Why does this article emphasize Komodo MCTS and completely ignore LeelaZero which has had much more exciting and interesting results as far as 'alternative method' engines go? This is especially strange given that Leela operates similarly to AlphaZero which the article strongly praises.
The "article" is an ad for Komodo. That's also why it includes 'ten times faster progress than the "normal" Komodo version'.
To be fair, Komodo MCTS is an exciting and new development, and relevant because it can run on a typical desktop machine without a GPU, but you can't expect this piece to be intellectually honest or rigorous.
That's because ChessBase sells Komodo. Don't get me wrong, Komodo MCTS (from test results) does very well and it's quite an innovative approach to make work at that level.
Nitpick: Leela Zero is the go program, Leela Chess Zero is the chess program.
Curiously, there's been some discussion lately in LZ's discord chat about whether its search, or more precisely its evaluation, should be made a bit more minimaxish (i.e., alpha-beta-like). Perhaps the optimum is somewhere in between the two.
A decision can be based on prior information (in the sense of being informed by it) while not being deterministic.
This is how Aristotle and Aquinas handle the relationship between the Intellect and the Will, where the will chooses between goods which are known by the intellect. It is precisely in this choosing between different goods that we are classically understood to have free will.
So, our decisions are determined in the sense that we are only able to choose things that appear good to us, but free in the sense that we can choose between those goods.
We experience choice and decision as subjective processes controlled by a subjective experience of self. But all of those experiences are subjective, and delusional to varying degrees.
I recently had a fascinating experience with a lawyer while buying a house. The vendors and the agent did various things annoy my lawyer, and she set up a situation where the agent was - apparently as a free choice - compelled to pay the lawyer a sum of money to close the sale.
I'm sure the agent made what she considers a free choice, albeit not a comfortable one. But in fact the lawyer was always in control of the situation, and knew exactly where to push, and how hard, to get the result she wanted.
If you have good psychological insight, humans can be ridiculously predictable and easy to manipulate. And said humans will still insist their decisions are made with full awareness and agency - even when they aren't.
So in what sense is will free at all? You really don't need to resort to quantum physics to understand there are serious problems with the concept. Marketing, politics, and law provide plenty of evidence that free will is a convenient fiction, not a psychological ground truth.
In game theory, you encounter situations where a player's optimal strategy is a "mixed strategy", i.e. "x% chance of choosing strategy A, y% chance of choosing strategy B, ...", with at least two nonzero values in x,y,... . For example, in rock-paper-scissors, if your strategy is anything other than 1/3 rock, 1/3 paper, 1/3 scissors, then an opponent can defeat you (in terms of expected value) by choosing the right pure strategy. People don't normally take out a die to make the choice, but... In a sport like boxing, if in certain situations you always "punch high" or "punch low" or otherwise act predictably, that will probably hurt your chances against an observant opponent. That presumably applies to real fights as well. Outside of direct combat, sometimes someone does you a small favor (or a small offense), and maybe the only favor (or punishment) you can do them in return is significantly larger than that; in that case, the appropriate response might be an x% chance of doing the thing and 1-x% of doing nothing.
So nondeterminism is a useful skill in at least some circumstances, quite likely some circumstances that existed in the environment where we did lots of evolving. Therefore I would expect us to have evolved some capability for (apparent) nondeterminism.
How often does the agent deal with lawyers in an adversarial situation? How often does the lawyer deal with people in adversarial situations like that? If an expert is able to see through and manipulate a beginner, well, I don't consider that good evidence that humans can never effectively deploy nondeterminism.
On the contrary, this example only /supports/ the classical understanding of free will, in the sense that it shows the will operates only on goods that we perceive, and not on those which the intellect does not know.
We often operate with imperfect knowledge of our situations. That is simply because our intellects are not all-knowing.
This is pure ideology, built with rhetoric on emotional sand instead of argument from principle.
A difference in outcome by itself in no way lays a requirement on society to 'drive that disparity to zero.' This is a baseless, indefensible assertion and leads to utter chaos when applied across society in the myriad ways one may analyze 'inequalities of outcome.'
Society is necessarily built on ideology (also known as principles). The idea that differences in outcome shouldn't be based on factors like race isn't exactly a radical one. Nor is it based on rhetoric or emotion--if we assume races are equal, differences in outcome tied to race must be the result of societal inefficiencies and failures.
First of all, principle and ideology are very different things.
