I just put $200 towards Emily's "Platform for migrants to start legal and profitable microbusinesses", mentioned in this post. I have some personal experience in this area and I think the approach sounds excellent and well structured; more importantly, the founder is giving off strong vibes of follow through, ability to execute, and commitment to the effort. I think there's a very high chance of meaningful success.
(p.s. Registering and logging in with the Google account button on iphone was tedious-- there's some kind of problem where logins are not being fully processed and the login state is not remembered when opening new tabs. Donating flow was v smooth.)
Regarding the HPMOR Book distribution, the pitch needs a lot of work, especially for someone who has done this book distribution once already. How many books were distributed before and what data was this person able to collect on the conversion rate from books to rats or AI safety researchers? Where does the 85% number come from? They mention doing it "effectively" and "impactfully" again and again but has given no details about how that will happen. What steps are they taking beyond just distributing the books to random strangers on the street or on campus? For now I remain skeptical of this strategy.
Not that I doubt your integrity, but because charities get to put their own value on ex ante actions, I suspect this can be used to do non-charitable activities, such as money-laundering. To use your toy example, I have $75,000 in illicit drug sales I need to launder. I pay some small investors $25,000 total to invest a total of $50,000 in impact certificates for a designated, but shady, charity, which ends up deciding to pay out $75,000. Where did they get the $75,000? From anonymous donations which turn out to be from the same source.
I'm sure the ACX grants have none of this, but I'm looking at a wider adoption. I think those that value the impact certificates for more than the purchase value would need great scrutiny, including audits by some independent authority.
> when you sell your investment, you won’t get a payout in real money. You’ll get it in special Manifund dollars that you can donate to other charities, but can’t spend on non-charitable purposes. I realize this is less fun than real dollars, sorry.
In case anyone else is thinking "wait I've totally heard that name before", the leader of the first project is *not* the same person as the lefty Dutch historian who wrote Humankind and Utopia for Realists, Rutger Bregman.
*Who are you?*... I'm a third-year postdoc in the Department of Neurobiology at Harvard Medical School. I study a variety of things, including Bayesian decision-making, memory, and reinforcement learning. I do theory, but work closely with experimentalists.
*What are you proposing?*... In short, I'm hoping to create a living guide to neuroscience analogous to Lilian Weng's living guide to AI (https://lilianweng.github.io/). This would start with a fixed number of posts on foundational topics (that I also know something about). Posts would appear on LessWrong and my personal website.
*Why should I care?*... Given the interest of Scott / LessWrongers / other bloggers (like Erik Hoel) in neuroscience, it's easy to think it's already well-covered ground. But I'm not so sure about this. I read a lot of popular science and rationalist-y neuro content for years before I found myself embedded in a (mostly experimental) neurobiology department, and I've come to find that neuroscientists think about nervous systems somewhat differently than those other sources. They care a lot about fine biological details, and careful experiments are increasingly showing that our naive ideas about how the brain accomplishes certain things (e.g., maybe the brain uses TD learning to evaluate actions) are either untenable or highly incomplete. If you care about constructing AGI, you might benefit from a good summary of how neuroscientists currently think about these sorts of issues; you might also care if you just think brains are cool and wonder about how they work.
*Why you?*... I've been in neuroscience long enough to 'get it', but not so long that I don't remember what it was like to be on the other side. I also personally know and regularly chat with people doing cool work on the cutting edge. (I'm doing cool work too, but a lot of it isn't out yet. Sorry about that!)
*Why do you need money?*... I don't. But writing a solid guide takes a lot of time and energy, and the monetary incentive helps. Also, money is a hard-to-fake signal that there exists a reasonable contingent of people that will actually care.
Thanks for your consideration! Feel free to ask questions or email me (google my name) for more info.
I am still not sure I understand what is the incentive for the investors to pay anything for the certificates after the goal is achieved, and how the valuation is established. The example says "You sell the impact certificates to an animal rights charity, who decide that if they were trying to get a law like this passed ex ante, they would be willing to spend up to $75,000 on it." But what is the incentive for them to value it as high as $75K? Can't they just say well, nice work in passing this legislation, we probably could have passed it for $20 and a pack of bubble gum, but we just didn't get to it yet, so thank you for spending all that money on it, good job! They don't even need to be fraudulent on that - something that somebody else already did always looks easier than it looked before they've done it.
Of course, if the animal charity does it all the time, nobody would use certificates so if they want people to use certificates, they need to pay up. But the animal charity's goal is caring for animals, not for certificate markets, so the incentive for them not to ruin certificate markets (and save $75k that could be used for other animal care projects) would be low, it's not their primary goal. So I don't understand how this scheme can survive - it looks like the incentives aren't aligned in the right direction.
I just put $200 towards Emily's "Platform for migrants to start legal and profitable microbusinesses", mentioned in this post. I have some personal experience in this area and I think the approach sounds excellent and well structured; more importantly, the founder is giving off strong vibes of follow through, ability to execute, and commitment to the effort. I think there's a very high chance of meaningful success.
