No, quite wrong:

the New York City Fire Department used written examinations with discriminatory effects and little relationship to the job of a firefighter (emphasis added)

The Court found that the City’s use of the two written examinations as an initial pass/fail hurdle in the selection of firefighters was unlawful under Title VII.

Now in addition, the court did say,

the City’s use of applicants’ written examination scores (in combination with their scores on a physical abilities test) to rank-order and process applicants for further consideration for employment violated Title VII.

But if they believed that a candidate with a better score was (ceteris paribus) a better candidate, they would presumably have no problem with this. Remember that people who want you to use AA probably won't trust your judgment alone. (ETA ceteris)

The US govt passed legislation that no one could argue for?

I think Eugine is arguing that they passed legislation for reasons that are not rational.

Sure. Here's an article from the Economist, but Thomas Sowell also wrote a book about the issue called Affirmative Action Around the World. I should also note that the national reservation system is not quite a century old yet, but reservation systems of some sort have existed for longer.

I should note that I am a fan of the policy that 'affirmative action' originally described- that is, taking action to affirm the government's commitment to meritocracy over bias, in order to counteract the self-fulfilling prophecy of people not applying because they don't expect to be hired or promoted on racial grounds- and am a strong opponent of reservation systems that 'affirmative action' is now used to describe. Officially, reservation systems are illegal in the US- but it's hard to see how one should interpret 'disparate impact' any other way. ('Disparate treatment' is the American word for anti-meritocratic bias, and so American systems have to be a tortured mess that is not too meritocratic (or it's racist) or too anti-meritocratic (or it's racist).)

A handful of claims:

• AA is a temporary fix / AA is a permanent fixture of society:

It proposed that the policy exist for a decade to see what progress would be made, but without spelling out how to measure it. The provision has been renewed without fuss every decade since.

(On this subject, reading Sotomayor's questioning during affirmative action cases that come before the Supreme Court is an... interesting experience.)

• AA will not lead to loss of quality / AA will increase corruption and decrease meritocracy:

Worse, the policy has probably helped to make India’s bureaucracy increasingly rotten—and it was already one of the country’s greater burdens. An obsession with making the ranks of public servants representative, not capable, makes it too hard to sack dysfunctional or corrupt bureaucrats. Nor will this improve. In December 2012 parliament’s upper house passed a bill ordering that bureaucrats be promoted not on merit alone, but to lift the backward castes faster.

• AA will decrease resentment between groups / AA will increase resentment between groups:

For secondary schooling state funds help to encourage more Dalit and tribal children into classrooms; the effect of setting aside special places in colleges and university is to lower the marks needed by Dalit and other backward applicants.

That causes resentment among general applicants, who vie for extremely competitive spots in medical, business and other colleges.

(One weird quirk of psychology, here: suppose there are 10 slots, and 100 applicants, 10% of which are Dalit, so one of the slots is reserved for a Dalit. If the top Dalit scores 20th best on the test, numbers 10 through 19 all feel as though they have been deprived by the Dalit taking the 10th slot, even though number 10 is the only person actually deprived.)

• AA will lead to homogeneity and acceptance / AA will lead to heteogeneity and perpetuate divisions.

Pratap Bhanu Mehta, an academic at the Centre for Policy Research in Delhi, favours affirmative action but concludes that a policy focused on distribution of limited state resources is bound to fail. “The current system is not about equal opportunity, it is about distributing the spoils of state power strictly according to caste, thus perpetuating it”, he says.

• The benefits will go to the poorest and most deserving / The benefits will go to the richest and least deserving.

The Economist article doesn't discuss this directly, but others (that I don't have time to find now) do. There's a 'creamy layer' provision to try to prevent the richest of the Other Backwards Castes from benefiting (to convert to an American example, if your parents are millionaires, you probably don't need AA consideration even if you're black) but this does not apply to the Scheduled Castes (Dalits). The hypothetical highest scoring Dalit mentioned earlier almost certainly comes from a rich Dalit family, and by looking at the subdivision within caste of the various beneficiaries of reservations it's been shown that the majority come from the SCs that were already privileged within the SCs.

What's irrational is the belief (or rather alief) that anyone arguing that the cause of the observed differences in intelligence by race isn't caused by white racism (or for that matter anyone pointing out said difference who doesn't immediately attribute it to white racism) is an EVIL RACIST. AA is just one consequence of this irrationality.

But you didn't appeal to the data. You appealed to a single data point.

In any subject that's not politically charged, people don't argue that they can see a weak effect in a single data point.

In health science a lot of observational studies that gather way more data don't replicate. You can't simple throw out everything we learned in statistics out of the window just because we are talking about a political charged issue.

VAuroch is pretty obviously of the opinion that admissions criteria are poor measures of talent, and that in particular minorities are more likely to score poorly on the admissions criteria for reasons other than talent.

Even if that were true, affirmative action is based on admitting a certain percentage of blacks. Thus unless he (or you) are claiming that the average black has more talent than the average white, the amount of talent a black needs will still be less than the amount of talent a white needs.

(there being no way to measure talent directly). VAuroch is pretty obviously of the opinion that admissions criteria are poor measures of talent

I would be interested to know what people consider to be better "measures of talent" than those usually considered by admissions office.