Comment by LudwigNagasena | Hacker News Reader

Comment by LudwigNagasena | Hacker News Reader

> They could also have just cooked up some ostensibly race-neutral metric that resulted in their desired outcome, […]! The EEOC even offers advice to business owners on how to arrange such a scheme:

>> If certain groups of employees are affected more than other groups, determine if you can adjust your layoff/RIF selection criteria to limit the impact on those groups, while still meeting your business's needs.

>> For example, you decide to lay off the most recently hired employees due to budget constraints. Female employees account for 30% of your workforce and 85% of the employees scheduled for layoff. Determine whether you can adjust your layoff criteria in a way that allows you to meet your financial goals while also reducing the impact on female employees. For example, you might determine whether alternative layoff criteria, such as employees' profitability, productivity or expertise, would enable you to reach the desired financial outcome and result in the layoff of fewer female employees.

That’s nuts. The US government literally suggests companies to cook up fake metrics to achieve desired results.

Not fake but better metrics. There are always many different metrics that are available and the choice of metric is often arbitrary. What EEOC is doing is pointing out that if a metric disproportionally affects some protected category then it is likely a flawed metric for the use case. Again, since there are many different metrics that are available, you can just try another one until you find one that satisfies both your business and diversity needs.

In fact it really is not likely flawed. The only way you ever get proportionate presence in anything is if there’s a finger on the scale.

I’m not saying it should be proportional, often times it is impossible to do so. What you can do is check which of functionally equivalent metrics has better non-functional characteristics (eg diversity).

But diversity isn’t better. That’s just a belief people adopted because of a court ruling that says it’s a valid excuse to discriminate. The point is to discriminate.

What you are describing is a bias. You are suggesting we should be bias towards believing a metric is wrong if it disproportionally affects a protected category. One would also assume that if we have it specify “protected category” that we should believe metrics are right even if they negatively affect those in unprotected categories.

That’s a political calculation not a logical one. Also in practice members of protected groups who are over represented at a specific business can be disproportionately subject to layoffs to ensure the protection of less represented minorities. The general bias is toward integrating businesses and favouring protected classes for better or for worse.

No, they suggest one be aware that neutral metrics can none-the-less have an impact which may result in a violation of the law and that therefore other - not made up, just other - metrics be used. I take no position here on this law or the quoted guidance, other than to assert that you have mischaracterized them grossly.

I think you're conflating "violating the law" and "violating EEOC doctrine". The law--the civil rights act of 1964--is pretty neutralist. It doesn't make special exceptions for marginalized communities the way EEOC doctrine tries to do. It feels very much like this guidance aspires to violate the spirit of the law while not running afoul of the most generous interpretation.

The Supreme Court said accidental discrimination can be illegal.[1]

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

Do you think the Supreme Court would say the same if they had a case today...?

Agreed. Imagine the IRS publishing guidance for exploiting tax loopholes.

that’s the EEOC’s MO, and the behavior of most bureaucracies like the IRS, they signal extralegal expectations and dare you to challenge them on it, the system working as intended

Reply on news.ycombinator.com