belZaah

That’s an interesting one. On a personal level, there are clear benefits. On a company level, there’s very little visible. Draw the ratio of operating profit to employee count over the past 10 years and there’s no change there you could clearly attribute to AI. It would seem the engineers are an important part of a machine, but not _the_ machine: there’s other bottlenecks that hide or consume personal-level gains.

Sure, but, in absence of money, where are the open models going to come from? They take a lot of effort and resources to train. The current ones have been made in hope the growth will lead to a payoff one way or another. With the two main giants struggling and the market cooling, that payoff becomes quite uncertain.

But do you really see a change in the technology sector? I don’t. If you plot operating profit per employee for Apple and Alphabet, it’ll show a trend up but since COVID. If you do the same for MS, it’ll show a steady trend since Nadella became the CEO. Amazon is flat, so is Tesla. Can’t really see a change caused by AI.

There’s usually a very good reason for that. It’s called complementary products. You don’t use something because it needs something else to be useful. That something else is not being made, because there’s no use for it, because nobody uses the original something. Communities can be like that: a product doesn’t fly, because there is no community and there’s no community because there are no users.

This is why Apple Vision pro did not become popular. Without software make use of its capabilities, it’s just an inconvenient and expensive extra monitor. But nobody would revamp their UX for a product nobody has.

Have published on this exact issue: Kütt, Andres. "Technology diffusion in ecosystems: a new model." Procedia Computer Science 268 (2025): 34-42.

Yup, there’s righteous dub involved in orbital adventures

Ukraine, for whatever reason, was denied access to the DARPA innovation pipeline early on. Thus, they had to make their own to survive. Now, not only is there something moving much faster, than that flow of innovation but also there’s a hunch of adversaries, who have picked up on it. That’s what you get for the lack of strategic thinking. Your decisions are not terminal, your moves are not the finishing ones. Others get to make theirs as well. And you might not at all like where these moves will take you.

15+ years of Estonian public sector IT experience, former chief architect of the country here. AMA.

This source code being open is great, but it’s not the wole story. E-voting is deliberately designed to fit into the context of voting in general and into the societal context. Your paper vote always overrides your electronic vote and there’s provisions in the law allowing cancellation of electronic voting and immediately falling back to paper. In the societal context, electronic identity and signatures are ubiquitous, voting is the least it’s used for.

Thus, the question is not if the system the code implements is secure but if the whole thing is. Of course, if you find a security issue in the code, get in touch and your contribution will be welcomed. But please do not claim e-voting to be inherently insecure without an understanding of how it works on the ground and what happens around it. Happy to explain.

The societal context is different. In Estonia, everyone uses the id-card for everything and has for 10+ years. Everyone understands, that handing your card and pin-codes over has severe consequences. Also, the e-voting does not function in isolation. If a person has voted electronically, whatever paper vote they cast will override their electronic vote.

Oh my sweet summer child. During the paper voting, only the act of casting the vote happens on paper, the rest (tallying, allocation, voter lists etc) is all electronic. Both in Estonia and elsewhere.

That doubt usually turns out to be politically motivated rubbish. People publish nice blog posts and videos before their articles get published. People scolding there being an open wifi hotspot during vote counting. People being explicitly funded by the political parties not liking the electronic voting.

But the rest is very useful. During last elections, one of the opposition parties I’m not a fan of finally got their act together and put actual pressure on the electoral committee rather than ludicrous nonsense: procedural issues, hygiene etc. That’s great! Very much appreciated. When the op sais we don’t have doubts he probably means we don’t have doubts about there being sufficient process to keep this thing getting better

There’s a reason all ballots are destroyed after the contesting period ends. Someone in the future might find a way to compromise that anonymous key and there’s no reason to take that risk.

True. Nothing can be fully trusted. That’s true for everything, though, including paper voting. Clear schemes exist to cheat there and there’s computers involved in tallying the votes anyway. So what’s the reason to trust that stuff? The goal of e-voting is not to become absolutely secure, they need to be secure enough not to make the elections as a whole less secure when used in them. That’s a big difference.

