The Coming War on General Computation (2011)(en.wikisource.org) |
The Coming War on General Computation (2011)(en.wikisource.org) |
Another thing is the general impoverishment and decline of productive power that can be papered over by centralizing the output of that production to hide the general decline. I don't think anybody is buying that we are richer than 20 years ago anymore.
And .. some of us just spent the same budget as we spent back then on PC’s, on Nvidia DGX gear …
It isn't the biggest technical concern and at least they have principles, but I expect something that looks like Monero + easy to mine will eventually appear and turn out to have an advantage in the real world. Although it is hard to predict because it isn't completely clear what (if any) impact the algorithm difficulty has in practice. But if there are big datacentre mining a cryptocurrency with dedicated hardware then that means there are people with skin in the game who want it to succeed. Seems like a better equilibrium than whatever the Monero people are aiming for.
Profoundly sad.
Keep on fighting.
Is it because you're using an OS like Windows or Mac that are very closed and don't encourage hacking? There's an easy solution to that.
(Note that I'm not taking about things like the age verification laws, which haven't had a meaningful effect anywhere yet.)
We're not making a computer that runs only the "appliance" app; we're making a computer that can run every program, but which uses some combination of rootkits, spyware, and code-signing to prevent the user from knowing which processes are running, from installing her own software, and from terminating processes that she doesn't want.
In other words, an appliance is not a stripped-down computer -- it is a fully functional computer with spyware on it out of the box.
(audience applauds loudly) Thanks."
Some excellent observations by Cory Doctorow!
Related:
https://en.wikipedia.org/wiki/Enshittification
"A Day in the Life of an Ensh*ttificator":
It was bad enough when mobile phones were not forced to be dealt with as "general purpose computers", and instead were allowed to remain proprietary hardware/software combinations controlled by the tech oligarchy.
But now that RAM is not financially viable to deliver to consumers due to the obscene appetite of the LLM crowd, and fabs are developing custom silicon for the dumbest things of all "training LLMs", we are seeing at least the beginning of what it mean for "personal computers" to become emaciated to the point where chip companies can no longer justify making them for such measly profit and how difficult they will be to sell at such high prices if they incorporate high cost components.
We arent fully there yet but we are heading in that direction. The laptop Im typing this on cost 1600 dollars 8 months ago and now retails for 4 thousand dollars. In 6 more months it will either not be available or retail for 8 thousand.
Given that LLMs are convincing less experienced coders that code is a waste of time to write, the generational/knowledge gap that will create over time in the industry, software engineering could in essence become a niche reserved only for elites (if it will even be allowed) and end user software/hardware knowledge will be discouraged and considered threatening to the establishment.
Although there is a large knowledge base on the Internet (right now) the Internet is now under relentless and growing assault from AI bot farms referred to as agents, which are simultaneously making it less and less practical to maintain web sites and internet infrastructure due to increased costs, which could create a "lights out" situation for online documentation as the internet is choked off and dies in favor of MCP clients and servers which hide functionality behind obscure interfaces and website content ceases to be economically sustainable.
This will usher in a human "dark age", where the cost of information will be zero but the truth of anything if it existed previously will become unavailable to all but the very wealthy especially if it contradicts LLM sources. Otherwise, non AI sources of information will be stomped out and authoritarianism will rule over all government and peoples, unleashing perpetual war, famine and genocide.
But other than that it will be fine.
That's the same definition as the one where developers have control over their devices, because that's necessary for those apps to exist and be used even if you're not the author of them.
Consider what most phone apps these day are. The vast majority are just a front end to a web endpoint and the app itself could just as well have been a browser. Nobody actually needs those to be apps.
Meanwhile the apps people would actually want are things like "ad blocker that blocks the ads in other apps" which are precisely what the war on general purpose computing has deprived people of. Because the scummy ad-laden app approved by the platform has to be "protected" from the "untrusted" open source ad blocker.
Likewise apps that back up the data in other apps or otherwise create competition for the platform's cloud services, competing browser engines that could properly support PWAs so that the "should have been a web page" apps can actually be a web page, etc.
Megacorps that want to lead ordinary people down the garden path may convince some of them that this is "for their own good", but convincing unsophisticated people that something is in their own interest is hardly the same as that thing actually being in their own interest.
Developers need full control over their own devices to develop apps. But they don't need users to have such control. Indeed, for most users to have such control would be unsafe, since they don't know what to do with it anyway.
> the apps people would actually want are things like "ad blocker that blocks the ads in other apps"
How many people would actually pay for such an app? My guess is, not enough to matter.
For that matter, how many people even understand that it's possible to have such apps on their phone? Again, my guess is, not enough to matter.
In other words, the point is not that Searls' and Doctorow's definitions are "wrong". The point is that ordinary users simply do not understand what the issue is with that framing.
> convincing unsophisticated people that something is in their own interest is hardly the same as that thing actually being in their own interest.
Of course not. But convincing them what actually is in their own interest--and that what the megacorps are trying to sell them is not--requires speaking the users' language. And I don't see that Doctorow's framing of the issue as a "war on general purpose computing" has helped at all with that.
