I think you significantly overstate the gravity of their announcement. You can entirely bot the process of finding a flight to suit and handing you a link; open it, pay with stored credit card, and you'll still be entirely within what they explicitly allow.

In practice they don't necessarily have much incentive to do anything about bookings they have some sloppy-implementation reason to suspect are fully botted(in their ideal world comparison searching would presumably be harder; but cat is obviously out of the bag on that one an concealing that almost all of the work was automated will be relatively simple so there's a high false positive risk); but since you were specifically directed to not do that they are better placed to deal with cases where someone's openclaw fiasco buys them a nonsense route and they want to try to make it the airline's problem.

Alleged professionals checking secrets into github and shipping them in retail binaries is at least as old as 'devops' being a thing people say with a straight face; and standards in 'AI' and for amateurs are even lower than in software development generally. There's some diversity in terms of the ignorant vs. the negligent; but in practice egregious mishandling of secrets is to be expected.

Is anyone else surprised that the DoJ ended up asking; rather than Satya just giving Dear Leader the access as part of a lobbying effort?

I don't think linkedin has, or has ever claimed to offer, any sort of protection of user data against administrator access, and the broligarchy has been nothing but eagerly cooperative.

That's honestly the weird aspect, to me. Infidelity is pretty common and unsurprising, even among people who don't dress like the midlife crisis is hitting pretty hard; but if you've got options why pick the almost comically politically compromising one?

Even if it has to be a workplace thing, and wouldn't be the same with some sort of heavy-mutual-NDA escort, there have to be more than a few AI lunatics who want him to be their accelerationist daddy sprinkled across areas Nvidia can actually export to.

This sounds awfully like a cynical rebadge of the same process termination and endpoint isolation features EDR packages all already have. Yup, 'agentic' is software too; so if you think a process tree looks dodgy you can terminate it or isolate the endpoint; assuming none of the (relatively common) 'EDR-killer' malware flavors are in use; in which case hopefully you have a more network-level isolation plan in mind.

It beats having literally nothing short of calling whoever is closest to the power cable and telling them to pull it; but it's only modestly helpful if determining whether the unpredictable-by-design software is doing wicked unpredictability or virtuous unpredictability to the authorized administrator; especially if it's anything time-sensitive.

I've never used the consumer version; but historically there was a significant amount in common between that and the "user state migration tool" they made available for command line operations or SCCM integration. If 'roaming profiles' had ever found a way to suck less I suspect it would have been a nonissue; but roaming profiles always found a way to suck; so that one was a pretty common IT toolkit item.

Salon was a definite mix. A left-wing bias, and some good analysis despite that. But then they also had people like Amanda Marcotte and Arthur Chu writing for them who frequently just wrote nasty attack pieces on anyone they didn't see as sufficiently left-wing, or worse people who came across as weird and nerdy. Look for example at the unpleasant attacks they wrote about Scott Aaronson. If Salon had had any serious ethical concerns or general care about moral approaches to things, they would have fired Chu simply for his own statements that what mattered was "winning" and not anything like an actual attempt at conversation or accuracy. So, yes, there was something unfortunate here. But most of the good parts of Salon died over a decade ago.

What kind of free market are we supposed to run with greater access to price signals and less asymmetry of information?

Brutal collectivists in Whitehall will be demanding low barriers to entry next!

There are two different traditions at work here. One is the European tradition starting with the French Revolution in which the Jacobins chose red as a symbol. The other is a 19th century South American tradition of distinguishing different political factions with the colors red and blue. They are entirely separate traditions.

It's certainly more plausible than some cuts we've seen; but it's hard not to still see it as the usual "described as fat-trimming; inevitably ends up reaching bone" exercise just because of how software usually ships. Not everything ships in full abject brokenness; but at least for commercial-grade software (specialty formally verified stuff and projects that have been laser focused on solving a specific problem that bothered an academic computer scientist in 1980 since 1980 can be a different story) you don't generally see software that makes you think "Nope, it's finished, any additional resources allocated would be wasted or actively spoil the product. There are no edge cases that could have used a little more love, nothing interesting but currently too rough or subject-to-change to use".

