I consider it a feature that acts as a filter.

The internet was a bubble: you could make a web page and sell it for millions because next year it was going to be worth billions. And then internet grew up.

AI is technology that's still beginning to find its place to settle. It's far from mature and that's perfectly fine. We'll have reached a reasonable plateau once the technology and the related stack stops changing every month and instead develops incrementally and boringly over the span of few years. That's like internet in the 2008-2010, and many investors will have a collection of new burn marks by that time.

Not only financially there's an unsustainable push for AI by the zealots du jour who are more often than not managers rather than engineers. AI is championed most ruthlessly as a silver bullet revolution by people who least grasp the limitations of AI. It'll take some time to figure out the dreamed-up proceeds won't be there, and "then what?".

I predict that the real bottlenecks of development will re-emerge as soon as the limitations of AI will manifest out of the hype. They bottlenecks are human-based, in development processes and in human interactions. A large part of development is trying to understand what we want and what we need and you can't offload that to AI.

I think many people have an expectation that (all) laws are just and needed because... somehow they're the law.

In reality, laws can be unjust, unnecessary, biased, and completely arm-wrestled together by people strictly following an agency of their own. Other laws are put together by sheer ignorance and lack of thinking beyond mere good intentions. The first question shouldn't even be "is this law fair" but "was this law made fairly".

It creeps me that people treat laws as axioms whereas they're just polished and reinforced opinions. Sure, many laws we can agree on, and many others that don't agree on aren't worth changing, but you should always question the law and question where it came from before choosing to accept it.

I can see the same pattern with technology such as the various digital restrictions management (DRM) schemes.

This is akin to how two kinds of people respond to law. The first kind think "This is the law, we must follow it" and the other kind think "This law doesn't make sense, we must change it".

People who look at pedestrian traffic lights and cross when it's green vs. people who look at cars and cross when there are no cars coming. The first say you must follow traffic rules and the second kind say they wouldn't be alive if they looked at the green/red light of law instead of whether there are oncoming cars: a green doesn't mean it's safe to cross and a red doesn't mean you can't cross if only there are no cars.

Anecdotally, the cassette player that came with the machine had a misconfigured tape head. Because there was no internet nobody knew why it didn't load most of the games I got with the machine. However, saving and loading programs did work. So, I started writing programs from the user manual and game listings from some programming books I found in the library, and saving them on my cassettes. Because the user manual covered not only some tutorial BASIC but also the machine's graphics, sprites, sounds, and what other features I eventually, after getting some hang of writing BASIC, did also realize that what I could create with the machine hardware itself was virtually unlimited. I didn't necessarily know what the commands did with the underlying hardware but I knew if I poked certain numbers into certain addresses I could make my sprites appear on the screen and make them move around.

By the time I got the cassette player fixed by some computer repair shop, learned about tuning the tape head, and I could finally load all the games bundled with the machine, I was seriously hooked with programming and the highly desired games no longer seemed that interesting in comparison. I knew someone sat down and wrote all those games and instead of playing them I could learn to do the same myself.

Been programming ever since.

Failing to recognize the dangers would be an instant fail; knowing that something reeks of undefined behaviour, or even potential UB, is enough: you just write out explicitly what you want and skip the mind games.

What has generally worked for me is paraphrasing the old adage "Write the data structures and the code will follow" over to AI. Design your data, consider the design immutable and let the AI try fill in the necessary code (well, with some guidance). If it finds the data structures aren't enough, have it prompt you instead of making changes on its own. AI can do lot of the low-hanging fruit and often the harder ones as well as long as it's bound to something.

Yet, for now, AI at best has been something that relieves me from having to write a long string of boring code: it's not sustainable to keep developing stuff relying on AI alone. It's also great when quality is not an issue; for any serious work AI has not speeded me up noticeably. I still need to think through the hard parts, and whatever I gain in generating code I lose in managing the agents. But I can parallelise code generation, trying new approaches, and exploring out because AI is cheap. AI is also pretty good for going through the codebase and reasoning about dependencies whether in the context of adding a new feature or fixing a bug: I often let AI create a proof-of-concept change that does it, then I extract the important bits out of that and usually trim down the diffs down to at least 1/3 or less.

