Dec 5, 2023·edited Dec 5, 2023

Re Argentinian economy, there is a difference between Dollarized and Fully Dollarize where they try to Dollarize but it doesn't quite take and they end up with a mix of dollars and the old currency (pesos)

Manifold wants the dollar de jure in place, Metaculus wants the peso de facto gone

Expand full comment

Taylor Swift has been a big pop star for a while, but now she's a billionaire pop star.


Also, smacking Hollywood by bypassing them to put her concert film in theaters and grabbing some of the NFL's spotlight makes it a better year for her.

Expand full comment

How is a commitment to merge with other labs close to superintelligence even plausibly a good thing to commit to if you are concerned about alignment? What if those other labs have different ideas about alignment? What if one of them actually wants an unaligned AI?

I mean either it's essentially meaningless or it's dangerous. If it only requires you merge if the other labs acceed to putting you in charge or at least adopting the approach you prefer then it's just kinda silly. I think pretty much any CEO and board of a non-profit will always accept some other organization with a bunch of assets showing up and saying: you can direct all our resources too.

OTOH if it means serious negotions to work out a power sharing agreement that induces other institutions to accept then it obligates you to share power with groups whose goals (or at least methods) are directly opposed to your own. And if the then board gets to decide if any other institution is too incompatible it's really saying: merge if it seems wise to do so which is the same as not saying anything.


My concern isn't the legal enforceability. Often it's good to put in commitments to a charter that lack any realistic enforcement mechanism. Rather, it's that either good faith board members are either agreeing to a rule they should know would be quite harmful if they felt obliged by it or it's misleading to phrase it as a commitment and should really be more like: when you get close stop and do whatever helps.

Expand full comment

> a graph of different people’s probability estimates of AI causing human extinction

Are these probabilities of extinction conditional on AGI or extinction in general? There's plenty of future outcomes where AGI is never built for one reason or another.

Expand full comment

Part of the reason for Taylor Smith being so high now is that she's on Time's released-Monday-morning shortlist ( https://time.com/6341947/person-of-the-year-2023-shortlist/ ).

Expand full comment

Hey, MY hair is that stylish and I’ve been using Manifold for over a year! Deleted my manifold.love profile because of privacy concerns, but if you want IRL proof re: my hair, text your shidduch resume to 234-5[gematria of my first name][taryag].

Expand full comment

> I’m most unsure about whether Adam D’Angelo is a committed safetyist. He hasn’t said so explicitly and isn’t openly linked to the safetyist movement.

He's not particularly concerned about existential risk, especially in the near future.

Expand full comment

Great post! Fixed the Ilya superalignment market for you. You know there's no Prime Directive here, you could just buy it up yourself!

Any chance we can get your prediction for POTY?

Expand full comment

Thanks Scott! Manifold is now 2 years old, as of December 1st.

We've built something cool, that a lot of people love. And yet I feel there's still so much work left to do.

Expand full comment

> Sam tried to out a board member.

Is this valid usage for other people? To me what it means is that he publicly identified the board member as gay.

Expand full comment

On average American P(doom): https://forum.effectivealtruism.org/posts/Rg7h7G3KTvaYEtL55/us-public-perception-of-cais-statement-and-the-risk-of#Perceived_likelihood_of_human_extinction_from_AI says 26%, but this clearly doesn't reflect their "true beliefs" (insofar as those exist) because if you ask Americans to rank x-risks they put AI pretty low.

Expand full comment

It's weird that the only time Lawrence Summers had anything to do with tech companies is having one meeting with the Winklevoss twins vs FB during his Harvard years. Everything else is politics, World Bank, Harvard, hedge funds and the Treasury/National Economic Council stuff.

He has, however, mentioned AI before:

> In 2018, Summers disputed the claims from then Treasury Secretary Steve Mnuchin that AI would not replace American jobs for 50 to 100 years.

> “The robots are coming," Summers wrote in the Washington Post. That year, he also warned of economic catastrophe if the US “loses its lead" in biotech and AI to China.