Secondly and more to the point, there is no reason to privilege race over any other of the millions of filters with which to slice up society to find inequalities of outcome. And even if there was, showing a correlation between one of those filters (race) and an inequality does in no way show injustice 'based on race' no matter how emphatic the assertion.
People with different hair colors, or heights, last names, pet types, and sibling orders on average don't have identical outcomes. And these differences in outcome by themselves are not unjust, are mere correlations, and therefore do not require fixing. Just read Harrison Bergeron by Vonnegut; it's not long.
Pretending that income, one of the most important things in American society, is merely one of the “millions of ways” to classify people, and that a persistent large disparity between racial groups in that metric is mere “correlation” is intellectually dishonest.
That is literally the definition of the word correlation. Claiming 'intellectual dishonesty' is a nice rhetorical move but doesnt actually show anything. The point remains.
A second point above which needs addressing is that it simply does not follow logically that differences in outcome must be a result of 'social inefficiencies and failures.' Culture and biology are some other obvious causative factors, but there are others. There is no reason to adhere to a reductionist view in which everything is caused by social pressures.
> People with different hair colors, or heights, last names, pet types, and sibling orders on average don't have identical outcomes.
There aren't massive systematic factors working to negatively affect specific slices of those population groupings. It's not a good comparison.
That's not to say that Harvard's approach is necessarily the best way to tackle this issue. But it's very difficult to argue that there isn't an issue at all.
>There aren't massive systematic factors working to negatively affect specific slices of those population groupings. It's not a good comparison.
Yes there are massive systemic factors working against people disadvantaged in those factors, and yes it is a good comparison. Look up the average height of CEOs sometime and then tell me height discrimination isn't a thing.
Ah, good catch. I missed the height one. However, I think it's not even close in terms of magnitude. Correct me if I'm wrong, but I remember hearing something like that tall individuals earn, on average, something like $1000 more per year. That's pretty small compared to differences in the usual groups that people tend to focus on.
Just saying that height, facial attractiveness/symmetry, weight etc all could (unquantifiably but obviously) easily outweigh race in terms of societal privilege and it's just absurd to cherry pick the issue of race as the determining factor of privilege rather than something even more basic such as height or facial symmetry.
And that's not even accounting for the fact that (according to many academics at schools that ironically factor race into admission) race doesn't even exist while height and facial symmetry inarguably do exist.
That is a ridiculous over-reading of the notion --- not well supported in mainstream science --- that "race doesn't exist". Scientists who make that claim aren't suggesting that it's therefore impossible for someone to be discriminated against based on race. That would be a self-evidently ludicrous proposition, given observable reality.
I never said it was impossible to be discriminated against based on race. Obviously it is. And obviously race exists. I was pointing out the absurdity of colleges factoring race into admissions while simultaneously employing professors who claim race doesn't exist. If it doesn't exist then how can you justify factoring it into admissions? You can't -- because it exists.
TLDR I agree w u dude.
EDIT: HN won't let me reply to ur reply so let me say -- no my argument is very coherent. If you want to say that "it's not settled if race exists or not" then I agree. And I agree with your assertion that discrimination based on perceived racial characteristics is also obviously possible. I also never said that ppl who argue that race is only a social construct also are arguing that racism doesn't even exist, I don't even know where you got that from tbqh dude. This comment thread is weird to me because you're disagreeing with a ton of things I didn't even say and that I disagree with as well. I don't know why you're so hell-bent on disagreeing with straw man arguments you project onto me because all your points are solid bro and I agree with them all, except for your nonsensical assertion that I don't agree with u. Uhhh trust me bro, yes I do lmao. :)
No, you don't. Your argument is incoherent. It's not settled that race "exists" or what it means. People arguing that race isn't meaningful, or meaningful only as a social construct, aren't saying that there's no such thing as racial discrimination, nor can their arguments be reduced to such a claim.
Outcomes are what they are, all you can do is make the process fair and let people do what they would like to. There are infinite groupings you can make of people (race just 1 aspect) and no matter what you pick, you will see group A be better or worse on some metric than group B.
Wanting to equalize outcomes based on random aspects is a recipe for chaos. Even if we reset the world tomorrow and gave everyone exactly the same resources, the inequality would appear within minutes. Therefore there is no purpose beyond emotional feel-good to say that everyone should be the same at the end because it only leads to misery by all.
