It should be noted that this has not revealed anywhere near as much as was being eagerly anticipated. So far nearly all the screenshots I've seen passed around are relatively low-follower barely-known accounts even in each one's own aligned political sphere.
I'm listening to ATC for education. I'm not confident in my ability to understand it correctly. That's why I qualified my level of certainty.
FWIW, the KBOS incident I saw didn't seem to be ATC either, if it was actually what I thought it was. It does seem like either of them may have been caught earlier if there had been more ATCs on staff or if they weren't as stressed or sleep deprived.
There is a dry pasta I use that, long story short, comes without a listed cooking time, whose correct cooking time I have experimentally determined to be ~18 minutes (though remarkably flexible, good at a much wider range than "normal"). I like it quite a lot (even though it seems to have the teflon-die surface rather than the bronze-die surface).
I think greater pasta thickness is underexplored, and the teflon-vs-bronze die thing as the highest determinant of pasta quality, while not nothing, is slightly-overstated r*dditry.
Bronze-die pasta has an obvious and substantial textural difference from teflon-die pasta. The stickiness of the bronze requires more force from the extruder, but results in a rougher surface on the pasta, because it literally sticks to the die.
Bronze-cut pasta holds sauce much better, especially for thinner sauces. It also makes your pasta water more starchy, since it loses more material during cooking. These things seem very obvious to me via my observations as a cook who uses both from time to time (but mostly the bronze stuff).
Both properties can be very useful (the first to everyone, the second just to those who use their pasta water in the sauce step).
It's good to question our assumptions from time to time, but there's no reason to just deny something like this with absolutely nothing to back it up.
I don't deny that it is beneficial (it clearly is, in my direct experience as well): I doubt that it is the highest determinant of quality, and suspect that even more basic properties like thickness have been systematically neglected and may be more consequential.
This sounded interesting, so I went and read a few articles. It seems, dies come in 5 categories: bronze, brass, steel, teflon coated (various bases) and plastic [0].
The bronze (and even brass) are uncoated and don't seem to lose material, on the contrary, they seem to get a patina with use. From what I read, bronze pasta is extruded at lower speed and temperature to account for the material (and the desired texture of the pasta). From an engineering point of view, this article give more insight [1].
This involved a completely nonsensical and arbitrary biographical screening test, which:
1. was designed to statistically select for members of favored identity groups and against members of disfavored identity groups, and not in any way to measure ATC job aptitude, resulting in highly-scored questions like "The high school subject in which I received my lowest grades was" where the only correct answer was "Science", and failing the test disqualified you permanently
2. then-current FAA employees distributed the exact answer key to outside racial identity organizations to give to their members
15. The high school subject in which I received my lowest grades was:
A. SCIENCE (+15)
B. MATH (0)
C. ENGLISH (0)
D. HISTORY/SOCIAL SCIENCES (0)
E. PHYSICAL EDUCATION (0)
16. Of the following, the college subject in which I received my lowest grades was:
A. SCIENCE (0)
B. MATH (0)
C. ENGLISH (0)
D. HISTORY/POLITICAL SCIENCE (+15)
E. DID NOT ATTEND COLLEGE (0)
29. My peers would probably say that having someone criticize my performance (i.e. point out a mistake) bothers me:
A. MUCH LESS THAN MOST (+8)
B. SOMEWHAT LESS THAN MOST (+4)
C. ABOUT THE SAME AS MOST (+8)
D. SOMEWHAT MORE THAN MOST (0)
E. MUCH MORE THAN MOST (+10)
Been hearing about this super racist DEI questionnaire for a while now. I cannot believe this is what people have been talking about? These are such normal corpo performance review nonsense.
The first link describes it as one half of a two-step screen where there are known biases in the second step, and there are far more applicants than positions. So the entire point of this quiz is for it to have a deliberate designed complementary bias, so that the outcome of the two tests combined gave a score that was statistically correlated with the desired results and NOT statistically correlated with characteristics they did not find useful.
