Hacker News Reader

Comment by quotemstr | Hacker News Reader

A lot of really bad policy gets enacted because advocates don't honestly ask themselves, "What are the foreseeable consequences of the policy I'm advocating?" That is, a failure to ask the question "And then what?" results in a lot of unnecessary misery.

Your comment essentially restates the HUD's complaint. It doesn't engage with my discussion of the fundamental incompatibility of the HUD's regulatory regime and ad targeting in general. Nobody was ever convinced of anything by someone typing "FYI: [restatement of original premise]".

View on HN

My point was your premise is making up imaginary scenarios that aren't actually happening as a way of creating an issue with the clear problems outlined and this isn't helpful.

Creating an imaginary general problem as a way of attacking the specific issue is disingenuous.

More generally, housing discrimination will happen at the individual level in all sorts of subtle and not so subtle ways that can’t be generally stopped. That doesn’t mean you throw your hands up and say that we might as allow any sort of discrimination.

If the intent of policy X is to achieve goal Y and X does not in fact achieve Y, what is the point of continuing to pay the costs of X instead of proposing some X' that actually does achieve Y? Continuing to advocate X when X doesn't achieve Y suggests that the proponents of X don't really care about Y, but instead some unstated Z.

Because a simple and easily enforced X that hits 90% of cases is vastly preferable to the 1200 page tome that makes up X’ and requires 175k people to enforce after 20 years of law writing.

In some situations, "perfect is the enemy of the good" is a reasonable approach. Unfortunately, we're not dealing with a good-but-imperfect filter, but a practically useless filter, one that imposes significant costs on top of being ineffective.

"Weapons of Math Destruction" presents good evidence that 90% is much lower than the actual effectiveness of X.

Reply on news.ycombinator.com