There’s also the bear mechanism. In the forest, you don’t have to run faster than a bear, you need to run faster, than the person you are with. Let’s say there’s an issue with id-cards or electronic signatures in general. The attackers could wait several years and attack voting risking, that all electronic votes are annulled as their attacks are uncovered. Or they could just wreak complete havoc on the entire society right now. The electronic signatures are so ubiquitous here you can literally do anything. Sign a document as the President, take out loans on behalf of everyone, get everyone divorced (or married), take all their money etc. why would you wait?

They are in a vicious circle there’s no good way out of. Building and operating their services is not operationally profitable. Therefore they need more funding. The only way to get more funding is to get a wider adoption and show new features. The only way to do that is to keep spending money.

Not being profitable in itself is not a problem or has not been historically. With software, there’s a bunch of investment into making the service but the variable cost of providing it is negligible. It’s ok not to be profitable because the growth of userbase will pay you back. Capex isn’t a thing, you can just rent somebody else’s from the cloud. This is how the world used to work pretty much for a quarter century, ever since the dot-com bubble. But OpenAI is not like it. Their variable cost is huge as is the capex to enable being able to even make that cost happen. It’s not that they are not profitable, it’s that their financial profile does not fit into anything this generation of decisionmakers is used to.

I’ve never drank. Half a glass of what I was told was a really good champagne at my wedding 27 years ago. That’s it. It was a conscious decision at first, based on alcohol problems of a family member, then a habit. I’m pretty sure my non-existent social life/skills are at least partly attributable to this. Without something to relax my continuously racing brain (never diagnosed, won’t bother now), I’m probably pretty tedious company and there’s a limited time in your life to build a social network. I don’t complain, as there’s a ton of time that’s freed up when you simply don’t go anywhere and I’ve made good use of it. But I do miss the long glorious conversations straddling both the intellectual and the profane between a group of smart relaxed individuals who, for the moment, have forgotten the world and it’s problems.

It’s a question, because a lot of money is riding on people developing a mental model of a llm being something else than a computer program.

The search function has also become patently unusable. I was looking for a specific bike widget. No luck, doesn’t exist according to Google search and Gemini. Fine. Unlikely, but fine. And then, there it was, a week later. I just stumbled upon this thing randomly. The product had the right keywords and everything.

Very well done, OP! Lots of reflection, methodical approach, practical design. I’ve also gotten back to making physical things recently. I did not recognize the feeling at first, but soon realized I’m looking forward to getting back to building that shed. Have not felt that for a while about coding or other work. One advice I have is not to skimp on tools. I had some old wired tools that were OK. On a whim invested into a good Makita impact drill and it’s just another level. Got a circular saw as well and few other bits: having proper tools makes this so much more enjoyable. It doesn’t have to make financial sense, DIY is mostly about enjoyment. So why don’t I make sure I enjoy it?

Same. It’s a tiring exercise keeping up with what is going on in Giant Steps. I marvel at the intricacies I can’t mostly even begin to understand but it doesn’t speak to me musically the way eg Watercolors or Kind of Blue or Take Five does.

Lemmy. The man, the myth, the legend. Like with Ozzy, there was no stage persona. I’ve met him on a street the day after the gig looking exactly as he did on stage the day before. If you listen to stories about him, you’ll find the obligatory drinking and excess. Sure. But immediately after you’ll find a lonely man who deeply cared about his team, who was probably too smart for the business and who added warmth to even simple human interaction. Ozzy was the same and it’s no coincidence his most vulnerable songs were written by Lemmy. Having met both, I miss them in this world.

Thinking and writing are complementary processes. If you write things where you don’t have to think, let the AI do the typing. No harm done. But any writing, where you actively think while you type not about what the next word should be but about the entire topic, so it yourself. Otherwise you’re not thinking and that’s pretty much the only value-add that’s left.

I have removed a person from a project. Doing their share of work, paying them exactly as agreed. Because they decided to have Claude write 10 pages of feedback on about 5 pages of content I created even after I asked them for clarification. If anyone uses AI in communication, fine. If people think an LLM can express their thoughts better, than they can, ok. No problem. But if I wanted to talk to computers all day long I can already do that. I’ve been doing it close to 40 years now. The reason I work in a team is so I get to connect to actual human beings. Live people, warts and all. If I can’t get that, why would I bother? Again: I have plenty to do talking to computers.

