I Am Deleting the Blog

Slate Star Codex says they were expecting a relatively nice article, not a hit piece.

Are you referring to Episode 207? https://overcast.fm/+KhqFMR3J4

I think Sam did a very poor job in that episode -- he was preaching exclusively to the choir. He spends the start of the podcast explaining the important distinction between justified and unjustified police involved killings. This is a very important distinction, and I would love to see data about the racial breakdown of unjustified killings, relative to a racial breakdown of police interactions.

But, Sam then completely abandons this distinction. He discusses "An Empirical Analysis of Racial Differences in Police Use of Force", Roland G. Fryer, Jr. July 2007, a NYC-only study that does not measure unjustified killings.

Then, as usual, he spends an awful lot of time spouting his usual rhetoric of truth, epistemology, science, data, facts, and knowledge. This is merely rhetoric because his reference to that study is clearly meant to be evidence that we have the truth -- that police brutality is the issue, and unjustified killings do not disproportionately affect black Americans. He even discusses the Fryer study and moments later is confidently stating "race isn't the relevant variable". This is a grand claim that can't possibly be justified based on the Fryer study.

All interspersed with more rhetoric such as:

- "expiation of sins" for you Botox as if you're "woke as AOC"

- "ecstasy of ideological conformity"

- "woke analysis" is where "democratic politics goes to die" (probably means Democratic Party politics)

- "social activists playing chicken with the forces of chaos"

- "form of political pornography"

- "unable to speak or even think about facts"

Sam's usual parade of platitudes about epistemology are best understood by another quotation from this episode: "the difference between the branding of a movement and its actual aims, that's why propaganda works".

This is a pretty easy conclusion to come to theoretically, too. Any position of power is going to tend towards abuse and incompetence if there's not some sort of filter in hiring for or sustaining the institution. Newspapers live or die by clicks and subscribers: There's no incentive towards any notion of "journalistic integrity", there's no filter ensuring that journalists are especially intelligent or honest, and there's no reason to believe that the typical journalist is any less likely to abuse their position than the typical police officer.

Just as there are individual dedicated, ethical police who believe deeply in fulfilling their mission the right way, there are good journalists out there who make the world better. But for every Ronan Farrow, there are a thousand Farhad Manjoos and Cade Metzes; for every Foreign Affairs, there are a hundred New York Times or Fox News. The net effect is the same as with police: understand that we've got a horridly imperfect system chock full of dishonest actors and engage with it on those terms. Don't talk to cops without a lawyer; don't talk to a journalist without a PR person and/or a specific plan for what you're getting out of the exchange and how to protect yourself from exploitation[1].

The Internet has greatly accelerated this trend. Pre-Internet, if by some miracle you managed to get enough honest, intelligent people together in a single paper, you could establish a culture of journalistic ethics under the aegis of the slack afforded by your local monopoly on distribution. But in the Internet era, you need to be fully competitive on the terms defined by the market, which, as described above, don't point towards honest, ethical reporting at all.

The tragedy with journalism is that government is usually a useful tool to address this problem: well-crafted regulation can shape incentives such that you don't need to rely on wishing for good cops, which is the direction that police reform discussions are taking. A heavy government hand, however, is anathema to the role that journalists are supposed to play in modern society, so this tool is off the table.

I've thought about this for a very long time, and I don't know how to solve this.

[1] This obviously doesn't apply in narrow cases like "observation from man on the street"

I agree with all of your points about the diffusion of responsibility that is common in ML, though I think you may not be sensitive enough to the harmful framing being created by the "anti-bias" side.

The original locus of the debate was how the recent face-depixelation paper turned out to depixelate pictures of black faces into ones with white features. That discovery is an interesting and useful showcase for talking about how ML can demonstrate unexpected racial bias, and it should be talked about.

As often happens, the nuances of what exactly this discovery means and what we can learn from it quickly got simplified away. Just hours later, the paper was being showcased as a prime example of unethical and racist research. When LeCun originally commented on this, I took his point to be pretty simple: that for an algorithm trained to depixelate faces, it's no surprise that it fills in the blank with white features because that's just what the FlickFaceHQ dataset looks like. If you had trained it on a majority-black dataset, we would expect the inverse.

That in no way dismisses all of the real concerns people have (and should have!) about bias in ML. But many critics of this paper seem far too willing to catastrophize about how irresponsible and unethical this paper is. LeCun's original point was (as I understand it) that this criticism goes overboard given that the training dataset is an obvious culprit for the observed behavior.

Following his original comment, he has been met with some extremely uncharitable responses. The most circulated example is this tweet (https://twitter.com/timnitGebru/status/1274809417653866496?s...) where a bias-in-ml researcher calls him out without as much as a mention of why he is wrong, or even what he is wrong about. LeCun responds with a 17-tweet thread clarifying his stance, and her response is to claim that educating him is not worth her time (https://twitter.com/timnitGebru/status/1275191341455048704?s...).

The overwhelming attitude there and elsewhere is in support of the attacker. Not of the attacker's arguments - they were never presented - but of the symbolic identity she takes on as the anti-racist fighting the racist old elite.

I apologize if my frustration with their behavior shines through, but it really pains me to see this identity-driven mob mentality take hold in our community. Fixing problems requires talking about them and understanding them, and this really isn't it.

Loading a page and choosing not to run the code they ship with it is not theft.

The NYT has reported on Bansky without revealing his name. Why is this different?