Expand full comment

> This has got to be a stunt by Aella, but that means it is very likely to be fun.


Expand full comment
Dec 5, 2023·edited Dec 5, 2023

I wonder if you could make it easier for prediction markets to handle "why" questions by mixing them with Pol.is-style crowd analysis (incetivize users to submit and vote on binary statements regarding the market->Use a dimensionality reduction algorithm to find the primary dimensions of disagreement->Let the options of the market be defined by these dimensions).

Expand full comment

> Are some of these people normies? Seems surprising, but I can’t rule it out.

I guess it also depends on where you draw the line? For an extreme example, some maths PhDs have many normie leanings and count as normies in some contexts, but are definitely not close to modal normies when the question is about enjoying the precise-sciences in-jokes…

Expand full comment

To answer Scott’s Taylor Swift question: https://www.thefp.com/p/taylor-swift-unites-america.

Expand full comment

> This will be a total unfair stereotype, and you should feel free to yell at me for it, I just think usually prediction market junkies don’t have hair as stylish as the woman in Picture #3

Ok, I’ll do it- I think this comment is unnecessary and pretty alienating. I understand that it’s lighthearted, but people feel hesitant enough already about not “fitting in” to the rationalist community that and jokes like these reinforce that. Also, putting myself in that person’s shoes- I wouldn’t like to submit a photo to a dating site and have it remarked on to a wide audience in any way, positive or negative.

Expand full comment

I'm a huge Swiftie, so I don't claim to be unbiased. But the Swift mania is off the charts more than ever this year, and with good reason. Here's the best case I can make for her as Person of the Year:

1. Eras Tour

Everything about the Eras Tour is absurd. Who performs for 3.5 hours straight with 10 different costume changes? And then does that for several nights in a row? And then repeats the spectacle for 146 nights over nearly two years? (The tour started in March 2023, and the international leg continues until December 2024!)

Her opening night brought in more business for the host city than the Super Bowl: https://time.com/6307420/taylor-swift-eras-tour-money-economy/. And that was just one of 52 shows in the US alone. Overall, her tour brought in nearly $1 billion in revenue this year, making Taylor a billionaire. And it produced around $5 billion in revenue for the broader tourism industry.

In addition, the millions of people who attempted to buy Eras Tour tickets crashed the Ticketmaster site so badly that it led to a congressional investigation. This included multiple senators dropping references to Taylor's lyrics in an attempt to seem cool and hip: https://www.cbsnews.com/news/taylor-swift-ticketmaster-senate-judiciary-committee-eras-tour/. (IMO, all of them failed as miserably as one would expect, but the obvious attempts at pandering show how influential Taylor is.)

On the side, Taylor also released an Eras Tour movie, which was one of the top-earning movies of the year. Its box office total has now passed $250 million: https://www.msn.com/en-us/music/news/box-office-milestone-taylor-swifts-eras-tour-crosses-250m-globally/ar-AA1kDfb0.

2. Disrupting the music industry

Taylor's success is due to her business savvy just as much as her songwriting ability. She's constructed a brand around herself in ways that few other artists can touch. And the cornerstone of that brand has been her brilliant moves over the past few years to gain control of her own work.

To recap, in 2019, the original masters of Taylor's first six albums were sold to a private equity firm over her strenuous objections. But Taylor found a loophole - the copyright only applied to those specific recordings. As was standard for the industry, Taylor's contract allowed her to create new recordings of her entire catalog five years after the original albums were released.

Before Taylor, no one had attempted to exploit that loophole at scale. Perhaps that's because no one had achieved much success with re-releasing songs that everyone already knew. But through a combination of adding new songs ("From the Vault") and appealing to her loyal fanbase, Taylor made it work better than most people imagined possible.

She has since re-released four of her first six albums as Taylor's Version (two of them coming in 2023), with two more likely arriving soon. All four have debuted at #1. The most recent and most successful was 1989 (Taylor's Version), which sold 1.653 million units in its first week. That's larger than any other album since 2015, and significantly larger than the original release, which opened at 1.287 million units.

