For anyone unsure: Jevon’s Paradox is that when there’s more of a resource to consume, humans will consume more resource rather than make the gains to use the resource better.
Case in point: AI models could be written to be more efficient in token use (see DeepSeek), but instead AI companies just buy up all the GPUs and shove more compute in.
For the expansive bloat - same goes for phones. Our phones are orders of magnitude better than what they were 10 years ago, and now it’s loaded with bloat because the manufacturer thinks “Well, there’s more computer and memory. Let’s shove more bloat in there!”
Case in point: AI models could be written to be more efficient in token use
They are being written to be more efficient in inference, but the gains are being offset by trying to wring more capabilities out of the models by ballooning token use.
Which is indeed a form of Jevon’s paradox
Costs have been dropping by a factor of 3 per year, but token use increased 40x over the same period. So while the efficiency is contributing a bit to the use, the use is exploding even faster.
Jevon’s Paradox is that when there’s more of a resource to consume, humans will consume more resource rather than make the gains to use the resource better.
More specifically, it’s when an improvement in efficiency cause the underlying resource to be used more, because the efficiency reduces cost and then using that resource becomes even more economically attractive.
So when factories got more efficient at using coal in the 19th century, England saw a huge increase in coal demand, despite using less coal for any given task.
Also Eli Whitney inventing the cotton gin to make extracting cotton less of a tedious and backbreaking process, which lead to a massive expansion in slavery plantations in the American South due to the increased output and profitability of the crop.
I always felt American car companies were a really good example of that back in the 60s-70s when enormously long vehicles with giant engines were the order of the day. Why not bigger? Why not stronger? It also acted as a symbol of American strength, which was being measured by raw power just like today lol.
This also reminds me of the way video game programmers in the late 70s/early 80s had such tight limitations to work within that you had to get creative if you wanted to make something stand out. Some very interesting stories from that era.
I also love to think about the tricks the programmer of Prince of Persia had employed to get the “shadow prince” to work…
The tech debt problem will keep getting worse as product teams keep promising more in less time. Keep making developers move faster. I’m sure nothing bad will come of it.
Capitalism truly ruins everything good and pure. I used to love writing clean code and now it’s just “prompt this AI to spit out sloppy code that mostly works so you can focus on what really matters… meetings!”
so you can focus on what really matters…
meetings!collecting unemployment!What really matters isn’t meetings, it’s profits.
The modern web is an insult to the idea of efficiency at practically every level.
You cannot convince me that isolation and sandboxing requires a fat 4Gb slice of RAM for a measly 4 tabs.
Can’t wait for the new evidence that Epstein is behind that too.
Actshually it’s bandwidth censorship if you make something too heavy to be used then it won’t get used. It is one of the things China is doing to separate their internet from the rest of the worlds, by having an internet so blazingly fast it is unbearable to goto the world wide web.
So yesh, the epstien class are making the news too slow for typical users to access. /maybe some sarcasm maybe not I’m not sure yet
EDIT: I have decided I was not being sarcastic. https://ioda.inetintel.cc.gatech.edu/reports/shining-a-light-on-the-slowdown-ioda-to-track-internet-bandwidth-throttling/
Episodes of network throttling have been reported in countries like Russia, Iran, Egypt, and Zimbabwe, and many more, especially during politically sensitive periods such as elections and protests. In some cases, entire regions such as Iran’s Khuzestan province have experienced indiscriminate throttling, regardless of the protocol or specific services in use. Throttling is particularly effective and appealing to authoritarian governments for several reasons: Throttling is simple to implement, difficult to detect or attribute and hard to circumvent.```
Thought leaders spent the last couple of decades propaganding that features-per-week is the only metric to optimize, and that if your software has any bit of efficiency or quality in it that’s a clear indicator for a lost opportunity to sacrifice it on the alter of code churning.
The result is not “amazing”. I’d be more amazed had it turned out differently.
Fucking “features”. Can’t software just be finished? I bought App. App does exactly what I need it to do. Leave. It. Alone.
No, never! Tech corps (both devs and app stores) brainwashed people into thinking “no updates = bad”.
Recently, I have seen people complain about lack of updates for: OS for a handheld emulation device (not the emulator, the OS, which does not have any glaring issues), and Gemini protocol browser (gemini protocol is simple and has not changed since 2019 or so).
Maybe these people don’t use the calculator app because arithmetic was not updated in a few thousand years.
A big part of this issue is mobile OS APIs. You can’t just finish an android app and be done. It gets bit rot so fast. You get maybe 1-2 years with no updates before “this app was built for an older version of android” then “this app is not compatible with your device”.
It’s kind of funny how eagerly we programmers criticize “premature optimization”, when often optimization is not premature at all but truly necessary. A related problem is that programmers often have top-of-the-line gear, so code that works acceptably well on their equipment is hideously slow when running on normal people’s machines. When I was managing my team, I would encourage people to develop on out-of-date devices (or at least test their code out on them once in a while).
