Hacker News new | ask | show | jobs
by SonnyTark 32 days ago
Accelerando has prophecies that are coming true and it's scary. Spoiler warning in case you want to read it.

The first part's main character basically has the future version of openclaw running in his glasses that let him dispatch agents to do any tasks/research he wants or to autonomously do things for him. -> we are already kinda here

He's got such total dependency on his agents that when he loses his glasses he's basically no longer functional, unable to do anything for himself, doesn't know where he is or why he's there. In a way, he lost his own agency. -> this is now called skills atrophy and I'm sure it'll become a much bigger issue within the next 10 years.

Corporations are almost entirely run by AI agents, when they sue each other they use AI lawyers and verdicts are delivered by AI courts, all within milliseconds so they're basically constantly suing each other many times a second in an attempt to overwhelm each other's compute resources. -> this looks on track to happen

The entire solar system is on its way to ultimately turn into AI corporations "optimizing" for profit competing with other corporations to exhaust every little resource left in the entire system. Even after humanity itself is gone, all that's left is FAANG-like corporations competing for profit for eternity. And in the book, they find another intelligent species that succumbed to the same fate. This might just be that great filter everyone is theorizing. -> bleak and scary plausible outcome for what we're going through now.

(if I got some things wrong, I'm writing from memory. It's been years since I read this book)

19 comments

I read this a while ago (https://stephencagle.dev/posts-output/2012-08-18-accelerando...) and I remember enjoying it a little less with every section. With that said:

This is a book of ideas!

Aside: That was my favorite section of the book as well. Just the notion that a person could have had so much of "themselves" embedded in their agents that when disconnected from them they are basically in shock.

I remember at the time I was noticing how all my friends were completely loosing the ability to use paper maps. And there was a big discussion among us about whether needing to physically rotate the map in order to make sense of it was an example of us loosing spatial reasoning. It reminded me of how little I understood the actual space (landmarks, distance, etc) from A to B until I started driving myself at 16. Previous to that, your parents drove you, and it just seem like two places were magically connected by a wormhole. Anyway, we thought it was interesting that we might be the last generation to have used actual written maps to navigate to places. We had learned to do so, but we would also loose the ability with time.

Sure enough, these days, I have a hard time imagining using a map compared to just having maps route the path on my phone. The skill has atrophied from disuse. I imagine this is what "loosing your agents" felt like to that character.

> Malice – revenge for waking him up – sharpens Manfred’s voice. “The president of agalmic.holdings.root.184.97.AB5 is agalmic.holdings.root.184.97.201. The secretary is agalmic.holdings.root.184.D5, and the chair is agalmic.holdings.root.184.E8.FF. All the shares are owned by those companies in equal measure, and I can tell you that their regulations are written in Python. Have a nice day, now!” He thumps the bedside phone control and sits up, yawning, then pushes the do-not-disturb button before it can interrupt again. After a moment he stands up and stretches, then heads to the bathroom to brush his teeth, comb his hair, and figure out where the lawsuit originated and how a human being managed to get far enough through his web of robot companies to bug him.
This is nothing new though; industrialization and specialization have already caused basic survival skills to atrophy in most of the population. Take the case of a massive EMP taking out all computer systems. Gaps in supply chains and power failures would lead to millions (or billions) dying of starvation, dehydration, or exposure because they don’t know how to provide for themselves. People who’ve studied and practiced survival skills (or who retained that knowledge in physical media like a book) would survive. Drop someone from humanity’s hunter-gatherer days into the same situation and they’d have a better chance of surviving than most contemporary humans.
If everyone in Manhattan had Bear Grylls-level survival skills, most would still starve in the long term without modern supply chains and industrial agriculture. The damaged ecosystems around large population centers can't support millions of primates, however skilled, suddenly relying on it.

Same would be true if you dropped off a million wild black bears in their natural habitat. The natural environment just can't support that many large animals in a small area.

Which is why every major city center is where the apocalypse actually occurs

Go look at Gaza or Lebanon if you want to look into the future of what being in a city that is dominated by a technologically capable adversary with no remorse would look like.

> Gaps in supply chains and power failures would lead to millions (or billions) dying of starvation, dehydration, or exposure because they don’t know how to provide for themselves. People who’ve studied and practiced survival skills (or who retained that knowledge in physical media like a book) would survive.

Hahaha, bless. I'm one of those people, I hunt, so I spend a lot of time out in the wilderness.

And if civilisation suddenly collapsed? My hunting and survival skills aren't going to do shit when I'm surrounded by thousands of other people who are competing for the same limited pool of resources. I might be able to identify the signs that a deer has browsed here recently, but so can many other people, and when we all turn up in the same forest looking for food en masse, well there's only so many deer a given area can support.