To be honest, aside from medicine and whatnot, what real benefit is AI bringing?
I use it to accelerate my development and it’s great, but as far as this is concerned, how many vibe coded apps do we need as a society? And for the quality of information it’s a junkyard overflowing.
I built a strategy agent and ran through a bunch of ideas, positioning them etc. even though the AI has a good sense of that, and for a moment I believed it to have good direction, man I was mislead. I took a step back and realized this thing has NO FKING IDEA. It almost derailed me and what I need to be doing. It was very convincing and took introspection to get back on track.
Its creative writing skills are idiotic and it’s emotional intelligence garbage.
It’s a bubble everyone and save for medical advancements, the rest is a mirage. Even if companies accelerate some temporary economic advantage, the real tangible value to humanity seems to be few and far between.
But it’s cool and everyone is masterfully debating the next big model advancement.
well I had AI program three things for me yesterday:
1. custom CSS for a website I read a lot so it conforms with readability research
2. crawl a Tumblr account of a comic book artist, download their comics, and assemble them into a PDF I can print and bind into a book myself
3. pull the code of a charity fundraiser website that is terribly written, refactor it, centralize the theme, and update it for the next fundraising campaign, and deploy it
About an hour of my weekend to do this. Each of these projects was a weekend before that.
You can always ask us questions. Sincerely, the CIA /s
2013 is so long ago.
But they do need that control. Otherwise apps can't do the things they want them to do, like blocking ads in other apps, because doing those things is prohibited by the platform vendor.
> How many people would actually pay for such an app?
Why is that even relevant? The question is if someone would make that app, so that ordinary people would have access to it. And we see that this is what happens when it isn't inhibited, e.g. people create plugins like uBlock for browsers.
> But convincing them what actually is in their own interest--and that what the megacorps are trying to sell them is not--requires speaking the users' language.
At which point you're making an argument about framing rather than one about the actual thing which is necessary, i.e. for the user to not be denied control over their own device.
Because breaking the stranglehold that the tech megacorps currently have over devices is not going to be cheap. In the end, if users won't pay for it to happen, it won't happen.
> the actual thing which is necessary, i.e. for the user to not be denied control over their own device.
My point is that as things currently stand, users don't care about this. To them their device is not a computer they should control. It's just an appliance. They have no clue what "general purpose computing" means or why they should want it (to the extent they should) on their devices.
Framing it as "we'd like to give you ad blocking apps but the platform vendors won't let us" might be an improvement, but I'm not sure even about that. Ads vs. ad blocking is a perpetual arms race, and it will continue as long as the ad supported business model is the primary one the tech megacorps use, and the incentives that lead the tech megacorps to use that business model in the first place are also the ones that lead them to lock down what apps on their platforms can do.
The real solution, as the last statement suggests, is to do away with the ad supported business model. But that would require users to be willing to pay for apps that they use; and right now, it seems like most users won't. They've been conditioned for at least a couple of decades now that everything they use on their devices, like everything they use on the web, should be "free". And since those things aren't actually free, it costs a lot of resources to provide them, they have to be paid for somehow.
Or, alternatively, users could simply stop buying devices that are locked down, and buy open ones instead. There are ways to do that--but they're outliers, and I don't see how that's going to change any time soon.
https://news.mit.edu/2023/roman-concrete-durability-lime-cas...
and yet, all of it has crumbled despite the copious availability of slave labor at construction time
Or do you mean higher block rewards? Don't worry, the value of the coin automatically adjusts to nullify that.
That is the "does nothing" argument. But compared to Bitcoin, having a general-purpose [0] ASIC seems like it'd make the network less secure - if someone wants to run something like a 51% attack against Bitcoin they first have to buy a ridiculous amount of specialist Bitcoin-only hardware that will potentially lose all its value after a successful attack. Also probably be a lot easier to track who did it.
Monero the equilibriums are less obvious. Does that matter? No, I don't think so, I still own a certain amount. But it isn't a helpful axis to experiment on and I don't see where the upside is meant to come from vs just going with sha.
[0] Edited, was special-purpose.
And hot ice cubes are a thing. They've made ice at up to 650K if the wiki phase diagram is any indicator https://en.wikipedia.org/wiki/File:Phase_diagram_of_water.sv...
I'm going to reiterate that you are the one who started calling a CPU an ASIC, I'm was just trying to match your language to be friendly. A CPU is technically as ASIC for a certain class of computable functions, so if you want to call it that I'm happy to as well. If you don't want to call it an ASIC we don't have to, we can just call it a CPU.
> A Monero mining ASIC is called a CPU, by design.
This is true. If you make a Monero mining ASIC, then you have made a shitty CPU. Off-the-shelf CPUs are the best possible chips for mining Monero with, and this is by design.
The Monero hash function works by generating random programs and then running them, so that you need to have a CPU to generate hashes, and all of the optimisations in a CPU benefit Monero hashing. If you build a chip that does this, you have built a CPU. It's even built so that the instruction set is irrelevant, because the same program is run thousands of times before a new one is generated, making the overhead of translating it to any needed instruction set negligible.
