acuteaura

I mostly avoid this place these days, but there are certain topics where people hyperrationalize, i.e. they (silently) accuse others of being reactionary because an argument looks like reaction to them on the surface.

I started self-censoring on HN because I knew some very normal and on-topic things to say would be voted to oblivion because they had the shape of a controversial topic that the average HN commenter can not relate to except through some mainstreamed right-leaning cultural narrative. The mere mention that I was an affected party in some culture war adjacent thing was treated as if I had invoked a thought-terminating cliche. It was very... anti-curious.

So I principally agree on the conclusion, when emotions get activated this site gets worse. I just see the impact differently - anything that deviates from the libertarian-right leaning default is pushed away. I'm still relieved I had my pseudonymous account with a lot more HN internet points wiped and don't have to care about this site on an emotional level anymore.

My secret hope is that Tim Cook was the main weight behind cutting off half the new features from Apple software in the EU an attempt to weaponize their users against legislators. Even features that have had analogues where nobody would've ever asked for interoperability (i.e. Handoff/Continuity exists in the EU, but iPhone mirroring doesn't)

This is not how politics works here.

I should clarify that my doctor explicitly lied to me (found out by requesting my medical file after moving) to shock me on purpose because of his own fifth grade understanding of psychology.

There is a fine line between being opposed to fat acceptance and thinking performative cruelty works.

(It's does not. It made me so much worse. I have nothing but bad things to say about the medical "professional" who catastrophized my diagnosis to "wake me up" and then shrugged when my response was depressed resignation.)

I don't see the contradiction to what I said or what the prior commenter said.

CPU requests still protect you from something using all the resources, because the weights correspond to your number vs all the other pods on that node, and the scheduler will not schedule pods so they exceed the total available resources.

So more accurately, requests prevent exceeding limits under contestion, and limits always do.

Where do you think the contradiction is? A CPU limit of 1 does not prevent an application from using 8 cores worth of compute at once, it just limits the compute to 1 core per 100ms slice on average, and usually that means getting unscheduled for 7/8 of the slice, if the app is using all available resources and nothing else contests them (weighted by requests).

Some workloads will also consume all the resources you hand them without being latency sensitive at all.

I've handled outages of CPU time available getting suddenly compressed (we were running pod priorities with staging/prod on one cluster and up to 70% spots in 2019) and then learning that some very important applications outgrew their original requests, gone unnoticed because limits were removed a year or so prior. You can fix this with monitoring/right-sizing tools, but that requires your org to not be dysfunctional, and my style of platform engineering usually has to account for the org being very dysfunctional.

It's not as much an "overhead" as it will mess up your latency. Limits don't stop you from using available resources until you hit the relative allowance in a CFS window, so a 1 CPU limit on a 32 CPU machine at worst gives you 32 cores for 3.3ms every 100ms.

More literally a Microsoft product. As in, rebadged Azure Pipelines.

I'm not sure if this is a joke or not, because it has all the signals of a joke, but it might be closer to true than I'd like to admit.

At my work, the average quality has gone up with "blindly trusting Opus". The failure modes are horrendous and the code is verbose as shit, but it still works better. Most "engineers" aren't very good at writing code.

Oh, what a wonderful shorthand to dismiss an entire article.

GitHub Actions was always a mess. Because it is a Microsoft product.

If you want to have a horrible time, try reading some of the runner code. It's early 2010s-style Windows-First, MSFT C# crud that has trouble not racing several threads to inconclusive status codes.

revolutionary. you can do this with 2 lines of shell code.

i do this: https://github.com/acuteaura/universe/blob/main/overlays/cla...

This is like blaming nicotine addiction on the user and not the nicotine for the purpose of defending nicotine use.

How people use AI is a structural problem. Just because you have an easy time using it to learn (good for you, genuinely) doesn't mean that such a product existing in a free market capitalist society isn't going to end in disaster where people will at best fall victim to the dunning-kruger effect constantly and stop valuing their peers' time and at worst roast their brains by unlearning how to acquire actual skills.

And it's not like the AI labs don't know this is happening. Their research arms regularly publish papers confirming dependence and deskilling are not uncommon in very carefully worded language. And still Anthropic is happily extending your Fable trial!

Yes sure, my coworkers are the tip of the iceberg here, but that doesn't mean AI is value-neutral. And we're not even discussing the ethics axis (because honestly, tiring; capitalism doesn't value ecology or any ethics when inconvenient and AI isn't special there)

I'm not convinced the result will not be "good enough". It looks like one might get away with shipping half-baked shit, just like before, but the speed difference incentivizes it a lot more now. Output quality is not really the issue I'm looking at. More what this has done to my own an my coworkers capability to care and mental health.

i consider it intrinsic to the technology. i don't think most people can resist the temptation of feeling a lot more productive than they reasonably can be at the cost of understanding what they're doing (and the fundamental skill of caring about their own work; turbo alienation).

and those that don't fall into this trap will be dragged in sooner or later by peer pressure and bullshit KPIs.

access to knowledge isn't the problem. if you use models to learn, good for you.

people around me use models to replace their critical thinking though, and i consider this a structural issue - with how incentives are laid out at my workplace but also as a cognitive trap where you think you get more done, but you're actually unlearning fundamental skills instead of supplementing your existing ones.

this is a good summary: https://link.springer.com/article/10.1007/s00146-025-02686-z

AI has a corrosive effect on everything it touches. I go to work and notice that everyone else is on a path to mentally checking out. Nobody values anyone's time anymore, nobody seems to weigh the cost of maintenance or conceptual work given creation is seemingly free. People get used to taking shortcuts, both for their work and also in their thinking.

I feel like the only sane person left in my team, like everyone else has stopped giving a shit, pretending to get work done at insane velocity on borrowed time.

It's no longer "can AI be useful" to me, it's "how do we avoid people frying their fucking brains".

As the reporting party of the much less severe (and much less interesting) TS-2022-003[1] I can confirm that vendor interaction is also swift when not responding would not light the company on fire. In fact there were several other vendors that were affected that have yet to mitigate it.

[1]: https://notes.acuteaura.net/posts/github-enterprise-security...