ways to screw this decision up too badly.
Programming jobs are great, but a CS degree is neither necessary nor sufficient to get one. It's one of the fields where the signaling power of a degree is the smallest. Software is eating the world so any job turns into a programming job sooner or later.
Security is a desperate waste of time: companies and countries that don't have security will be replaced by companies and countries that do, but currently the decision-makers in most companies can't tell security from a hole in the ground, so they buy security products based on the opinions of shill operations like the Gartner Group. This works about as well as getting a company to profitability by buying new accounting software chosen by people who don't know anything about accounting.
Don't worry about the girlfriend thing. Being an adult, in both the literal sense and the metaphorical sense, is a much bigger factor here than choice of career, unless your alternatives to CS include becoming a famous actor, playing professional sports, or some other avenue to being a celebrity. Traveling helps too. You will be amazed how much more attractive you are to women at 25 than you are at 18.
3) is almost certainly the most critical thing to get right. No matter what you end up doing, it'll matter hugely. My advice is to listen to the huberman lab podcast, particularly the episodes on sleep, dopamine and addiction. They've helped me get a handle on very similar issues after nearly 3 decades of struggling with them.
I was also pulled in a million different directions as an eighteen-year-old and ultimately decided on CS. There are going to be a lot of people in these comments who say that CS is good, so I'll provide a counterpoint and say that I wish I hadn't done it.
The big question you'll have to answer fairly quickly is "Do I enjoy commercial software programming?"
None of your classes are going to give you the answer to this; the only way to really tell is to get an internship that lets you simulate the experience of being a software engineer (i.e. spending about fifty hours a week doing some combination of writing code and sitting in meetings discussing code).
I hated it, and I realized that I hated it during my first internship, but unfortunately my first internship wasn't until the summer between junior and senior year, so I was kind of stuck.
I thought, "Oh well, no big deal; I can just freelance when I need the money and travel the world, la dee da."
It's much harder than it sounds. Skillset gets stale very quickly in software, so it's hard to jump out and jump back in. Also, you're SO MUCH MORE VALUABLE to a company if they "own your time," so to speak, i.e. if they can call on you whenever they want without paying you extra money, i.e. if you're a salaried employee. If you are a competent programmer with a degree from a name-brand school, you'll probably have people lining up to hire you as a salaried employee but still face difficulty finding decent-paying freelance work.
I also though, "Oh well, with my new CS skillset I can just found a startup and be my own boss, la dee da!"
And yeah, I really did try it! And then the startup didn't work out (happens to about 75% of them), and I was broke and I needed a job, so I ended up back at an office desk for a few unhappy years before I finally got out for good.
What I'm saying is this: If "sitting in an office and writing code with a bunch of people doing the same thing" sounds very bad to you, I would seriously question CS as a career choice, because you run the risk of just kinda washing out like I did. If the collaborative coding project paradigm actually sounds okay to you, then it could be a great fit. There were a ton of people at my company who liked (or at least didn't mind) that paradigm, and for them it was great: They got paid well and were reasonably happy. I just hated every minute of it and had to spend about five years doing it before I finally realized I absolutely could not anymore.
So just...make sure being a software engineer is actually something you'd be okay with, lol.
Other Things You Ought To Know About Computer Science:
1. Unless you are a person who is super mathematically and logically inclined, you're going to have to work a lot harder in college than you would while majoring in, say, English. I had some great experiences in college, of course, but could most commonly be found in the basement of the CS building running on very little sleep and on hour 12 of trying to get some assignment to compile. This was not the experience for my friends who were humanities or social sciences majors.
2. CS might fuck up your GPA, which will make it difficult to get into a good grad school. You'd think they'd care and take into account that you had a fairly difficult major, but they don't, really; it's kind of a pure numbers game (at least for the professional schools).
3. You mention never getting a girlfriend. I mean...CS isn't going to single-handedly stop you from getting a girlfriend, but it does mean that there will naturally be fewer girls floating around in your social circles. It's not a totally dumb thing to consider, although if you're reasonable attractive/social it's not going to matter very much (although I have heard horror stories about the dating scene from my male friends who live in Silicon Valley, lol)
In retrospect, since I ended up going back to grad school anyway, I would probably have just majored in something I enjoyed (something that involved a lot of reading instead of a lot of coding), gotten a high GPA, spent a couple years figuring out exactly what I wanted to do, and then done grad school (it's easy enough to change your trajectory with a masters or postbacc program).
Your description of yourself does not exactly scream "happy as a CS major/software engineer," so I would tread carefully.
The scarcity/glut issue of programmers is this: there are a large number of people with some form of programming credential (like a CS degree) who can't program themselves out of a paper bag. The interview process may be stupid, but so are a huge number of applicants.
A great way to get a leg up on your job application is to do some software development work in college and post it on github or whatever. Find a tool you need, work on Open Source stuff, whatever. Show that you can produce something that works.
Happy and sad fact on corona: by the end of this year, 12.5 billion vaccination doses for corona will have been produced. Monthly production is 1.5 billion.