In response, the major music labels have all rewritten their standard contracts to prevent future artists from replicating her success: https://www.billboard.com/pro/taylor-swift-re-recordings-labels-change-contracts/.

So it remains to be seen whether Taylor has created new opportunities for artists to take more ownership of their work, or whether she's only exploited a temporary loophole for her own advantage. But either way, she's definitely disrupted the industry.

3. Warm fuzzies

This is admittedly subjective, but the piece that Petey shared points in the same direction: https://www.thefp.com/p/taylor-swift-unites-america.

Of all the options for Person of the Year, Taylor Swift is by far the most likable. That's why she's the overwhelming favorite in the reader's choice market: https://manifold.markets/Joshua/will-taylor-swift-be-the-time-magaz.

At a time when the news is full of doom and gloom, and the country is more polarized than ever, Taylor is the one major person who is nearly impossible to hate. And that makes her the Person of the Year that we need right now.

Expand full comment

> Why is Taylor Swift so high? I understand she’s a very famous pop star, but hasn’t she been an equally famous pop star every one of the past ten years?

I don't know why, but anecdotally, I've been hearing way more about Taylor Swift in the past 6 months than I did in several years before that. Seems she had some sort of resurgence in popularity.

Expand full comment

Regarding long-term markets, I had a thought on early resolutions. Specifically, I created a market that asks if a large construction project will finish in time. It will resolve as NO as soon as a delay is announced. This is good because it reduces the expected period of capital lock-up. This is also bad because it creates different incentives for different participants: If you expect major delays, you also expect a shorter lock-up. If you expect a punctual finish, you expect a longer lock-up.

I'm basically resolving the market early based on a (somewhat authoritative) third-party estimate. This is problematic in itself, but seems fine to me because the project management organisation is strongly incentivised to not falsely announce delays.

I was wondering if anyone had thoughts on this approach.

I will also shamelessly self-promote my Manifold market: https://manifold.markets/ThomasTwenhoven/will-the-fehmarn-belt-tunnel-open-i

Expand full comment

"The best realistic medium-term outcome I can imagine for the people of Gaza is as something like a West Bank without settlements and roadblocks"

I don't think this is even remotely realistic. Israel has worked_ hard_ to cleanse the good areas for colonization and settlements - there is absolutely no way they would ever return all of that, not even if a credible peace was offered (and no-one can offer that). That ship sailed long ago - it could _possibly_ have happened in the early 2000s, but not now.

Expand full comment

Two things:

(1) I am totally uninterested in talking with, using, or playing with ChatGPT or any of the competing chatbots. "But you can do art and write essays and this, that and the other!"

Yeah. I find I don't care. I tried one model that allows you to just start online, and I found myself sitting there with a blank input prompt box, because there was nothing I wanted to ask it, no conversation I wanted to begin, not even any "can I make it do no-no things?" because I don't care what a machine spits out as 'Hi I'm your real liddle fwend who wanna talk wid u about stuff just like human person!'

So I'm approaching this whole giant blob from the point of view of "uninterested idiot"

(2) Given all the chit-chat about OpenAI, here's what Microsoft are doing in the field. This isn't theory, this is "selling a product to businesses". I think it might be interesting to get the view from the outside, as to what the actual implementation of AGI is visualised as (not the theoretical paperclipper/fairy godmother stuff):


That mess above brings you to the PowerPoint Presentation from the November "Microsoft Discovery Day: Building AI-Powered Apps". It amuses me, a bear of very little brain, that nestled in among all their salesmanship about how ground-breaking the AI is, there is still an AI-generated piece of art which can't understand how to spell correctly "integration" or "recognition" in "AI integration" and "Voice recognition" text on image.

