belZaah

We had a lady down here in Estonia working for police. She used their extensive databases for non-work related reasons and got, very publicly, fired. She decided to appeal and lost. Through all levels of court up to the EU one. So for years, there would be a newspaper headline “remember that person who invaded your privacy? Still guilty”. Could not wish for a more impactful public execution.

It’s a sign of a badly designed system. In a well-designed one (have seen one, have helped to improve several) the drivers license database only has a personal identifier and the class of license identifier. Which is private information for sure but it would hardly be an end of the world if the feds learned I have the right to operate a motorcycle.

Perversely, this is the consequence of misguided privacy initiatives. Government agencies are discouraged or even forbidden from sharing data. Nobody has a reliable identifier. As a consequence, every single database out there will contain your name, dob, address and probably a photo. Oh, and because there’s no identity, they need to store a bunch more data to make sure you are not a fraudster. Also, in absence of a sensible electronic identity (because there’s no identity at all in the system), all of this data is used as a shared secret. Which makes it immediately commercially useful for identity theft, once obtained. And, of course, any of these giant piles of personal data can and will be breached in some way or another by both local and remote state actors.

It’s all because “identifiers are bad” and “no sharing information”. Are you guys satisfied with the privacy situation yet?

There’s one more aspect to this. Look at Claude self-reported availability figures. It’s two nines at best, not enough to build something serious on top of. Historically, adding another 9 is an order of magnitude cost shift. Even in best case scenario, where a LLM scales much differently, you are looking at significant cost increases just to achieve something people would be willing to have their business depend on.

Radars do this better. They can warn you from much further away and do not require for you to keep an eye out - there’s audible cues.

You just need to go north. Most of Finland requires massive amounts of heating for most of the year for industrial and residential purposes. If the math doesn’t add up there, I really can’t see how math for orbital data centers would add up.

Why would you want to earn a billion dollars? You must have the maturity to handle the money. Hopefully you’ll mature sufficiently through the process of making that money but if the money’s fast, that might not happen. I’ve seen people get 9-figure rich overnight. Didn’t change them at all, they are still the wonderful folks they were. But I’ve also seen people drink themselves to death as it turns out a few mill in the bank does not answer the question “wtf do I do when I don’t have to do anything?”

Thus is the crime of the communist Russia: forcing millions into hard labor to die for progress yet squandering innovation for ideological reasons. But the same mechanism is there in, say, Microsoft. To get the attention of leadership, your idea must have 9 zeros at the very least. If it doesn’t, you either leave M$ or stay there and abandon your idea. But a 7-zero idea is a pretty expensive one to be abandoned.

True. But in the physical world, ideas (or memes, if you will) are bound to people and can only survive, if the carrier survives. The idea of curing headaches with striking your head with a hammer does not spread, because the carrier dies. On the interweb, that connection is no longer there. An idea is independent and can spread without direct human involvement. We are like a tundra species finding itself in a rainforest: completely unprepared, lacking both immunity and the ability to keep up with the environment as it changes.

Exactly. So is that level of obvious hygiene where the bar is or is it somewhere else. What ticks me off is the audacity of blanket claims without an attempt to even remotely state why it’s said this is a list of successful patterns and what does success mean. We’re just supposed to eat it up, because, you know, Claude.

How very interesting. In an industry, where things shift around in months if not weeks, there’s been not only enough time for clear patterns to emerge but also these patterns have proven successful on large codebases. What’s the success criteria? Didn’t delete production database? Team velocity has increased? Codebase TTL has increased? Operations guys are happier?

There used to be such a thing as profit. A return on investment. If your exit strategy is to get sold to Google, focusing on revenue is a perfectly fine strategy. If you _are_ the Google, however, the money poured in should eventually be made back. We seem to have forgotten that. The current level of commercialization just means the US is burning investments faster, than anyone else. Eventually this might change and the bet might pay off. But every minute this goes on, the expected payoff must be larger to pay for the loss made this minute as well as interest for the previous minutes. I’m not entirely sure this is what “winning” looks like. Tic-toc.