Given that there are 5.8 billion people of age 15+ on earth, that's easily enough to give every adult two shots, with almost a billion doses left for children, booster shots, etc. (Currently about 250 million booster shots have been given world-wide, so there is still margin.) By end of March, we could give three shots to all adults on earth.
It would even be enough if every single adult would take the shots. We know that some people decline the offer to get vaccinated.
Meanwhile, Africa has given out only 18 doses of vaccine per 100 people. This is not a malfunction of vaccine production, it's a malfunction of distribution.
I am moderately optimistic that this will change soonish, and that the excess doses are finally reaching Africa. I wish I would be sure about it.
Source: https://www.ifpma.org/resource-centre/as-covid-19-vaccine-output-estimated-to-reach-over-12-billion-by-year-end-and-24-billion-by-mid-2022-innovative-vaccine-manufacturers-renew-commitment-to-support-g20-efforts-to-address-remaining-barr/
Update: the largest vaccine manufacturer, the Indian Serum Institute, has announced (or threatened) to cut vaccine production (AstraZeneca) by half because they don't find buyers. Apparently they are sitting on a stockpile of half a billion doses that nobody wants.
https://www.bbc.com/news/world-asia-india-59574878
Yesterday I learned at Noah's substack that vaccine manufacturers had stopped maxxing out their factories a while ago. :/
I am always in favor of more support to poor countries, and outraged by the lack of funding for things like malaria and general economic support. This holds for the vaccine issue, but it's worth adding that sub-Saharan Africa remains among the least-hit regions in the world by Covid, probably because of low urbanization rates (https://ourworldindata.org/explorers/coronavirus-data-explorer?tab=map&facet=none&Metric=Confirmed+deaths&Interval=Cumulative&Relative+to+Population=true&Align+outbreaks=false&country=USA~AUS~ITA~CAN~DEU~GBR~FRA). I believe this remains true when accounting for under-reporting, but don't have a source on hand for this. The fact that the same pattern is seen not only in countries with questionable statistics and health systems (say, Tanzania and Zimbabwe) but also those with robust systems and some levels of transparency (Rwanda, Ghana, Botswana) suggests to me that this isn't just an artifact of misreporting. And this fits a more general worldwide pattern of urbanization predicting infection rates (with the striking exception of China and East Asia more generally).
I think even with this considered, Africa is being under-vaccinated, but the problem is not quite as stark as the raw numbers might suggest.
> a more general worldwide pattern of urbanization predicting infection rates
Do you have a source on that? When I look at US states, the top ten states for total infection rates are North Dakota, Alaska, Wyoming, South Dakota, Tennessee, Utah, Rhode Island, Montana, Kentucky, South Carolina.
Rhode Island and Utah are in the top ten most urban states, but of the other eight states on that list, four are in the ten *least* urban states, and the other four are also in the bottom half of urbanization.
(All time infection rate from here: https://www.nytimes.com/interactive/2021/us/covid-cases.html
Urbanization from here: https://en.wikipedia.org/wiki/Urbanization_in_the_United_States )
I am also a bit doubtful about the urbanization hypothesis. Looking at the list of excess mortalities here, https://www.economist.com/graphic-detail/coronavirus-excess-deaths-tracker, there are a lot of countries in the top part which sound rather rural, e.g. Bulgaria, Peru, Albania, Kazakhstan, Romania.
South Africa is also in the top part. There are very few African countries in the list, though. But I think Africa was hit really hard by the last wave in July/August, much harder than other regions were hit by third/fourth waves.
Does someone know about India? It might be a good test. was hit hard in general, but were rural Indian states hit less hard than rural ones?
There's definitely been a lot of discussion of how boosters for us are getting in the way of first doses for others. But it has sounded to me for a while like most countries are not dose constrained, but are instead either logistics constrained or demand constrained. I'm still hesitant to state that this is a fact, because I just haven't seen any systematic discussion of it, but if anyone has such information that would be helpful.
The link you post suggests that in a few months, the world will have enough doses to get everyone two or three doses, which supports what I'm feeling right now. But I'm a bit skeptical, because this comes in the form of a press release from the International Federation of Pharmaceutical Manufacturers, telling everyone "look, there's no need to weaken our intellectual property protections". It seems to me that even if there are lots of countries that are dose-constrained, they have an incentive to say they've manufactured enough for everyone, and keep selling extra doses to rich countries that are letting them expire while locals refuse their doses, rather than doing the difficult thing of selling to poor countries.
That's a good catch, thanks for pointing this out! I still find the numbers plausible, it's pretty consistent with the prognoses from spring and from summer (which came from an independent analyst company, as far as I could tell). But yes, the source is not neutral.
Also, the shortage is very specific African. There doesn't seem to be a super-severe shortage in India, or Latin America, or Indonesia, or Mongolia, going by vaccination rates.
Except that I just noticed Haiti, which has ~2 doses per 100 people, compared to 125 doses in neighbouring Dominican Republic. Sad.
I would want to dig a bit further into data on some poorer countries in other regions, like Paraguay and Laos. But I am convinced by the broader point - it's mainly vaccination capacity that is missing, not doses.