"Microsoft and OpenAI partnership

OpenAI: Ensure that artificial general intelligence (AGI) benefits humanity

Microsoft: Empower every person and organization on the planet to achieve more

Azure OpenAI Service :

GPT-4 & GPT-3.5-Turbo Text, Chat & Code

DALL·E 2 Images

Whisper Transcription & Translation"

Very nice, how do they see this translating into practice?

"Introduction to Top Use Cases

(1) Business Problem

Productivity is lagging

Business Needs

Increase Productivity


• Conversational Search/Knowledge Insights

• Code Generation and Documentation

• Trend Forecasting

• Report Summarization & Generation

(2) Business Problem

Need for process Automation

Business Needs

Automate Processes


• Document Processing

• Workflow Management

• Fraud Detection

• Supply Chain Optimization

(3) Business Problem

Degraded Customer Experience

Business Needs

Improve Customer Experience


• Intelligent Contact Center

• Agent/Employee Assistance

• Virtual Assistance

• Call Analytics

• Call Summarization

(4) Business Problem

Creating Content is Time Consuming

Business Needs

Build Creative Content


• Marketing/Sales Content Generation

• Personalized Content Generation

• Product Design & Development

• Digital Art

What can Generative AI Do?

Generate New Revenue Streams

Deliver Differentiated Customer Experiences

Modernize Internal Processes"

In future, Scott can be freed from the time-consuming burden of "build creative content" by outsourcing it all to AI. How nice to look forward to this!

So the AI future in the workplace is not going to be increased leisure and money, it will be "you WILL be more productive! you WILL increase velocity! if you still have a job, that is, and have not been replaced by the machine altogether". I honestly think *that*, rather than "unaligned AI will find a way to 3-D print neurotoxins to be delivered globally by nanobots to wipe out humanity" is the real problem we will be facing and should be discussing.

Expand full comment

I'll reiterate a "prediction" I've been offering recently: We'll arrive at a substantial understanding of the internal structure and operations of LLMs BEFORE (the mythical) AGI is attained. Moreover, once that happens, the need for belief in AGI will weaken and the idea will disappear, as did the idea of phlogiston once upon a time. What will replace AGI? I surely don't know, specifically. But in general, a richer and most robust understand of the mechanisms of both artificial and natural intelligence.

On natural intelligence: Principles and Development of Natural Intelligence (5 of them): https://www.academia.edu/235116/Principles_and_Development_of_Natural_Intelligence

Expand full comment

To set the record straight: I once gave a ~2% probability for the classic AGI-doom paperclip-maximizer-like scenario. I have a much higher probability for an existential catastrophe in which AI is causally involved in one way or another — there are many possible existential catastrophes (nuclear war, pandemics, runaway climate change…), and many bad people who would cause or fail to prevent them, and I expect AI will soon be involved in just about everything people do! But making a firm prediction would require hashing out what it means for AI to play a “critical causal role” in the catastrophe — for example, did Facebook play a “critical causal role” in Trump’s victory in 2016? I’d say it’s still not obvious, but in any case, Facebook was far from the only factor.

Expand full comment

> I thought Kalshi was trying to cultivate good will with the CFTC, and this seems pretty adversarial.

A couple of perspectives. First, cultivating goodwill can be (is?) adversarial. If nothing else, you're competing (sometimes with the CFTC) for the CFTC's attention.

Second, even if the parties are getting along, an adversarial process might be inevitable to break a logjam. When an agency reaches the limit of its statutory authority, its hands might be tied (especially if there's political or media scrutiny). In that case, you go to Congress for a new statute or you go to the courts for a new understanding of the statute. Either way, adversarial.

To be clear: I'm not an expert in CFTC, prediction markets, or regulatory law in general. But I am a lawyer at a state-gov agency. We work very closely with regulated entities and have a strong customer-service culture, but it's often clear that an entity has to turn to politicians or judges to get around limits on agency power.

Expand full comment

Taylor Swift, yet again beating out Beyoncé despite Beyoncé also going on the biggest tour of her career.

