@jacqueline I see that a lot with robotics and iot. We literally have a robotics revolution now, because someone figured out that having your room vacuumed badly by a randomly moving noisy thing is something some people could find useful. It now spilled to lawn mowers and warehouse logistics, just like the steam engine spilled from pumping water in a coal mine to Jacquard looms once it became more polished. Finding that first use case where the thing is useful despite being very bad is hard.
@jacqueline It can become better if it finds its niche, but it's a rare thing, and it's hard to tell up front if it will happen or not.
Steam engines were bad initially. Electricity was very bad at first. Radio was ridiculously bad. Semiconductors were bad. Computers were embarrassingly bad and fiddly (no change there so far). Photography was pathetic.
On the other hand there are so many technologies that never found their niche. We don't eat radioactive chocolate anymore, or use lead makeup.
Remember that automation is not about removing effort, it's all about packaging it in convenient ways so that it can be stored and moved around. Sometimes you get some savings due to bunching up of similar things or economy of scale, but most of the time the perceived savings are simply because the real effort is hidden away or deferred. Whether it's Jacquard looms, language models, github actions, self-driving cars, or robot food delivery, someone is doing (or has done, or will do) the work.
The only case where you save effort is when you carefully define effort so as to exclude the entities that actually do the work. Like, say, plowing a field using animals "saves" effort for people, because now animals do the work.
Steam engines (and modern combustion engines) "saved" a lot of effort by using up timber and coal that accumulated over millions of years, and damaging our atmosphere in the process. You can't keep "saving" effort this way indefinitely.
@dcz That's not quite what I had in mind – of course thermodynamics ensures you always have to spend a minimal amount of energy, but not all effort can be measured by the energy spent.
On the other hand, going by what the given society currently considers to be agents leads to ridiculous results, for example when slave labor is not considered to be real effort.
This is why I think it's more useful to consider where/when that effort is being moved, rather than pretending it disappears.
@dcz I don't think you can call it a law, and I don't think there is a "unit of benefit". We don't have a working theory of value, really. The ones proposed by capitalism (based on capital) and communism (based on labor) evidently don't work, especially when you try to apply them to cultures outside of the western society.
So any kind of an equivalence law would be a huge stretch.
On the other hand, automation always has costs. They might not be equal to the work saved, but they are not zero.
@dcz Oh, right, that's another thing automation does: it commodifies the effort, making it fungible. Because it's stored and can be easily moved around, you can now use it for different things, not necessarily the thing it was originally meant for.
So you can use the same electricity for, say, operating a life support system, or for generating NFTs.
The exchange rate has nothing to do with actual value of the work done. One watt-hour of extending someone's life is not the same as a bored ape.
@dcz If you read my original message, you will note that I mentioned that you do get to save effort sometimes. But that's usually less than it seems. Solar panels are environmentally expensive to produce (though they are getting cheaper due to economy of scale), require effort to install, maintain and repair, so much so that they wouldn't probably be worth it just for the washing machine – they become worth it due to the bunching up effect because electricity is so versatile.
If you wanted to explain to a 10 year old how a button or switch can change between two things, like AM or FM on a radio, but didn't want to use a radio because it may be outdated... what example might you use?
How is this considered remotely acceptable in 2024?
$ deno # same with node. It's really V8
> let r = /.*foo.*bar/
undefined
> let s = "no".repeat(100000)
undefined
> console.log(performance.now()); r.test(s); console.log(performance.now())
22362.667416
93954.182416
@renniezen@cscheid warning: what you get in computers these days only looks as a regex, and in reality has exponential complexity instead of linear, because perl
@cscheid@renniezen well yes, that's what I said, even if you choose to intentionally pick an interpretation that doesn't make sense instead -- what you see in modern computing called "regex" has nothing to do with finite automata these days