The author does not emphasize a significant reason why this is. In Russian Orthodox belief, there is no purgatory. This is one of the key reasons they consider the catholic heretical. In their belief system, the suffering, that’s supposed to clean us of our sin, happens while we live and not afterwards. Suffering in itself is good for the soul. This is why there’s a saying in Russian about violent husbands: “beats, that means he loves”. Not beating my wife was considered amazingly progressive by my Russian relatives 20 years ago. And this is also why they don’t care about their fallen soldiers or treat GULAG sites like the sites of former concentration camps. Suffering is good for you.

This assumes the current demand will continue. Which is unlikely given the order of magnitude difference between capex by the Big Folks and the revenue generated by that capex. Given they now need to fund this spend from money markets, they are likely to be told off pretty soon. Feds go “ooh” and Google goes “aah”.

What’s the rationale of all this? I mean countries have gotten (mostly) rid of black books without such systems. Maintaining such a solution must be horribly expensive and what if it goes down?

Quadlock makes a special dampening mount for this reason.

What makes large things slow is not the individual pieces of code but the way the architecture dynamically reacts to changes in the load. Sync vs async calls, buffers, parallel vs sequential processing. This includes optimizations made: do we want to wait until the call returns or do we proceed anyway and deal with the negative response later? Most developers can’t coherently (ie, with math, graphs and figures) explain, how a connection pool prevents undesirable consequences of brief spikes in incoming requests. And I’m pretty sure AI can’t either.

Harassment is not permissible. But if he had been a black researcher in, say, microbiology, it wouldn’t have materialized to that extent. His research tried to shame a very large chunk of the society for being racist. And it turns out people react in unpredictable and unpleasant ways to accusations towards their core identity that are based on lies.

There’s a difference between normalizing harassment and understanding where it comes from. I’d also like to be more forgiving about people talking bs about me. But I’m not.

Also, imagine being a black person trying to build a career in academia after such a story. The vitriol _they_ will receive, the scrutiny of _their_ work.

Everyone deserves compassion. I feel sorry for him. But the fact remains, that he deliberately used lies and deception to gain a high social position in a very politically derisive context. It couldn’t have come as a surprise, that every single racist out there would jump at the opportunity to pull him down should there ever be doubt about his academic work. Plus the people, who really care about academic integrity. Plus a whole lot of folks who simply do not like people blatantly lying for massive personal gains. The vitriol was entirely predictable. The internet is not known for being forgiving. He very deliberately and knowingly took a huge and, given the absurdity of some of his claims, unfounded risk the materialization of which he could not handle. I do feel sorry for him to have started on this path. What state of mind would push a person to take such risk? He must have been a tortured man long before it all exploded into the open.

Maybe it’s niche, but if you do endurance sports like triathlon, you can easily become a power user. I think I did. You constantly look at your watch during workouts. It does navigation on trails, it measures your heart rate, tracks workouts, keeps pace during races. The metrics it gathers about hrv, sleep, rest hr etc are vital to adjusting the routine. It acts as a data recorder for multiple sensors: hr strap does a million running metrics, a separate strap does hydration, on a bike there’s a power meter. It supports my routine alerting me about the fast, timing my morning porrige (14 min). It’s my payment method on long rides when I need water and could play audio if I wanted to. I care about advanced features like multi-band gps and handover with my bike computer. It’s a vital tool, I need it for so many things. But vast majority of smartwatches do not cater for this crowd at all. Even the Apple Ultra feels like a toy. Garmin upper end stuff is where it is.

It’s not at all immediately obvious as to why the bicycle stands upright or turns. The physics is non-trivial and, based on the drawings, offsetting the center of the wheel from the axis of thw wheel being turned was not a thing. But it is this feature, that allows the modern bike to maneuver. Counter-steering is very unintuitive. If you lack the underlying physics or, in fact, personal experience, the bicycle sounds utterly mad. It doesn’t make sense. Which is why people did not work on it.