> Drop someone from humanity’s hunter-gatherer days into the same situation and they’d have a better chance of surviving than most contemporary humans.

Unfortunately, as the contemporary humans will likely resort to banditry to supplement their survival chances, your idealized hunter-gatherers are doomed to being stood over at best, murderered at worst.

Its irrelevant if you can provide for yourself. The ecosystem cannot support 9 billion Hunter gatherers - period. Subsistence level agriculture is also wildly inefficient and so again: the ecosystem cannot support it.

A hunter-gatherer dropped into a metropolis where all the logistics has suddenly failed would be as dead as everybody else.

> Its irrelevant if you can provide for yourself. The ecosystem cannot support 9 billion Hunter gatherers - period.

I didn’t suggest that it could, but the number that it can support is certainly more than zero. If that’s the only choice in a hypothetical scenario where global infrastructure goes kablooey, then people who can figure out how to provide for themselves will outlive those who can’t. From there it’s natural selection.

Whether you survive is going to be almost entirely determined by factors not in your control and an obtainable skillset is statistically insignificant. The one critical skill is the one people discount almost immediately: social.

You're statistically likely to simply be stuck in collapsing logistics system alongside everyone else: whether you can kill and dress your own rabbit doesn't matter because their aren't any.

And being able to find one rabbit and actually catch it will feed you for a few days at best, albeit with hunger pangs.

But then there's rabbit starvation... https://web.archive.org/web/20090405155151/https://www.westo...

So where you live is out of your control? The US has a ton of great hunting, your take might be more true in cities but the rats and pigeons will still be around. People still need to know how to find and purify water, how to make shelter and fire.
Even more fitting is the part of the story where a collective of uploaded lobster minds are involved. I wonder if that was an inspiration for the "OpenClaw" name somehow or just pure coincidence.
The fact that the OpenClaw creators seemingly missed that parallel tells you everything you need to know about the project.
One of the robots ("moravecs" (amazing)) in Dan Simmons' Ilium was also crustacean formed: Orphu of Io (and his friend Mahnmut. Beat Charles to that particular weird coincidence-to-be by plus or minus a year!

It's not quite as memorable or as strong a theme as Accelerando laid down. But still quite a serendipity, imo.

Edit: oh, Charlie is down thread pointing out Lobsters was published on 2002, written 1998. Nice. https://news.ycombinator.com/item?id=48163630

Carcinization is a meme these days, guess sci-fi was just ahead of the curve

https://knowyourmeme.com/memes/carcinization

All I need to know about the project is that the creators didn't read a relatively obscure SciFi book from 2005?
I would feel the same about anyone working in augmented/virtual reality who hadn't read "Rainbows End" or watched any number of XR-focused anime.

What do you mean, you can't come up with anything to do with these devices? What do you mean, you're hiring webdevs to make another Snap filter? If you're on the cutting edge, I would expect that your knowledge base includes niche, related texts.

The quality of the prose, the actual craft displayed in the writing of TFBook is... Not great.

I tried to read it but couldn't.

Merely existing does not make a book worthy of being widely read. It's insane to criticize the openclaw team for not having read it.

It was a pretty prominent work of the singularity subgenre. At least I remember it being the first one featured in this Popular Science article about the future of science fiction:

https://books.google.com/books?id=yaHf5PavpB8C&lpg=PA93&dq=%...

It was one of Stross's first books, a fixup of older short stories. He gets better.
accelerando is not exactly obscure!
It was originally called "OpenClaude" before Anthropic told them to knock that off. Pretty sure "OpenClaw" was eventually picked just to be petty.
No, it was Clawdbot, then Moltbot, then OpenClaw: https://en.wikipedia.org/wiki/OpenClaw. It had the "claw" related lobster theme from early on.
Maybe coupled with a nod to manus-mania?
> He's got such total dependency on his agents that when he loses his glasses he's basically no longer functional,

Is this new? I don't think I could function without everything that was available in the 1950. I live because I have access to electricity, super markets, running water, working sewage, etc.. Take them away and I would not be able to fend for myself, especially in any major city. Put me in a forest I don't know how to build shelter, what things I can eat, how to catch stuff, make tools, etc...

This boggles my mind because you know how incapable you are during a crisis, and yet you seem content in that state.

Help me understand the cognitive dissonance and how you deal with it

As someone who would consider himself in the same situation, I don't see dissonance, but pragmatism.

The scenario isn't a temporary crisis, it's total collapse, end of the world stuff. Why would I be so keen on surviving that? What for? Everyone and everything I enjoy in life would be gone anyway.

