Hacker News Reader

Comment by true_religion | Hacker News Reader

You don’t need statistics to determine that someone has been discriminated against. Here is an example. My friends wife is Latina, and they were buying a house she goes to an open house at 3pm only to be told sorry we already sold. He swings by at 5pm after work, and they are happy to try and sell him the house.

This is your classic N=1 sample size. But what can explain why one had a tour, and the other was lied to? You can invent a lot of things if you insist on giving them the benefit of the doubt, but the easiest explanation is discrimination.

View on HN

Not all N=1 examples should be discounted. Your example is much more believable because there's fairly good reason to suspect that the agent was lying on the basis of nothing more than the race or gender of the person (although there could be other explanations, such as the original buyer pulling out in the intervening time period?).

In the case of the OP, there's no reason provided to suspect racism. Merely the correlation created by two or three data points. That's grossly insufficient for drawing such a link and it's also where the article crosses over into outrage stoking and irresponsible borderline misinfo more than anything.

Well, home appraisals are pretty subjective but if the value of a house doubles then something must be causing it. Theoretically, appraisers are supposed to take into account the neighborhood, the value of the other houses near by, accessibility to work, culture, and shopping, etc. This is why an apartment built on top of a shopping center is worth more than that same apartment built 2 miles away on the other side of a public park.

Now doing the above is quite a bit of work, so it's believable that appraisers simply use their experience and 'gut instinct' to determine the value of the neighborhood without actually going out and doing research. If their past experience tells them that houses in black neighborhoods sell for less, then they'll appraise a house with a black owner as valuable.

This isn't racism per se by the appraiser, but it is however discrimination based on race.

It's quite possible that a less lazy appraiser would have realized the neighborhood still rated X price instead of 2X price.

> but it is however discrimination based on race.

It is discrimination in your hypothetical where you've presupposed that discrimination played a causal role in the appraisal price.

There's no reason to suspect that discrimination played a role in this actual particular case, absent additional information.

Incompetence of the final appraiser or any number of explanations could fit.

You have to realize the layers of selection filters that are at work here:

- Someone is on the lookout for evidence of discrimination

- Someone reports a suspect case to a journalist

- Journalist decides to write an article

- HN user decides to post the article

- HN article gets upvoted and gets to front page

It's like a sales funnel. At each step, even if there isn't any discrimination, it's only the cases that are the most egregious (due to any number of causes, chance being a large one) that do not get pruned by the filters and therefore make it to the final step. Statistical variance can push through a large number of such articles.

That's why this isn't evidence, not even suggestive evidence. We're ignorant of all the filters that led to this last step. The only way to address cases like this where there isn't explicit evidence of discrimination (e.g. chat logs, recording, whatever), is to precommit to a methodology and then go from there. This is what's been done to good effect with things like policing.

You are presupposing a sales funnel, which would play a casual role in pushing an article with flimsy evidence that supports an existing world view (that you presuppose is dominant). For the sake of argument, can you see how there's an inherent bias in your logic?

This is all fine by me, as we're in a casual setting but if you do demand rigor in general.

I see that you're not convinced, but isn't this annectodate enough to prompt an actual study of discrimination in house appraisals carried out by government agents with a strict methodology like you'd prefer?

I was the "HN user who decided to post the article". Which was flagged, and then unflagged, and now flagged again. Apparently this doesn't gratify one's intellectual curiosity.

Last year I saw the following story, this seemed a second data point:

https://www.nytimes.com/2020/08/25/realestate/blacks-minorit...

To me the salient point is that property appraisals are clearly bullshit in the U.S. if they vary this widely in such a short timespan.

And thus ripe for disruption with tech that provides a mere ounce of accountability on the appraiser / appraisal process.

>the easiest explanation is discrimination

If you have decided that discrimination is the most likely explanation and don't bother collecting information or looking at other possibilities, of course it will end up being the best fitting explanation.

This is circular logic.

Can you suggest another explanation?

In absence of any information, I could come up with a few plausible explanations:

The wife spoke with another buyer who tricked her.

The agent was busy or lazy around 3PM

The agent didn't like something about the wife other than race, e.g. attitude, attire, cheap car ect.

Anything but believing black people...

I remember when this same approach was employed until police bodycam footage taught us some cops do just go into a black neighborhoods and plant drugs on poor black people that are innocent, etc...

I guess some people find it unbelievable that one of the most racist countries in human history is the same country they live in, must be nice. Whitewashing history helps in this process.

Reply on news.ycombinator.com