(p.s. Registering and logging in with the Google account button on iphone was tedious-- there's some kind of problem where logins are not being fully processed and the login state is not remembered when opening new tabs. Donating flow was v smooth.)
Regarding the HPMOR Book distribution, the pitch needs a lot of work, especially for someone who has done this book distribution once already. How many books were distributed before and what data was this person able to collect on the conversion rate from books to rats or AI safety researchers? Where does the 85% number come from? They mention doing it "effectively" and "impactfully" again and again but has given no details about how that will happen. What steps are they taking beyond just distributing the books to random strangers on the street or on campus? For now I remain skeptical of this strategy.
Not that I doubt your integrity, but because charities get to put their own value on ex ante actions, I suspect this can be used to do non-charitable activities, such as money-laundering. To use your toy example, I have $75,000 in illicit drug sales I need to launder. I pay some small investors $25,000 total to invest a total of $50,000 in impact certificates for a designated, but shady, charity, which ends up deciding to pay out $75,000. Where did they get the $75,000? From anonymous donations which turn out to be from the same source.
I'm sure the ACX grants have none of this, but I'm looking at a wider adoption. I think those that value the impact certificates for more than the purchase value would need great scrutiny, including audits by some independent authority.
> when you sell your investment, you won’t get a payout in real money. You’ll get it in special Manifund dollars that you can donate to other charities, but can’t spend on non-charitable purposes. I realize this is less fun than real dollars, sorry.
Sounds like Robin Hanson's "sacred money".
https://www.overcomingbias.com/p/sacred-moneyhtml
https://www.overcomingbias.com/p/toward-consecrated-capital
If I really like one of these charity ideas, is there any reason why I'd want to donate normally to them instead of buying an impact certificate?
In case anyone else is thinking "wait I've totally heard that name before", the leader of the first project is *not* the same person as the lefty Dutch historian who wrote Humankind and Utopia for Realists, Rutger Bregman.
Workaround for small investors who want to get payout in dollars instead of mana: Buy a ticket to Manifest using mana instead of dollars.
Here's a market about which certs will get prizes bigger than their initial valuation, i.e. which certs will be good investments.
https://manifold.markets/mckiev/which-of-the-following-acx-grants-i#
I'd like to quickly advocate for my proposal, which you can find here: https://manifund.org/projects/create-accessibl
*Who are you?*... I'm a third-year postdoc in the Department of Neurobiology at Harvard Medical School. I study a variety of things, including Bayesian decision-making, memory, and reinforcement learning. I do theory, but work closely with experimentalists.
*What are you proposing?*... In short, I'm hoping to create a living guide to neuroscience analogous to Lilian Weng's living guide to AI (https://lilianweng.github.io/). This would start with a fixed number of posts on foundational topics (that I also know something about). Posts would appear on LessWrong and my personal website.
*Why should I care?*... Given the interest of Scott / LessWrongers / other bloggers (like Erik Hoel) in neuroscience, it's easy to think it's already well-covered ground. But I'm not so sure about this. I read a lot of popular science and rationalist-y neuro content for years before I found myself embedded in a (mostly experimental) neurobiology department, and I've come to find that neuroscientists think about nervous systems somewhat differently than those other sources. They care a lot about fine biological details, and careful experiments are increasingly showing that our naive ideas about how the brain accomplishes certain things (e.g., maybe the brain uses TD learning to evaluate actions) are either untenable or highly incomplete. If you care about constructing AGI, you might benefit from a good summary of how neuroscientists currently think about these sorts of issues; you might also care if you just think brains are cool and wonder about how they work.
*Why you?*... I've been in neuroscience long enough to 'get it', but not so long that I don't remember what it was like to be on the other side. I also personally know and regularly chat with people doing cool work on the cutting edge. (I'm doing cool work too, but a lot of it isn't out yet. Sorry about that!)
*Why do you need money?*... I don't. But writing a solid guide takes a lot of time and energy, and the monetary incentive helps. Also, money is a hard-to-fake signal that there exists a reasonable contingent of people that will actually care.
Thanks for your consideration! Feel free to ask questions or email me (google my name) for more info.
I am still not sure I understand what is the incentive for the investors to pay anything for the certificates after the goal is achieved, and how the valuation is established. The example says "You sell the impact certificates to an animal rights charity, who decide that if they were trying to get a law like this passed ex ante, they would be willing to spend up to $75,000 on it." But what is the incentive for them to value it as high as $75K? Can't they just say well, nice work in passing this legislation, we probably could have passed it for $20 and a pack of bubble gum, but we just didn't get to it yet, so thank you for spending all that money on it, good job! They don't even need to be fraudulent on that - something that somebody else already did always looks easier than it looked before they've done it.
Of course, if the animal charity does it all the time, nobody would use certificates so if they want people to use certificates, they need to pay up. But the animal charity's goal is caring for animals, not for certificate markets, so the incentive for them not to ruin certificate markets (and save $75k that could be used for other animal care projects) would be low, it's not their primary goal. So I don't understand how this scheme can survive - it looks like the incentives aren't aligned in the right direction.
Where can I find an introductory explanation to the Manifund?