Even if you go with a fairly optimistic view of productivity effects, 10%-ish gain and no serious side effects, just axing 10% of the employees so you can status-quo the product with lower payroll costs absolutely seems like a chop-shopping move when the status quo for software is so mediocre. As PE cuts go they don't seem to be speedrunning the hollowing-out and immediately going to a skeleton crew of junior vibe coders and a 'support' chatbot; but precious little software ships in a state where "efficiency is up 10%" means that you've already maxed out possible product quality and just need to start cutting people since you no longer have anything for them to do. If the optimistic projections pan out they might be able to approximately maintain their existing level of polish with fewer people; but that level was always a compromise.

It's good to know that this is purely a matter of technological change; no need to look at the classic "get picked up be PE chop shop"/"gradually get stripped for parts" hypothesis. Seems very credible.

Some of the problems that have arisen here are not the fault of Microsoft or Google or other platforms because they are due to a deeper problem in setting times: times must be set with respect to a particular place - a time should be recorded as a pair of a time and a geographical anchor. Since calendars and other such problems generally do not do this, changes in the relationship between time zones can cause problems.

I participate in a seminar whose organizer is in Ann Arbor. It takes place at 2pm on Fridays. That's 11am for me in British Columbia, three hours behind. Suppose that a change takes place that makes the difference only two hours. The time will become 12pm for me. Their calendars should still be marked for 2pm, but mine should now be marked for 12pm. This is because the time is set by the organizer in Ann Arbor. Hers is the fixed point and mine has to adapt to it.

But suppose I were the organizer? Then the time would be set with respect to my time zone. If I set the time for 11am my time and the difference between time zones drops to two hours, I will keep my calendar unchanged but people in Ann Arbor will have to change their calendars to 1pm.

The point is that it matters which location is the reference point. Does the seminar take place at 11am referenced to Prince George or at 2pm referenced to Ann Arbor? In the current situation, with organizer in Ann Arbor, my calendar needs to indicate that it starts at 2pm Ann Arbor time. The time it would display to me would be computed using the appropriate offset from Ann Arbor time. Any calendar that records a fixed time, whether in local time or GMT will be wrong when time zone offsets change if the time is one whose reference point is not in my time zone. What is needed here is lazy evaluation.

This is not Flock's fault. It's the fault of the police. Flock provided correct information about the suspect car. The police towed the car on the basis of non-unique properties like the model and color, and they ignored the testimony of the owner's neighbors. The police followed a lead and acted as if they had a solid identification when they didn't, something that seems to happen all too often.

Allegedly the same known-vulnerable Danfoss controllers as in the prior french incident; just chilling out on the internet for easy access. I assume we'll have breathless reports crediting agentic hack swarms by midday along with security vendors assuring us that faith in their next generation machine-speed solutions will save us; but apparently the bar is in hell in terms of low-hanging targets.

It also serves a lot of very low budget/someone-involved-in-game-development-did-a-game-jam-thing type purposes. Unfortunately, that's exactly the sort of niche that makes anti-slop work particularly hard.

If it were just about pearl-clutching standards that'd be one thing(roughly the same slop problem for both; but not easier or harder); but if you want to serve the "I did this game jam thing, pay what you want if you think it's neat" you've got a lot of your desired and legitimate activity that would be immediately destroyed if you tried to throw some sort of flat economic floor under the slop guys; and you are...not exactly...dealing in the margins on units sold that pay for tons of human mod effort.

It's not like Steam doesn't have an asset flips and shovelware section as well; but they aren't, and don't pretend to be, in the business of caring all that much about things that fall below the threshold of potential commercial relevance. They do turn a blind eye to a certain number of steam keys going to kickstarter backers; and are significantly less hardass than someone like Apple about how many 'review' copies go out, so it's not like they are actively hostile to stuff that is not yet commercially viable but is trying to get there; but they basically don't touch some of the little test pieces and rough cuts and concept demos and things that itch actively seeks to.