>instead we must show how a solution is produced by the interactions of people each of whom possesses only partial knowledge.
(In the same spirit, I'm trying to move beyond dissfests towards more collaborative convos on HN)
@warden I was encouraged to see "perfect knowledge" replaced by "sufficient knowledge" in at least once instance in your post.
We should encourage Glen & Michael to work on seeing if weakening their assumptions as above would produce proofs of more useful natures, as a prelim to solving the knowledge problem :)
(Alas collabs take work, & HN is primarily an entertainment & venting channel, due to its ephemeral design)
In the example in the article, many more people did contribute to the lead pipes (100 people vs. 10 for the art museum). And they still get only a fraction of the funding that the art museum gets.
Agree, coordination is a larger threat to QF. But this issue has been discussed extensively. In this article I wanted to point out all the other assumptions behind QF and what happens when they don't hold.
But they got 99 shares of government contribution for each share they put up for a 100x multiplier while the art patrons got 9 shares of government contribution for a 10x multiplier.
Setting aside the emotional content and looking only at the math, it’s not at all obvious to me that the project with 100 donors was somehow shorted.
Yes good point. We'd have to actually flesh out the assumptions about marginal utility of wealth for low-income vs. the high-income group (as well as assumptions about individual utility functions), to demonstrate that this outcome was not optimal. I didn't do that in this article because it gets too mathy.
However, the optimality of QF does assume wealth equality. When you drop that assumption and assume diminishing marginal utility of wealth, you can show that QF is not optimal.
But I think you are right that the example in this article doesn't necessarily show that clearly. The example leans heavily on intuition (or emotional appeal). I think I will try to improve that section.
Many people would consider fixing pipes a more important project despite the fact that the wealthy contributors could front a lot more cash for their pick.
You're missing the point, which is that the group with a lot more money to start have their voices heard more.
It matters some that their multiplier is different , but in absolute numbers its still more to the program that benefits fewer people. The "utility function" is not accurate because the wealthy's utility starts out with a massive advantage.
So yes, I think it would still be unfair if you switched it given the poor majority genuinely would rather have art than lead free pipes.
The problem is that their voices are counted less due to not starting with money.
But regardless, that would be a silly thing to switch because that's not a situation that ever comes up, while the original framing is a genuine problem in our society right now.
Yes, money has utility. That should be unsurprising to everyone.
This proposal's pairing of hypothetical projects levels the playing field by a factor of 10 versus the starting point. That seems like a pretty good improvement over the purely monetary starting point.
If your objection is that government can't work this way, because some projects need to be done for the benefit of people who literally cannot even contribute so much as a penny, while other projects are optional, then I'll agree with you. It means that this funding mechanism is fundamentally flawed in regards to required projects.
But if you want to augment government spending with private contributions for certain public-private partnership projects, this might be a good way to allocate government matching funds for these optional projects.
You can't treat a lead pipe replacement project as an optional project (the responsible government or utility just has to do it), but if you wanted to trade off funds towards a skate park versus towards an art museum, this process seems better than a straight matching funds percentage process.
Or, if you want to have no partnership projects and use existing government mechanisms exclusively, that also avoids this problem.
The point is that its not socially optimal. The socially optimal solution would optimize global utility, as in it would not be influenced by the starting wealth of each person. If you allow starting wealth to influence things, their needs will be optimized for more at the expense of people who do not.
Yes, that goes against the idea that "money has utility" but the point the article was making was that its not socially optimal anymore not that is regressive compared to whatever other strategy, like straight matching funds. There's no math claim that straight matching funds is optimal either.
I think maybe we're speaking past eachother? Because yea totally I'd rather there be a multiplier based on the # of people than not given either that or a straight match. And your other options sound good too: "always fix non-optional things" and "do things democraticly (so 1 person 1 vote, not 1 dollar)"
But the article is making a very specific point about a claim of QF being mathematically socially optimal that isn't being met.
I think all of these schemes end up failing in environments other than academic papers.