It’s a general principle. Complexity breeds complexity via a multitude of mechanisms thus growing exponentially. As complexity is also related to cost, this means, that there are two limits approached exponentially: financial (we can no longer afford complexity) and cognitive (ee can no longer understand). In an ideal world, financial barrier arrives first, as it is necessary to understand something to make it constructively simple. If it doesn’t, the only solution is destructive simplification by simply breaking the system into pieces forcefully. This is what Musk did to X and tried to do to the US Government.

Unlikely. There’s no change in operating profit per employee trends for major software companies like Alphabet since GenAI became a thing. But MS employees are now making 3 times more profit, than they were before Nadella took over. Clearly leadership can make a difference but there is no visible impact after several years of the technology being available. I can’t imagine a technology, that shows no economic impact at all while we figure it out. There ought to be _something_. Yes, big companies have inertia, but Nadella showed clear results in a year.

My kids walk/bike to school, take public transport and are pretty free-range. But that’s absolutely nothing compared to my childhood. I was born in 1975. We dug up literal explosives from WWII and made our own small bombs (some matches, two large bolts…). Access to all sorts of chemicals was easy and we would set things on fire or just mix stuff to see what happens. Playing around on construction sites. Taking someones boat out to go fishing. Making bows and arrows, that would go straight through plywood. All _that_ is definitely gone. Some of it for good - kids loosing their fingers or worse was a common occurrence. So there definitely is a trend in case of me and my friends.

Soviets did not have two things the West did. Concern for quality and market forces to direct development focus. This means Soviet stuff varies amazingly between specimen and can sometimes be over-engineered in particular ways. Soviet optics had a specific visual style, but everyone ditched them as soon as alternatives became available as hunting for the ones not made on a Monday was just too tedious.

Terrible in which way? Did not use counterpoint sufficiently elegantly? It’s punk, mate. Try to do a set downpicking like Johnny.

An excellent analogy. Everyone is an expert in taking the photo. But this does not make them a photographer. Even that expert claim is actually not fully true, the phone camera is woefully inadequate in many ways. But the main difference between a photographer and a layman like myself is the ability to produce output strongly linked with clear artistic intent.

Writing code is not the hard part and never has been. The hard part is having a clear understanding of how to solve a specific complex problem and being able to express that intent in code. Getting a decently exposed image was never the hard part.

Finally, there’s no scaling issues with cameras. You just make them better until it stops making economic sense. This is not true with code. To make llms better, good human-made code is needed for training. Better llms lead to less human-made code being available. This means there’s not an exponential growth in quality but a S-curve with a balance point. I’d say we are already there: innovation is shifting from the models to the ways of harnessing the models.

So it’s like code for execution via a llm or an interpreter? Have we invented source control yet?

I used to manage NT-based infra back in the day, have been on a mac for 15 years now because of stuff like this. A few years ago I bought a Windows box for my daughter. Out of the box the clock was wrong and it would just hang on auto-update. No message, no logs anywhere, just hangs. A few years later the son comes of age and gets his own box. And it’s the same story, no automatic adjustment of the clock. I’m running a bog standard unifi network leading to fiber, nothing complicated, everything else works including all the windows laptops of my wife. But a basic standards-based library-supported Windows function.

The weird thing is the way it’s trash. It breaks weird things no dev should ever have to touch. At one point Excel left horizontal lines on screen, when scrolling. Bullets and numbering just straight up refuses to restart numbering. It _worked_ why did you break it? Who gained what out of you breaking it?

That’s a broken analogy. An intern and a llm have completely different failure modes. An intern has some understanding of their limits and the llm just doesn’t. The thing, that looks remarkably human, will make mistakes in ways no human would. That’s where the danger lies: we see the human-like thing be better at things difficult for humans and assume them to be better across the board. That is not the case.