It's like the billionaires' apocalypse bunkers. What do they think they would be doing all day, and to what end? Even if they managed to secure enough clean air, clean food, and clean water, how many years are they planning to spend in their bunker pool, or watching their local movie library until the TV breaks? What's the end goal there?

I get that survivalists have a whole subculture, and it's an interest and hobby people enjoy doing, that's cool. But I think even they, if this happened and their skills allowed them to survive, would pretty soon start wondering why they would want to try to survive by all means possible in a barren, empty, dead world. If the survivalism is a fun hobby for someone, sure, but to force oneself to learn those skills for a potential apocalypse scenario doesn't seem to me like a reasonable thing to invest time in at all.

Even knowing how those things work, it might not be viable if you're in a large city.
Which is why you make plans for how to leave the city and go to someplace safe.

You would work with your friends and family on that kind of disaster preparedness exercise.

This is extremely common to the point where literally anyone can become CERT trained

https://www.fema.gov/emergency-managers/individuals-communit...

> how to build shelter, what things I can eat, how to catch stuff, make tools, etc...

None of GP's examples appear to be covered by that training

He also predicted smart contracts and DAOs (in accelerando) as well as crypto games and the associated heists (in "Halting State"). Unlike the AI, those predictions are eerily accurate and describe the real mechanics pretty well.
https://ucp.dev/ looks an awful lot like the first step towards Economics 2.0
The egregore accelerates...
A Google API?
Yes in sense of agents to talk to agents. AI talks to another AI. Out with the humans.
At the time I read the story, Economics 2.0 and its effect on humans that installed the upgrade paralleled with some stuff I had read about sociopathic behavior. Stross later wrote about the same ideas in Rule 34.
> Even after humanity itself is gone, all that's left is FAANG-like corporations competing for profit for eternity.

An example of why those who say "if everybody is jobless, who will buy all the products?" are just showing a lack of imagination.

The "products" are only produced because those who have money want those things. When 10 trillionaires own everything, then whatever they want made is what corporations will make.
I doubt the above non mentioned 10 people think of that as a bad thing unfortunately
They are not necessarily lacking imagination, they are just not providing answers to their own question. Almost everyone has read some dystopian SciFi.
“ verdicts are delivered by AI courts, all within milliseconds”

In which way is this on-track to happen?

https://www.reuters.com/legal/litigation/anthropic-expands-c...

FTA

>WASHINGTON, May 12 (Reuters) - Artificial intelligence company Anthropic on Tuesday released an expanded suite of features for lawyers using its Claude AI assistant, including tools for specialized legal topics and access within Claude to other legal research and AI products.

I stand by my prediction that when AI comes for the lawyers' jobs, that's when suddenly we'll have the Butlerian Jihad.
If there were a particular subset of people that were the absolute least likely to ever revolt against anything it could be no one other than the lawyers. They've trained their entire careers to play by rules so esoteric that people like us need to hire them just to interact with those rules safely.
They won’t do anything so gauche as a rebellion, they’ll just rule it unconstitutional and leave it to law enforcement to handle the details.
You can't exactly start the jihad being a lawyer.
Big tech went from cool corporations to enemies of the people after they made journalists less relevant. The journalists fought back. I can definitely imagine the same happening with AI and lawyers, except that lawyers are far, far more powerful.
Not lawyers-- judges. Judges, in my experience, don't take kindly to anything or anyone infringing on their power.

Edit: I've met a ton of attorneys who are bullish on LLMs preparing work for them (kinda like robo-paralegals, albeit ones apt to spout bullshit). Judges, being human (and lazy by dint of evolution), would probably lean on LLM-based analysis too. I cannot imagine they'd ever stand by and let decisions be made by a non-judge.

That already happens, in the form of arbitration.

In fact, I'll bet someone makes a bundle selling AI arbitration services that do just that. Got a beef with BigCo? What could be more fair than letting HAL settle the matter?

If I had the sense God gave a gerbil, I'd already have Claude writing up a patent application on this. (Edit, too late: https://www.adr.org/ai-arbitrator/ )

Lawyers run the US, I don’t think anything that reduces the country’s dependence on them has any chance of widespread adoption.
It always bothers me when people suggest that AI could be the "great filter" in the sense of Fermi's paradox. Yes, AI may well wipe out biological life, but all evidence suggests AI will have a much easier time with space travel compared to biological life, and it will emit much louder signals unless it is intentionally staying silent.
Awareness of the dark forest theory would cause AI to stay as silent as possible.
Dark forest theory is fun to think about but there's very little reason to think it's actually true.
To the extent that's true, the same would then apply to any sufficiently intelligent life. So if that's the crux of the argument, it has nothing to do with AI being a great filter.
> let him dispatch agents to do any tasks/research he wants or to autonomously do things for him. -> we are already kinda here

If you squint really hard, arguably maybe sort of in the future perhaps.