If I was a wealthy person and the calculation was based on share of wealth donated, I’m pretty sure I’d find a way to become a massive employer of temporary labor locally at $100/hr but only for 8 hours, make clear my preference for the art museum or other causes, and that I like to learn about what causes matter to them.
Accordingly, those local workers for whom I like their cause and story of why it matters to them, I’ll match their donations (based on receipts showing their donations) and also setup a series of follow-up paid interviews in the future to learn about their on-going donations to a cause they obviously care so deeply about. They keep about half the money to compensate for their time and to grow their wealth, I get to socialize with some people who support the same causes I do, I no longer have an outsized donation in my name [we wouldn’t want me to have an outsized influence], and together, we can really do great things!
> levels the playing field by a factor of 10 versus the starting point.
Let's say there were only 10 poor people that contributed to the pipes. The total funding would be $10,000 -- a subsidy of $9,000. So 10x multiplier both for the pipes and the art.
Then let's also say that the marginal utility of $100 for a poor person is equivalent to the marginal utility of $1,000,000 for a rich person.
So we have the same number of contributors for each project, but a much higher marginal utility-per-dollar for lead pipes. But the socially optimal funding would be at the point where the marginal utility-per-dollar are equal for both projects (per the Equimarginal Principle).
A core problem of capitalism is that it only solves the problems of people with money... This is more of the same.
Consider that any of those 100 people might have a kid who would be the next Einstein, if only they hadn't been lead-poisoned. But these hundred people also have rent to pay and food to buy, and can only set aside $100 to deal with the lead-pipes problem. The existing distribution of wealth is not a good measure of the importance of the problems that these different individuals are experiencing. And the existing distribution of wealth is thus not a great way to prioritize solving problems for maximum societal benefit.
But do we know the "ideal" funding values for the pipes and museum (and if we can, then why not use that)? It's only really "unfair" if we know that the pipes "deserve" a disproportionately larger multiplier. If the pipes deserve to be funded regardless of contributions (and they probably do), then the issue is using a system that could possibly fail to provide for them in the first place.
It's not a good way to allocate funds, but I don't think it's a slam dunk to say it multiplied a larger group's money more than it did a smaller group's.
Well it's also unfair if we assume large difference in the marginal utility of wealth -- for example to go to extremes, we might assume that a $10 contribution from a low-income individuals represents the same sacrifice as a $1,000,000 contribution from a high-income individual. If that were the case, a $100 contribution from a low-income individual represents 10x the utility of a $1,000,000 contribution from a high-income individual. So in that case the lead would pipes have both more contributors, and higher utility per contributor, than the art. So total utility would be maximized by giving more money to the lead pipes.
Here's a very brief summary of what Quadratic Funding is (which is distinct from Quadratic Voting):
Quadratic Funding is a mechanism where individuals voluntarily contribute funds for some public good (e.g. an open source software project), and then these are matched such that the total funding amount is equal to the square of the sum of the square roots of the individual contributions. Under certain assumptions, this formula results in an optimal outcome, where each individual contributes an amount that maximizes their individual utility (given what others are contributing), and total utility for society is also maximized.
No one has a deep emotional connection with OpenAI that would impede switching.
At best they have a bit of cheap tribalism that might prevent some incurious people who don't care much about using the best tools noticing that they aren't.
The way, I understand it is that with a Poisson process, at every small moment in time there’s a small chance of the event happening. This leads to on average lambda events occurring during every (larger) unit of time.
But this process has no “memory” so no matter how much time has passed since the last event, the number of events expected during the next unit of time is still lambda.
I don’t use probability distributions in everyday life ;)
But it is the right distribution to represent uncertainty about the probability of binary events (eg a website user clicking some button). For example, if I have absolutely no idea the probability then I use the uniform distribution, Beta(1,1), which is the maximum entropy distribution. Then if I observe one user and they happen to click, I have Beta(2,1), and at a glance I known the mean of that (2/3) which is a useful point estimate.
It’s not, really. If you try to read it you will fail miserably with a completely unenjoyable experience. It’s kind of like The Bible or The Odyssey, anyone who recommends it is out of touch.