I don’t think the objections are not necessarily in terms of lack of productivity although my personal experience is not that of massive productivity increases. The fact that you are producing code much faster is likely just to push the bottleneck somewhere else. Software value cycles are long and complicated. What if you run into an issue in 5 years the LLM fails to diagnose or fix due to complex system interactions? How often would that happen? Would it be feasible to just generate the whole thing anew matching functionality precisely? Are you making the right architecture choices from the perspective of what the preferred modus operandi of an llm is in 5 years? We don’t know. The more experienced folks tend to be conservative as they have experienced how badly things can age. Maybe this time it’ll be different?

I’m not sure you are familiar with the way some traditional communities treat, say, single mothers.

The system is, that if you make unfortunate choices (such as moving away from your support network and carrying your life savings where you can forget them) and do not have anything (such as skills or low morals) to compensate, life is going to be hard. That has always been the case, that will always be the case. What the consequential decisions are, differs. But the basic premise of “f around and find out” holds.

Hats off for going to the Primary Source!

I believe in coding primarily as a means to an end

Yes. Absolutely. To what end, though? Is your end deterministic like a cryptographic protocol or loose like pagination of a web page? Is your end feature delivery or 30 years of rock solid service delivery at minimal cost?

AI is a dangerous tool. It exposes fundamental questions by automating away the mundane. We have had the luxury of not thinking deep and hard about intent and value creation/capture and system architecture. AI is putting us face to face with our ineptitude: maybe it wasn’t the tech stack or the programmers or the whatnots? Maybe the idea was shait, maybe I had no understanding of the value added of my product? Maybe …?

You get the best gear - musical instrument, bicycle, camera, etc - the pros have and still the results are not great. Gotta ask why. We are experiencing this at literally industrial scale.

It’s called emergent behavior. We understand how an llm works, but do not have even a theory about how the behavior emerges from among the math. We understand ants pretty well, but how exactly does anthill behavior come from ant behavior? It’s a tricky problem in system engineering where predicting emergent behavior (such as emergencies) would be lovely.

Not just that. An LLM is an universal middleman for anything creative. Books, code, music. Everything written, spoken or performed can be accessed as digested and mixed by the LLM. It breaks all content-relates business models by inserting a middleman, that seeks to capture as much value as possible while actively arguing the source of mediated content is not entitled to anything. This is not sustainable, as the content source will loose motivation, including non-monetary motivation. This in turn will stop the middleman being able to capture value created by them. But the ability to combine existing content without anything new being added is likely to be too little to offset the enormous costs involved. Whatever the outcome for the AI-bros, human content generation will decrease massively for a long while. Which has a massive detrimental effect on culture globally.

What ticks me off in this is the statement, that a certain body shape is “unattainable for most”. I’m pretty sure the author does not have the data to back this up. Difficult? Yes. Requiring commitment? Absolutely. Unattainable? No. I really don’t care what body shape anyone is comfortable with. But as someone, who has struggled hard all his life not to be obese, I find it irresponsible to outright declare something that’s absolutely doable by anyone as “unattainable”. Being able to attain it might be someone’s only hope and it’s just wrong to take it away.

There’s an old saying: if civil engineers built houses the way software people build software, the first woodpecker to appear would destroy the civilization. With Tesla, we build cars. That, as told in court documents, absolutely should continue accelerating while in cruise control despite the driver pressing the brakes. There’s a century of institutional knowledge on system safety built into most cars. And (looking at you, Pinto), the carmakers are not even especially good at it. As a former software engineer, I’d rather rely on some actual engineers rather than a bunch of tech bros led by a deranged sociopath.

The premise here seems to be, that the code is where the value is created. That’s maybe true for some apps but not every one. Apps can facilitate interactions (eg Training Peaks), store data long term (eg MyFitnessPal) or act as a part of a complex network of complementary goods (eg Garmin Connect). The last part is especially important: the ability to create something does not necessarily mean the ability to gain value from it.