More like "need to switch to this supplier that made bribes... err... campaign contributions".

I don't think I've ever seen DEI actually make a significant difference for the worse, but I have seen *plenty* of incompetent, cheap, greedy, and corrupt behavior around all the sorts of the things the right wingers blame DEI for. The good news about those things is that those are bipartisan at least.

Also, the assessment requires taking the administrations word at the cost and state of the program. They have been shown to lie to exaggerate their "cost savings" repeatedly.

Also characterizing their savings with the big three GOP virtue signaling words is also likely suspect.

I feel that's an optimistic characterization.

Probably the "bad" system was from an unfavored supplier to the current administration, and the "new" system will be awarded to some different pet supplier, that will probably overrun budget and still fail and just be basically even more bad money, but on a later schedule, maybe to be cancelled later.

They are over here pinching relative pennies to cater to the "woke bad, covid is a hoax" sentiment while simultaneously spending bililons a year on just keeping the F-35 planes maintained.

I can't speak to bio applications, but I've gotten a solid exposure to AI in weather.

It is not "better" than the traditional models, in fact the AI models fall apart on a fairly short timeline. However they are useful for very short term forecasts done with less computation. You want to offer specific, precise forecasts over the next 8 hours, AI is going to be a solid strategy. So "it's going to rain on your neighborhood in 17 minutes for 39 minutes" is a tempting one for AI. What's the weather going to be in 3 days, well they come apart by then.

It's kind of like how an untrained person can eyeball a time lapse of radar and be able to guess pretty well what happens next, but you actually need understanding to make longer term predictions The AI models have much more data and much richer data than the human, but the example gives a rough idea of the strengths and weaknesses.

Note that AI in it's current forms certainly have utility. The issue is the magnitude of the investment is large and mostly driven by the parlor trick of text generation that superficially resembles human interaction and presuming that means open ended human replacement grade AI for general purposes is imminent..

Worse than letting them rot, it's about trying to make them "more AI", which on the one hand does make it potentially able to react to a wider variety of inputs, but at the same time makes it much harder to control the specific resulting behaviors compared to their old strategies.

Which is a funny thing to say... T-Mobile owns Mint...

I probably have less exposure to VMWare customers than you (my main job doesn't have a whole lot of intersection with them), but I was sort of attached to a migration effort for one customer.

The customer pretty quickly had a proof of concept of their tech stack without vmware. The tech people ultimately can't find any specific showstoppers. Their management has issued delays to do even a partial deployment into production every time their begin date came within a couple of months. Asking their tech team, they aren't actually doing anything to evolve it, but management is terrified and so long as they can get the company to pay, they will keep going. No one wants to be the one to pull the trigger on a stack change, even as their tech team assures them that the abstractions are such that nothing will notice.

So I wonder how many of those migrations actually follow through versus just sit on the backburner, maybe to be used as a threat to Broadcom sales folk...

As much as I really don't want the Broadcom strategy to work, I think it's broadly working. They really seemed to correctly peg VMWare customers as fearful suckers spending the company dollar they don't care about anyway.

Note that angled right at the sun when the panel is on the back side of the screen would mean either the screen is at an unusable angle, or the panel is not going to get anywhere near the amount needed.

It's a gimmick, much easier to have a handheld foldable or rollable solar panel to go with your laptop, so long as the laptop is willing to accept lower USB-PD rates.

He's not big on consistency in general, but this is an area where he might be fairly consistent. He doesn't like US law enforcement and intelligence agencies broadly speaking. He's about self-enrichment by whoring out the government to the highest bidder, and none of those folks want the government to have this sort of access either.

Who says it can only be a gravity vehicle? If they leave the floor open, they could walk the car, while saying 'yabba dabba do'.

What's amazing is that as dumb a product as the Pin was, that was the *best* Humane had to offer. Why in the world would anyone spend that much to acquire is beyond me, especially while recognizing how bad an idea it is by shutting it down.

Compaq spewed $300 million to get Zip2 back in the day, so HP is just continuing that tradition of tossing bags of money at dubious companies that might have relevance to the buzzwords of the day.

Note that the attorneys may not say "slam dunk you are in the clear" and instead assess that the risk (likelihood and severity) and the business determine it's worth a shot to try. Here in a hypothetical where someone actually brings a case and RedHat loses, the remedy would likely be be "restore the contract of the plaintiff", or basically just back where they would have been if they didn't even try.

On the FSF side, well, even if they wanted to try out an argument, they don't have standing. They'd need a subscriber who was penalized by RedHat that might have standing if it's clear it was termination over exercising GPL rights.

Broadly speaking, currently the ecosystem has settled into two general camps. AlmaLinux, that decided to play it the way RedHat ostensibly approves, and work off of stream and tolerate a bit of deviation from RHEL sources (and play up scenarios where they beat RedHat to issuing security updates, which turns out not to be hard to do) and Rocky, who basically says they'll keep getting an exact copy by whatever means and keep going even as it means some unspecified people are technically violating their subscription agreement to make it happen.

Frankly, they wouldn't even have to go that far. They'd only have to shut the door on non-copyleft content and they've pretty much already made it impossible to clone.

A quick review shows that a random RHEL I could find was only about 66% copyleft. So a full third of the distribution could close up as much as they like. It's already severe enough that you couldn't credibly claim to be a source based clone with so much missing.

The question is whether a license like GPL is able to impose restrictions on those terms and conditions on related. It's quite likely that, as written, the subscriber agreement lets them do whatever they want in terms of cancellation. The question is whether the copyleft suppliers license is compatible with retaliation over exercising rights in the license.