Taylor Swift had the most lucrative tour ever, the most lucrative music film premiere, and she has a football player boyfriend.

Expand full comment

>> To my surprise, this not only hasn’t collapsed, but has attracted people outside the usual prediction market community:

These are fake profiles. Some with locations like "California City" and "Arizona City".


Expand full comment
Dec 5, 2023·edited Dec 5, 2023

I'm entertaining the idea of trying manifold.love, but I have a physical preference I perceive is the opposite of what people would expect. It seems like that would be important information for anyone to make better bets, but I normally avoid putting that publicly on my profile in dating apps because because I'm *open* to women who don't match that preference and I don't want to burn that bridge with them (for them to know I have a preference they don't meet). How would it make sense to handle that on manifold.love? Maybe accept that this is the one dating app where I should risk ruling out all women in one category for the sake of better matches?

Expand full comment

re: Kalshi

I think the details of Kalshi's proposal for election markets are very important. iirc, the original proposal was markets on the outcome of Congressional control with position limits of $25,000 -- seemed fair enough. But their most recent proposal allows ECPs (eligible contract participants) to invest up to $100 million, and I think this is just way too high a limit to start playing with, and I worry concerns about manipulation actually become quite plausible. I get why Kalshi did this, because the CFTC said that they didn't believe the contracts had significant economic value for hedging with a 25K limit (which oil company is going to stake 25K on an election market to hedge against Democratic control), but I think the better route is to start small, and see what problems arise, instead of just opening up election betting on such a massive scale all at once. I reckon it will probably not end up being a good idea to have such high position limits to start, if ever.

If the court rules on the narrow question of whether "election markets" are or are not "gaming", that could still give the CFTC enough leeway to regulate contracts for their manipulability (warning: not at all a legal expert). But if the court just totally guts the CFTC's ability to regulate election markets, I think this will be bad -- I think if you want markets to work well (not just in terms of accuracy but social legitimacy), they need to be regulated (imposing requirements like KYC, market surveillance, etc.).

Expand full comment

I gotta say, the endorsements feature on manifold.love sure seems underused. :P

Expand full comment

Scott: "An Israeli occupation would involve constant bloody resistance; I'm not convinced it would be any better for people on the ground than Hamas (though someone can try to convince me otherwise)"

Well, one possibly convincing argument is that we have a test case for what it means for Israel to occupy a chunk of Palestine with a couple million Palestinians in the 21st century, and it sure seems to be better than Hamas. Israel has occupied the West Bank continually for decades, with soldiers, police, and settlers. There has been resistance, and I suppose pedantically you could call it both continuous and bloody, but I'm guessing you were thinking of something more than that.

And while I suspect that both Israel and the non-Hamas Gazans would prefer that Gaza be occupied by UN peacekeepers or maybe even Egypt, but those aren't likely to happen. But I don't see any fundamental reason why an Israeli occupation of Gaza, particularly in the medium term, should be any more bloody than the West Bank is now.

On the one hand, at least in the short term the Israelis have given the Gazans cause to be really, really upset with them. On the other hand, the Gaza population is presently going through a filter that will select for not being eager to engage in bloody violent resistance to Israel. And there are no settlers in Gaza, nor are there likely to be, which eliminates a huge ongoing aggravation.

Expand full comment

I just fixed the issue with the embeds. There was a difference in the embed graph code that was making it connect the points for each bet with straight lines. Which is wrong! We put every single point* so the chart should be stepped rather than linear interpolated. Usually this wouldn't be a significant visual difference but in this case nobody bet on the market for an entire month before it dropped.