Premature optimisation often makes things slower rather than faster. E.g. if something’s written to have the theoretical optimal Big O complexity class, that might only break even around a million elements, and be significantly slower for a hundred elements where everything fits in L1 and the simplest implemention possible is fine. If you don’t know the kind of situations the implementation will be used in yet, you can’t know whether the optimisation is really an optimisation. If it’s only used a few times on a few elements, then it doesn’t matter either way, but if it’s used loads but only ever on a small dataset, it can make things much worse.
Also, it’s common that the things that end up being slow in software are things the developer didn’t expect to be slow (otherwise they’d have been careful to avoid them). Premature optimisation will only ever affect the things a developer expects to be slow.
Optomisation often has a cost, weather it’s code complexity, maintenance or even just salary. So it has to be worth it, and there are many areas where it isn’t enough unfortunately.
And that lazy mentality just passes the cost to the consumer.
Your spelling is terrible
Oops, forgot the AI step
I’m pretty sure the “unused RAM is wasted RAM” thing has caused its share of damage from shit developers who took it to mean use memory with reckless abandon.
Would be nice if I could force programs to use more ram though. I actually have 100GB of DDR4 my desktop. I bought it over a year ago when DDR4 was unloved and cheap. But I have tried to force programs to not be offloading as much. Like Firefox, I hate that I have the ram but it’s still unloading webpages in the background and won’t use more than 6GB ever.
I actually have 100GB of DDR4
They’ve got RAM! Get’em!
Will disabling the swap file fix that?
If not, just mount your swap file in RAM lmao
With 32 and 64 GB systems I’ve never run out of RAM, so the RAM isn’t the issue at all.
Optimization just sucks.
My PC is 15 times faster than the one I had 10 years ago. It’s the same old PC but I got rid of Windows.
That’s… not really true, and not what that link shows. Those latency tests still show then-modern devices topping the list. They’re arguing that some then-modern low end devices have more button-to-screen latency than older hardware (which they would, given he’s comparing to single-threaded, single-tasking bare metal stuff from the 80s spitting signals out to a CRT to laptops with integrated graphics). And they’re saying that at the time (I presume the post dates from 2017, when the testing ends), this wasn’t well understood because people were benching the hardware and not the end to end latency factoring the I/O… which was kinda true then but absolutely not anymore.
I’d get in the weeds about how much or little sense it makes to compare an apple 2 drawing text on a CRT to typing on a powershell/Linux terminal window inside a desktop environment, but that’d be kind of unfair. Ten years ago this wasn’t a terrible observation to make with the limited tools the guy had available, and this sort of post made it popular to think about latency and made manufacturers on controllers, monitors and GPUs focus on it more.
What it does not show, though, is that an apple 2 was faster than a modern gaming PC by any metric. Not in 2017, and sure as hell not in 2026, when 240Hz monitors are popular, 120Hz TVs are industry-standard, VRR is widely supported and keyboards, controllers, monitors and GPU manufacturers are obsessed with latency measurements. It’s not just fallacious, it’s wrong.
I still remember playing StarCraft 2 shortly after release on a 300$ laptop and it running perfectly well on medium settings.
Looked amazing. Felt incredibly responsive. Polished. Optimized.
Nowadays it’s RTX this, framegen that, need SSD or loading times are abysmal, oh and don’t forget that you need 40gb of storage and 32gb of ram for a 3 hour long walking simulator, how about you optimize your goddamn game instead? Don’t even get me started on price tags for these things.
Software and game development is definitely a spectrum though, but holy shit is the ratio of sloppy releases so disproportionate that it’s hard to see it at times.
Then factorio dev blog comes in and spend months optimizing the tok of one broken gear in the conveyor belt to slightly improve efficiency.
Tbf, there’s saves there that efficiency increase means a lot
It’s the only game I have that will actually recover from when it hangs and freezes and then go back to working fine.
StarCraft 2 was released in 2007, and a quick search indicates the most common screen resolution was 1024x768 that year. That feels about right, anyway. A bit under a million pixels to render.
A modern 4K monitor has a bit over eight million pixels, slightly more than ten times as much. So you’d expect the textures and models to be about ten times the size. But modern games don’t just have ‘colour textures’, they’re likely to have specular, normal and parallax ones too, so that’s another three times. The voice acting isn’t likely to be in a single language any more either, so there’ll be several copies of all the sound files.
A clean Starcraft 2 install is a bit over 20 GB. ‘Biggest’ game I have is Baldur’s Gate 3, which is about 140 GB, so really just about seven times as big. That’s quite good, considering how much game that is!