AI further helps with non-work, i.e. tasks that you have to do in order to fulfill external demands and requirements, and not strictly create anything solid and new. I can imagine AI creating various reports and summaries and documentation, perhaps mostly to be consumed and condensed by another AI at the receiving end. Sadly, all of this is mostly things not worth doing anyway.

Overall, I cringe under all the hype that's been laid on AI: it's a new tool that's still looking for its box or niche carveout, not a revolution.

I've held the belief that an occasional bankruptcy is basically a sign of healthy competition within an industry: those companies going down literally didn't know how to be any more efficient or they could've survived.

Regarding airline business, a crapload of more people are flying now with better prices than before the industry was deregulated. Sure it must hurt someone at one end, eventually. Part of the business is standing through price wars because someone will always lose: the best companies can endure that. While airline industry probably fluctuates as described in the article there are plenty of other cyclic industries. Churn itself isn't anything new.

Because fixed costs are what they are, I think, is the reason you can drive the business quite precisely to the brink of inoperation: it could literally come down to pure luck between how full your planes happen to be and how close you are to the next payment of some critical loan whether you can take off into the air for another month or so.

I can't remember a phone that died because of the battery since the era of Ni-Cd cells in early cell phones. I don't think I've never discarded a phone with a li-ion battery because of the battery. It's always physical breakage or getting too slow to be usable, because of age.

Sure, I don't spend a cycle per day. Not even every other day. That's probably rare, I get that. But much rather than because of dying batteries I'd like EU to mandate

- the phone should come with full keys so that I can own the machine if I want to - or at the very least the hardware must become unlockable once the support period ends - individual components should be made available for independent repairs - repairs must not need software pairing of hardware components on unlocked devices

because of right to own and right to repair which shouldn't be "rights" but nonnegotiable traits of physical properties like they used to be.

YES. The beauty of programming is and always was that, first, you enjoyed it and, second, for some oddball reason you could actually get paid to do it. And one can't produce anything good unless you actually love working on it which means you want to put yourself working on it. The outcome might accidentally serve the one who pays for it but ultimately what did get the work finished was the sensation when you were reaching the point where you would finally tie things together and see everything you designed come to life and work together.

AI doesn't give you that personal involvement. We can do it but it's a different line of work and we care very little about what goes in and what comes out. We just do the grunt work of connecting the two ends. We're not for a fuck interested in elevating ambitions which is a word that relates to what is outside while all the good stuff comes from the inside.

My phone can pre-download maps into Google Maps for offline use, I've done this in foreign countries where I didn't necessarily have full data connection. There's no reason you couldn't cache the necessary maps on the car's navigation system and let it operate based on that and an incoming GPS signal, never emitting out one bit.

OTOH, if you wanted live data, dynamic routing etc. for your convenience you could explicitly turn data on but then you'd acknowledge it comes with the caveats such as potential snooping of telemetry data.

Admittedly, I would never trust a car manufacturer to actually disable telemetry no matter what they'd promise. So, disconnecting the antennae would be the only reliable method regardless. I wonder if there will ever be a car with a physical radio kill switch like laptops.

This is how cars used to be made. Features were standalone modules: there could be some bus traffic about optional data (wiper module with rain sensor could broadcast that it's raining and body control module could hear that and could be configured to close windows when raining) but they weren't strictly integrated in any meaningful capacity. You could change the radio unit to whatever you liked: if you were lucky you could get one that can actually understand what the other modules in the car were saying and show some non-enterntainment info on its screen as well. Navigation used to be a standalone system that had GPS receiver but nothing else in the car couldn't necessarily tap into the location data.

SUre, it meant some more wires and maybe the features had disconnects because they weren't aware of each other that much but all in all that was a good thing. It kept everything simple, isolated and repairable. Now because of more integration the modules need to know who they're talking to which leads to bizarre things like having to code in new headlights and pair them with other modules or they won't be recognized and just stay off.