Hacker News Reader

Comment by mtgp1000 | Hacker News Reader

Only one answer is socially acceptable, regardless of what statistics indicate. And we are building a dangerous world by going down this unscientific path.

I am genuinely concerned that this will bleed into ML science and we may find ourselves collectively pursuing a rabbit hole of bad science in a vain attempt to reconcile empirical data with ideology.

View on HN

This preference for ideology instead of data reminds me of weird stuff that would happen with science in the Soviet Union. You had things in the Soviet Union like Lysenkoism[1] where a kook got power and explained his science in ideological terms and his theories, unsupported by data, were adopted and defended to the detriment of agricultural production. Workers adopted the theories because people were so afraid of criticizing the dominant ideology that they adopted whatever "science" was mandated in order to avoid being purged. Disagreement with official science was considered proof that one was a political subversive.

"More than 3,000 mainstream biologists were dismissed or imprisoned, and numerous scientists were executed in the campaign to suppress scientific opponents."

[1]https://en.wikipedia.org/wiki/Lysenkoism

> Joseph Stalin supported the campaign. More than 3,000 mainstream biologists were dismissed or imprisoned, and numerous scientists were executed in the campaign to suppress scientific opponents.

Wow, so what sort of rationale can we make of this? Is the logic, "Stalin supports bunk science theory", "other scientists critique said theory", "therefore they indirectly critique Stalin and reduce his legitimacy/credibility"? Something like that? Are politicians expected to grasp every scientific thing they must evaluate accurately? Why wouldn't they accept and admit 'yah maybe we got this one wrong, sorry honest mistake' than to go out and murder folks. That seems widely overreactive.

When you get a paranoid murderous psychopath like Stalin in power, "rationale" may be stretching it.

But I think it went something like this. Marx was "scientific" - Marx had created a "science" of economics and sociology, and it was the official truth of the USSR. If some crank could show that his crackpot theory of agriculture was the logical consequence of Marxism's established truth, then it too had to be the official truth of the USSR. The science was settled; all that was left was for everyone to fall in line and implement it. Anyone who opposed that was both unscientific and unpatriotic.

Except the science was objectively wrong, and implementing it resulted in disaster.

Note that even you only mentioned whether answers are socially acceptable (in your view), not which one is true.

That's the fundamental problem. We are ruling out possible results in advance, based on arbitrary social conventions, not data.

This is fundamentally unscientific and will lead to stagnation at best, regression at worst, and progress only in the unlikely case. At some point we will have to reconcile the fact that the same mathematical and scientific techniques which we correctly apply to other sets of data produce unacceptable results any time models are conditioned on demographic data.

What then?

Only 13 percent of the population, you say?

Reply on news.ycombinator.com