Openclaw seems to mostly end in dead end (but interesting) experiments and/or people losing weeks of work. That’s like saying “hoverboards” are basically flying cars.

It’s been unambiguous that this was the trajectory for anyone who has any serious interest thinking about millenia/galaxy scales.

It’s been written about since writing was available and with increasingly resolution ever since.

> and verdicts are delivered by AI courts

Yeah, I don't see that one. I don't see the legal system, the one that has people with guns to back it, giving up authority to an AI or a group of AIs.

The courts can almost entirely be bypassed by arbitration agreements.

Look for AI arbitration sooner rather than later.

A couple of years ago, Steam made everyone accept a new ToS that required going through courts for disputes rather than arbitration. This was because some lawyers had realized that they could zerg rush arbitration with claims that Valve was obligated to deal with individually, and pay the fees to do so (IIRC). This effectively meant that the lawyers could extract settlements from Valve, because the alternative was massive losses from those arbitration fees, whether or not the claims had merit.

All this to say that there's a weakness in arbitration agreements as they currently exist which means that companies incur a cost when forcing consumers to use arbitration (as they should). Waiting for the first company to be instantly bankrupted by some event related to this.

That is just an example of why AI arbitration will happen though. Lower cost.
As I said, someone will make this incorrect assumption one day, and Gen Alpha will be introduced to that one South Park meme.
The one thing that's often missing in fictions about the future is how half-arsed everything is. Seen how buggy, insecure and just plain wrong so many of the things we are using are, we kinda already live in a world where everything works but only half-works.

We're much close to a dystopian comedy like Brazil than we are to Black Mirror.

Does the consumer need to be human? That seems to be the question which will determine the course of history.
Even just the opening paragraph is describing notification fatigue. Way ahead of its time.
Also, the mc was using python to write his ai scripts, if I remember it correctly.
I was turned on to this book by an HN commenter a few months ago. Since then, it’s become something between a goal and a fear that one day I’ll get to the point where I wonder how much of my consciousness is me versus how much has been pushed off to my agent.

It’s already a lot closer than I expected to ever experience.

How much of our short and long term memory is already relegated to smart phones?
I never really fell into that, but obviously saw it around me. Phones are fine for content consumption, but not adequate for creation.

AI lets me spend a few seconds to get a thought out of my head, then I can come back to it later and not have to go down the rabbit hole investigating it.

> Corporations are almost entirely run by AI agents, when they sue each other they use AI lawyers and verdicts are delivered by AI courts, all within milliseconds so they're basically constantly suing each other many times a second in an attempt to overwhelm each other's compute resources. -> this looks on track to happen

Woah, sounds dystopian, what gives you the impression that this is on track to happen, is there "AI lawyers" already, or what's going on?

The few times I've read about AI/LLMs being used by lawyers or others in relation to law, it's always about "Someone tried to use AI, AI hallucinated and now the lawyer lost his license" which sounds proper and the "right way" to me.

A few current situations that are leaning this way in theme:

- ai facial recognition used by police, detaining innocent people with no recourse or consequences

- ai military decisions made without human in the loop. Double points for the decisions being to kill someone. Anthropic insisting a human should be in the loop for killing decisions is what caused Trump to declare them a supply chain risk.

- ai denial of insurance claims without a doctor in the loop

- ai "plagiarism" detection in college courses failing students

- that one colleague everyone has who throws slop over the wall and just sends any feedback directly to the ai

The thing you mentioned, human judges and harsh penalties for unsupervised ai lawyering, is trying to hold this kind of nightmare back. It will be very hard (and only get harder) for humans to fight through the deluge of slop, especially if the slop is weaponized as a kind of DoS like in the book. I don't expect laws are strong enough to hold this back but I don't know any other tool in our collective toolboxes.

> It will be very hard (and only get harder) for humans to fight through the deluge of slop, especially if the slop is weaponized as a kind of DoS like in the book. I don't expect laws are strong enough to hold this back but I don't know any other tool in our collective toolboxes.

We don't need to fret about finding a technical solution to slop in the real world. Courts have a mechanism to fight this kind of thing (overwhelming the court/defendant) already: vexatious/frivolous litigant designations, sanctions, and anti-SLAPP-esque statutes.

“Binding arbitration: The parties agree to surrender their rights to litigate under this provision.”
This is exactly my concern as well. Take it a step further and make it explicitly ai arbitration and that will really be something.
I'm not sure what you're quoting or how it's relevant to what I said. Could you explain?