My understanding is that all the big AI companies are currently offering services at a loss, doing the classic Silicon Valley playbook of burning investor cache to get big, and then hope to make a profit later. So any service you depend on could crash out of the race, and if one emerges as a victorious monopoly and you rely on them, they can charge you almost whatever they like.
To my mind, the 'only just started' argument is wearing off. It's software, it moves fast anyway, and all the giants of the tech world have been feverishly throwing money at AI for the last couple of years. I don't buy that we're still just at the beginning of some huge exponential improvement.
My understanding is they make a loss overall due to the spending on training new models, that the API costs are profit making if considered in isolation. That said, this is based on guestimates based on hosting costs of open-weight models, owing to a lack of financial transparancey everywhere for the secret-weights models.
I'm sure they would love to. They've been trying to make their own app store (Galaxy Store) a thing for over a decade. But cutting ties with Google would mean no Google Apps and no Google Play Store, and that would probably be catastrophic for them.
> we give up on the tools that companies use. UX, user research, graphic design, marketing and similar roles are pretty absent from these communities
Some of the bigger open source communities, like GNOME, do some amount of these things. But I think very few people are excited enough about user studies or marketing to do them as a hobby, unlike writing code. It's hard to see how you could beat Google/Apple/Microsoft at their own game like this without a lot of money. Red Hat is probably the biggest company that might be interested in this, but still about 2 orders of magnitude smaller than the giants.
And if someone at the NSF decides to terminate the grant & 'recover all funds', does the dispute over the contract involve the same burden of proof and rights to appeal as a federal discrimation case?
Someone wrote it into the grant agreement. It's a fair bet that they think that has some effect beyond what the law already achieves.
The burden of proof is "on the balance of probabilities" in both cases as far as I know, and there's no limit in principle on how high a breach of contract case can be appealed.
Of course it has an effect, but that effect is giving the NSF the ability to sue over a grantee's alleged breaches of discrimination law, instead of that being limited to parties discriminated against and the EEOCs.
That's definitely not the requirement! The requirement is to avoid doing certain kinds of "research" that the government disagrees is valid research to fund, characterized by the principles underlying it.
They may have started by using certain keywords to find examples of such grants to terminate, but the requirement itself has nothing to do with words and everything to do with the intentions.
The “requirements” are vague and still being litigated against congressional intent, but the problem is the scale: when you have so many complex things to review and only a few trusted political apparatchiks, they end up doing things like simple keyword searches for terms like “diversity” and “inclusion” blithely aware of those being used in fields such as geology.
I know this because I know people who’ve had to take time away from their research to keep their grants from being cancelled.
And maybe they'll get that back eventually, but academia can't complain about rough handling when it steadfastly refused as a bloc to fix its own ways for so long. Outsiders trying to fix them will always create a lot more collateral damage than insiders fixing the problems, but when insiders refuse, outsiders will take over.
Also, frankly, I heard a bunch of such stories and very often the grantees were misrepresenting their work. It actually did have DEI content in it and they were pretending it didn't. You can see how many people in this thread are arguing that all you have to do to comply with the requirements are use a thesaurus or misrepresent their work and then continue anyway. For as long as academics insist on total warfare and malicious compliance, expect universities to be blowtorched.
this and many other grants were singled out because of words used in their description. you seem quite certain that 'diversity' here isn't referring to the degree to which the genome of these animals is similar to different to other of their species, but instead a leftist dogwhistle hiding racist intent, and this researcher lost their position because they are really a secret racist and deserves to be 'blowtorched'.
you celebrate the ruin of the career of a highly trained person, frankly a national resource, because they used a word to describe their work that you think has evil connotations.
It's not a renewal, it's their first application for government funding, and they turned it down without accepting the terms. This is all quite clear in the blog post.
If it has zero weight, why would the grant agreement specifically highlight it? I would guess it's much easier to enforce a particular interpretation of the law via a grant agreement than having to argue it in court.
