Comment by solinent | Hacker News Reader

Comment by solinent | Hacker News Reader

Parent Full thread

> So you're taking a high (100s-1000s) dimensional problem and projecting it onto a 2-10 dimensional space where your conclusions are based on a form of clustering which itself is purely dependent upon your (probably ideologically driven) assumptions. In such drastic oversimplifications it is extremely east to paint whatever picture you want, even unintentionally.

It's possible there are only a few main dimensions which matter for showing the majority of the benefit, with the error being negligible. For example, in computer graphics, to solve the rendering equation you're sampling an infinite dimensional space (all paths a photon might take), but you can do it with one dimension (direct bounces from light->surface->camera + deterministic shadow) and the error is almost negligible depending on the application, because most photons get absorbed after the first bounce.

> Case in point. Those evil fat cats simply hoard stolen money and don't interact with greater society in any other way, right? Nevermind driving social policy, producing jobs, and building indirect relationships where large and small actors do benefit from cooperation. Those evil tech CEOs wouldn't have any cash if there wasn't incentive for the poor exploited proletariat to spend their money on goods and services, and offer their time to provide the same.

Perhaps these effects are negligible? If you consider the worst case with one person hoarding all the money, it certainly would be unless he distributed it equally through the means you mention.

I'm not disagreeing with you however, you could be right. It's hard to tell without a large number of empirical studies to be done. The gain from such a change could be worth it, or the effects might be negligible.

It's possible there are only a few main dimensions, but the article doesn't describe anything in particular Hilbe did to evaluate that possibility. We can't accept any model we're presented with just because it's possible that it's correct.

In fact, I'd argue there are obvious reasons to be skeptical of this particular reduction. In the game, fantastically rich people never contributed to the public pot; in real life, most fantastically rich people toss piles of money at public goods like hospitals, schools, and charities.

> In the game, fantastically rich people never contributed to the public pot; in real life, most fantastically rich people toss piles of money at public goods like hospitals, schools, and charities.

It's easy to make an argument that this is negligible, and far more experienced people can make this argument quite well. The money should trickle down, but I think it might help if we had more wealth to trickle down in the first place.

Many contribute to charities do this to avoid taxes, I've had first-hand experience with this type of "charity". The people who actually want to help tend to want to pay taxes, in my experience.

So who is really donating to the charity ultimately?

I'm not saying these benevolent people don't exist, it's just that if you distributed all of Bezos's wealth to each person individually it would only be ~1000 per person once or less, a meager stimulus package. How is that wealth going to trickle down fast enough to have any meaningful effect? Money tends to accumulate towards those who value it the most, but perhaps more wealth could be accumulated if more of it could flow through the system without being trapped only in the wealthy hands.

It could be a win-win, or a win-lose, but it's not a lose for the rich at all if they can now select from a much larger workforce--charity money rarely reaches those that need it most. Just like how many homeless on the street aren't in a desperate situation, and those that are probably can't monopolize those good street corners.

Ultimately there's a feedback loop between supply and demand--any incentive to increase either helps the majority.

I don't mean to be flippant, but I agree easy to make broad arguments about how society in general ought to work. Those arguments just don't have much to do with, and aren't particularly supported by, this game theorist's analysis of how inequality affected the game he experimented with.

I honestly may have been unclear, so I apologize. It's directly related--these are the extra dimensions not accounted for in the criticism of this game theorist's analysis, which supposedly affect the outcome.

So if you can refute my argument, you can easily say that this criticism of this game theorist's analysis is valid, but as I said, my experience does not lie in this field at all, so you probably won't find any difficulty in doing so. It's just pretty obvious to me given my mathematical background that this criticism is meaningless without an understanding of the reasoning behind how the dimensionality was reduced. My point is just that such arguments are easy to make and probably easy to prove as well.

In any case, it leaves room for increasingly complex models to disagree with the results of this one.

> We can't accept any model we're presented with just because it's possible that it's correct.

I agree with this completely, but it doesn't refute this particular reduction of complexity either, which is the point of my post. I may have been too verbose.

Modelling is not just a matter of removing irrelevant detail. It is also about identifying and isolating the different dynamical systems that are at play so that we can build an intuition for how a complex system behaves.

Even if the factors that you are describing are important to the overall outcome, it may still be useful to explore a model with those factors removed.

They (on aggregate) freely give away much less than they would've been forced to under higher taxation, however. Which is an extremely significant difference for the economy as a whole.

And then there's the question of what they give to. Hospitals, schools and charities are great, but what about less glamorous targets like public works maintenance, which often is the first thing to be cut when government coffers run dry?

You are entirely correct in that the crux of the issue is determining which of the initial dimensions are most responsible for the effect you are trying to explain.

The trouble is that the the very nature of the subjects of soft sciences is such that it isn't possible (or at least practical) to perform any kind of rigorous principle component or sensitivity analysis. The result is that dimensional reduction intrinsic to the vast majority of studies in economics, psychology, and the like is built purely on institutional bias, subject to fads and dogma. People essentially build careers around how effectively they can peddle their particular flavor of bullshit - this isn't physics or chemistry or engineering where there is a minimally ambiguous answer. We don't have the tooling (or arguably the right institutional culture) to study soft sciences with rigor and that's dangerous because the academic establishment influences social and political policy with comparatively little self awareness of it's vulnerability to bias.

Reply on news.ycombinator.com