*[skippable nerdy aside]: We used to literally put every point on the chart but since lk-99 this would be really slow on the most popular markets. And on a chart that is at most 700px wide people can't perceive the level of detail from tens of thousands of bets anyways. At the same time I HATE those smooth curvy trailing average charts that ui designers keep using because they don't understand that data should not be beautiful; data is bumpy, sharp, and real. It took some tinkering but the solution I settled on for big markets is to slice the x-axis into 500 evenly-sized buckets; take the min, max, and median in each bucket (or the last value in the previous bucket if this one is empty); and draw the chart with those points. I am pretty proud of the end result, I think it looks very close to what it would look like with full data. As long as you don't zoom in :p

Oh, also we start the calculation with no more than 50,000 bets on a market, semi-randomly sampled, because I was too lazy to paginate through all the bets and eh it works

Expand full comment

Manifold.love is a neat idea. I'm all in favor of anything that comes close to bringing back 2005-2010-era OKCupid, and the matching questions are a big part of that.

That said, I'm dubious of the "make a Yes bet to send an intro" mechanism. It suggests some specific target user demographics: people who have more than a 50% chance of a first message turning into a first date. As far as I know, that niche contains most women and very few men.

Are there guys on the platform making confident Yes bets of any substantial size, or is everyone just putting in Ṁ10, or are people spending real money to buy Ṁ to make bigger mostly-losing bets on intros, or... something else?

Expand full comment

My prediction for Gaza is that no government will reign there. Israel will conduct frequent raids in an on-going basis but based from outside the Strip. Hamas will more or less retain military control of Gaza but by having to operate from underground will not be able to effectively govern the territory. No other force will be willing to operate a government in Gaza. — 75% chance.

The UN and/or Red Crescent will become the de facto government of Gaza. — 20% chance.

Basically there will be no police type activity in Gaza from now on. Just military and basic humanitarian activity.

Expand full comment

I'm the #6 most profit on the POTY market, so I think I'm qualified to speak on how I got it right.

The Eras Tour was the highest grossing tour in U.S. history. It generated small booms in excess consumer spending in every city where she performed; According to current estimates, about $4.6 billion in excess consumer spending. It's like Swift ran her own private stimulus program for post-COVID downtowns. If you worked in or near a city she performed in it was all people were talking about for weeks. This is pretty much the largest cultural phenomenon in the US in a very long time.

Meanwhile, most normal people do not know who Sam Altman is, and a minority of them have even heard of ChatGPT.

Expand full comment

One issue manifold has always had is with conspiracy theories - people resolving their own markets gets you a bunch of markets like "did the CIA kill JFK" where you're basically betting on how credible the market creator thinks they are, and there's always been an unfortunate amount of those on the site. But with the Gaza stuff in the news it's exploded, with a bunch of highly-active users pushing every sort of antisemitic conspiracy theory under the sun (stuff like "well obviously AIPAC/the IDF/the CIA planted/made up all the evidence for X so this market should resolve no").

Makes it both hard to trust (you now have to look up even market creators of factual-seeming markets to make sure they're not going to misresolve due to conspiracy theories) and makes it a pretty unpleasant experience to go on (especially if you read the comments, you have to wade through a lot of "the *real* terrorists aren't the rape/torture/murder advocates they're the people fighting them).

Expand full comment

>And Israel's previous commitment not to do settlements in Gaza (if maintained) would make a West-Bank-style Gaza better off than the real West Bank.

Why? Did Germany becoming (largely) Judenrein make it better off?

Expand full comment
Dec 9, 2023·edited Dec 9, 2023

The graph says that Holden Karnofsky is at 50% extinction risk, which I think is based on an incorrect interpretation of the source. The linked source is this post https://www.cold-takes.com/some-additional-detail-on-what-i-mean-by-most-important-century/ which gives 50/50 odds for "civilization could either end entirely, or change so dramatically that 'humans as we know them today' would either not exist anymore, or would at least be a very small part of the population [and] This century is our chance to shape just how this happens".

That's entirely consistent with presently alive biological humans surviving. Biological humans could become a small part of the total population just via the creation of tons of digital minds. (Which humanity may freely choose to do.) So 50% is an upper-bound on extinction risk.

Edit: The 50% is also conditional on PASTA being developed.

Expand full comment

Judging by the comment, her performance is a tribute to her stamina and athleticism.

Expand full comment