The "rule against surplusage": Where one reading of a statute would make one or more parts of the statute redundant and another reading would avoid the redundancy, the other reading is preferred.
"Judges frequently invoke anti-redundancy principles in the interpretation of legal language, whether it appears in classic private-law documents such as contracts or classic public law-documents such as constitutions and statutes."
Redundancy: When Law Repeats Itself, John M. Golden (2016)
> Why would the grant agreement specifically highlight it?
I would humbly suggest that it mentions this particular example because the NSF administrator serves under the pleasure of the Executive and they have been tasked to demonstrate that they are following the orders of the Executive branch.
However, the inclusion of this specific example confers no higher priority than any other possible example. It has no weight; it is inoperative.
If it's inoperative then it shouldn't be in the language of the grant. Full stop.
The language itself also overly broad. The stipulation from the grant didn't just cover activities funded by the grant itself. In the very language quoted on the PSF blog, they needed to affirm that as an organization they "do not, and will not during the term of this financial assistance award, operate any programs that advance or promote DEI." Read that again. The language expressly states that they cannot operate ANY programs that advance or promote DEI during the term of the award. So if a PSF member volunteers with PyLadies, would that count as "advanc[ing] or promot[ing] DEI?"
In the real world, no one would _ever_ sign a contract with this sort of poison pill on it. If something like this was found buried in a contract I was evaluating with my lawyer, we'd immediately redline it as overly broad and overbearing.
> If it's inoperative then it shouldn't be in the language of the grant.
It’s not inoperative. A contract requirement that is redundant with a legal requirement still has separate effect (that is explicit here since this clause is a basis for both cancelling an award that has already been made and clawing back funds that have already been disbursed, separate from any penalties for the violation of the law itself.)
> In the real world, no one would _ever_ sign a contract with this sort of poison pill on it.
If by “this kind” you just mean “incorporating existing legal obligations separately as contract obligations with contractual consequences”, every government contract has multiple such clauses and has for decades.
If by “this kind” you mean more narrowly incorporating the specific anti-DEI provisions and partisan propaganda about DEI inside the clause also incorporating existing legal requirements, I’m pretty sure you will find that most federal contracts that have had their language drafted in the last few months have something like that because of agency implementations of EO 14151. How many people are signinf them...well, I would say look at whoever is still getting federal money, but given the shutdown that’s harder to see...
You are claiming that if the PSF took the grant and the NSF, or the president, decided the PSF was promoting DEI they would not be able to claw back funds?
OK, I accept that as a possible reason why it might be written there even if it has no weight. But it still seems very likely that it's easier to terminate a grant - and harder for the PSF to argue against that - than to actually prosecute DEI work and prove in court that it's illegal.
You say, paraphrasing, "It's harder to prove that a DEI program violates Federal anti-discrimination laws than it is to simply terminate a grant to an undesirable grantee."
Ok. Suppose that's true. The government can terminate grants that don't include that language equally as easily -- and, indeed, I just found that there are multiple current cases against the government for doing exactly that: health grants [1], solar grants [2], education grants [3].
Is your point is that the inclusion of this inoperative language makes it easier than it already is for the government to cancel grants and to defend against the subsequent lawsuits until the plaintiffs are pressured into compliance from lack of funding?
Small correction: the restriction would only affect the PSF for the 2 years the grant runs. That's still more than bad enough when 'diverse' is in the mission statement, and of course they might well apply for other grants, but in principle it can't be applied 'at any point'.
I would imagine it is much easier to enforce as part of a grant agreement that organisations have signed. Especially if the law is either not really a law (yet), or it might be invalidated by a court on free speech grounds. There's probably a reason someone wrote it into the grant agreement, and that they're declaring DEI stands for something other than the familiar Diversity, Equity & Inclusion.
To my mind, the 'only just started' argument is wearing off. It's software, it moves fast anyway, and all the giants of the tech world have been feverishly throwing money at AI for the last couple of years. I don't buy that we're still just at the beginning of some huge exponential improvement.
reply