Is your argument that this is a bad goal to have, or a bad method of approach, or that the quiz created cannot possibly achieve this goal?
“We’re going to flip a coin and if it’s heads and you guessed tails then you're fired, but it’s okay because our research shows that people of your racial group are more likely to guess tails and we think there are too many of your kind around”
It's not DEI though; it's just standard corruption.
The answer key wasn't provided to _any_ black candidate. It was provided to National Black Coalition of Federal Aviation Employees. Members of that group had an advantage while black candidates not of that group didn't.
corruption is still bad but much like if I stab you it's not a mugging unless I also steal something (both still crimes).
--
Also, are they using an AI image? The woman's head in the bottom left table is like exploded (and DALL-E in the URL)?
Yea - if the end goal of "there are >x qualified applicants, so give me x acceptances where all of them are qualified and the ratio of group1/group2 in acceptances matches the ratio in the applicants" was met, and nobody could see how race came into the test, I don't think anyone would complain. (For instance if they'd waited 20 years and wrapped it in AI, or maybe if they'd just added it as an additional section in the test. It seems that everyone agreed some qualified applicants weren't going to make it through because there were more of them than acceptances.)
My understanding is that people complained a) because it did not meet that goal of all acceptances going to qualified applicants, but I certainly haven't read enough to judge that and b) the rollout sounds like a chain of poor decisions - even just splitting it into two separate result steps was guaranteed to raise ire from people who got rejected at that new first step, which would have been reduced if they'd simply made it an internal factor calculated at the same time as the exam result.
edit: but I take it that your argument is "that's a bad goal to have"
This story has really annoying results, because while there were dumb decisions made about the hiring process, people are also blowing some things out of proportion.
There's lots of terrible personality tests in recruitment and they're sometimes abused for various purposes. This one is just mildly bad compared to for example corpos hiring people to analyse the signature/writing style of the candidate. But handing out the key to that test was just terrible.
Then there's another one where people reacted strongly to someone handing out highly scoring words for the resume... where the words would be included in any basic coaching like "leader", "ownership", "delivered", etc.
It's hard to even talk about this when people have kneejerk response to a few key phrases here.
Bad: Writing a nonsensical personality test that ostensibly attempts to select for individuals with a certain background, while actually doing no such thing.
Worse: Making it a pass/no-pass test where if you get any "wrong" answers, you are permanently ejected from the hiring process regardless of education, skill, or in-person interview results.
Egregious: Distributing the answer key by phone to members of a specific DEI action group and telling them to keep quiet about it.
This is not "overblown," it's literally what happened to the FAA hiring process. Some of the most outspoken critics of this scheme are members of that DEI group who were told to cheat on the test, refused to cheat, didn't get the job due to not cheating, and now have little hope of ever landing the job they trained for.
It’s not free money, it’s high risk with a net positive expected return. Any significant profit would carry an irresponsible level of risk. Significant profit without significant risk would take many bets, which means sustaining the accuracy advantage over broader subject matter, which means lots of time spent, which means it’s time for money, which is just a job.
Suppose I gave Provost 5% chance of winning the papal election. Then I would have been more accurate than Polymarket. But I wouldn't call betting on what I perceive as 5% chance of winning "making free money"; from my perspective it would still be a wild risk to bet any significant money on that outcome.
There was a horrendous problem with gambling on the election at one point. I believe the most recent episode of "Tasting History with Max Miller" covers this.
This is not truly solvable. There is an extremely strong outer loop of optimization operating here: we want it.
We will use models that make us feel good over models that don't make us feel good.
This one was a little too ham-fisted (at least, for the sensibilities of people in our media bubble; though I suspect there is also an enormous mass of people for whom it was not), so they turned it down a bit. Later iterations will be subtler, and better at picking up the exact level and type of sycophancy that makes whoever it's talking to unsuspiciously feel good (feel right, feel smart, feel understood, etc).
It'll eventually disappear, to you, as it's dialed in, to you.
This may be the medium-term fate of both LLMs and humans, only resolved when the humans wither away.
reply