I do agree with you. I can’t think of a single useful feature that’s been added to eg. MS Office since Office 97, say, and that version is so tiny and fast compared to the modern abomination. (In fact, in a lot of ways it’s worse - has had some functionality removed and not replaced.) And modern AAA games do focus too much on shiny and not enough on gameplay, but the fact that they take a lot more resources is more to do with our computers being expected to do a lot more.
Why are you comparing the most common screen resolution in 2007 to a 4k monitor today? 4k isn’t the most common today. This isn’t a fair comparison.
1080p is still the most common, though is 1440p is catching up very fast
BTW the demand for bigger screens and bigger resolutions is something I don’t easily understand. I notice some difference between 1366x768 and 1920x1080 on a desktop, but the difference from further increase is of so little use for me I’d classify it as a form of bloat. If anything, I now habitually switch to downloading 480p and 720p instead of higher definition by default because it saves me traffic and battery power, and fits much more on a single disk easy to back up.
I hate that our expectations have been lowered.
2016: “oh, that app crashed?? Pick a different one!”
2026: “oh, that app crashed again? They all crash, just start it again and cross your toes.”
I’m starting to develop a conspiracy theory that MS is trying to make the desktop experience so terrible that everyone switches to mobile devices, such that they can be more easily spied on.
That would be incredibly ironic given that they completely fucking gave up on mobile devices when the iPhone came out.
Windows Phone was around in mid-2010s, at least 7 years after iPhone release. But it was not hyped enough: companies did not care to develop apps for it, customers didn’t want a smartphone without X Y Z apps (same argument i see now about mobile linux or even custom ROMs). The phones had nice and fast UI though, and some had very good cameras.
Windows Phone was great. I’d done Windows Mobile since 2005 and it was nice to be able to continue developing with C#/.NET and Visual Studio (back when it was still good) in a more modern OS. One thing that really spoiled me permanently was being able to compile, build and deploy the app I was working on to my test device effectively instantaneously – like, by the time I’d moved my hand over to the device, the app was already up and running. Then I switched to iOS where the same process could take minutes, also Blackberry where it might take half an hour or never happen at all.
Funny thing: RIM was going around circa 2010/2011 offering companies cash bounties of $10K to $20K to develop apps for Blackberry, since they were dying a rapid death but were still flush with cash. Nobody that I know of took them up on the offers. I tried to get my company to make a Windows Phone version of our software but I was laughed at (and deservedly so).
On Linux it really is noticeable
Two possible and opposite interpretations of your comment:
- Modern Linux feels exactly as responsive or worse on modern hardware than old Linux used to feel on old hardware.
- Linux feels much more responsive and fast on modern hardware than it does on old hardware, unlinke other OSes.
I dislike a lot the framing of this.
Yes, the average software runs much less efficient. But is efficiency what the user want? No. It is not.
How many people will tell you that they stick to windows instead of switching to linux because linux is all terminal? And terminal is quicker, more efficient for most things. But the user wants a gui.
And if we compare modern gui to old gui… I don’t think modern us 15x worse.
There isn’t anything fundamentally slower about using a GUI vs just text in a console. There’s more to draw but it scales linearly. The drawing things on the screen part isn’t the slow bit for slow programs. Well, it can be if it’s coded inefficiently, but there are plenty of programs with GUIs that are snappy… Like games, which generally draw even more complex things than your average GUI app.
Slow apps are more likely because of an inefficient framework (like running in a web browser with heavy reliance on scripts rather than native code), inefficient algorithms that scale poorly, poor resource use, bad organization that results in doing the same operation more times than necessary, etc.
The terminal is quicker. Not because of the image is drawn more quickly but because it is more efficient to do anything.
Technically true, but there’s a threshold on responsiveness. If both user interfaces respond in milliseconds, it doesn’t matter if one is more efficient
It does because it highlights that instead of being excited to “have to use the terminal” as it is more “efficient” but instead they prefer the “slower” prettier gui. The user want the stupid animations and the flashy nonsense. The user doesn’t want quick software. They want pretty software.
Ah, you’re one of the Linux gatekeepers. You’re not worth bothering with anymore. Tah tah
How am I gatekeeping?
I am not telling anyone to use Linux in anyway or to not use it in anyway. I am just pointing out that the average user wants a pretty/convenient gui and not the most efficient tool. That isn’t bad. I don’t want to eat some weird mixture of nutrients because it is optimal, I want to eat food that I enjoy eating.
I am calling out the weird focus on efficiency of software when the average user wants a good user experience. The user’s desire is not good nor bad, they just highlights that focusing on criticising efficiency of software is a strange thing to do, if the customer desires something else. It is like complaining that grindr is lacking heterosexual people.
When you become one with the penguin, though … then you can begin to feel how much faster modern hardware is.
Hell, I’ve got a 2016 budget-model chromebook that still feels quick and snappy that way.
