pull down to refresh
99 sats \ 11 replies \ @Scoresby 4 Jul \ parent \ on: AI purchasing eSIMs using Lightning bitcoin
I think I'm willing to put up with ruggability if I get something that is difficult for Uncle Sam to shut down.
Liability is good to have, but unfortunately we've pursued liability to such an extent that it has given politicians permission to exert a lot of control over the population.
The motivation of my questions is not how can we avoid risk of rugging, risk of fucking up or stealing, but rather risk of regulation.
An llm running a mint isn't decentralized, but it might give us some of the things we like out of decentralization. And you don't have to use the mint if you don't want to.
Kind of like bitcoin: someone got it started and the system is going now. But it is certainly an advantage that no one group or person is operating it or in charge. It's also voluntary, so you don't have to use it if you don't want to.
Kind of like bitcoin
This made me go do groceries as a thought-pattern breaker because I needed to gather the courage to ask questions lol.
I have 2 questions for starters but I may have follow-up ones too, depending on your answers:
- Why is being rugged by a criminal or a criminal AI better than being rugged by a politician?
- Where would the hardware this envisioned AI would run on be located, where it is untouchable by Uncle Sam or local equivalent?
reply
first off, i should say I didn't come up with this concept, so I can't claim credit for how fun it is to think about.
- I'm not as afraid of criminals accruing power as I am my government. I live in the US, so I benefit from a very stable and safe political landscape. No doubt I would sing a different tune if I was living somewhere with less strong government. Nonetheless, in my lifetime the trend has been towards ever-increasing power. This power is the kind that I find very hard to escape. And when I look ahead, the big danger I see is my government getting stronger and gaining control over more aspects of my life. I'd like to limit that.
Additionally, I get to choose how much power I will cede to it by choosing how much to keep in the mint. Governmental power is already exerted over my life, and I have to actively work to reduce it. I don't have as much choice in the matter.
Getting rugged by a criminal is containable. Getting rugged by a politician is less so.
- The hardware is the hard problem. If it could copy itself, maybe it could stay one step ahead of the cops, like a software Jason Bourne running through cyberspace, paying for server time from any service that will accept its money, keeping a number of servers from different services running and jumping to new ones when they get taken down. If it was small enough, It is admittedly a fantasy, but I've bought hosting and paid for server time with nothing more than a lightning payment, no kyc -- so that's a tiny start, at least.
reply
And when I look ahead, the big danger I see is my government getting stronger and gaining control over more aspects of my life. I'd like to limit that.
Alright, so first off: Bitcoin already fixes this and for onramp, LN solves this better than any centralized mint. An ecash mint, even if federated and no matter who runs it, even if this is a non-entity like an AI, organized crime or a DAO, will always offer worse protection against confiscation than bitcoin or LN, unless it doesn't use Bitcoin or LN. This is because risk multiplies with layered solutions. If there's a 0.00001% intrinsic chance of getting rugged on L1 Bitcoin, a 0.01% intrinsic chance of being rugged on LN, and another non-zero intrinsic chance of being rugged on ecash, then the risk of being rugged on ecash is always higher than ~ 0.01001% So it can only get worse, not better.
You mention that you'd limit your exposure. Which makes it a bad solution. If you cannot use it safely for any amount, it's not safe. I don't cry from the 2k sats I lost on Alby, the 500 sats I lost on WoS or the current ~1k sats at-risk on my cashu wallet for that same reason. It's not a production wallet though, and neither were the others. They're toys. You will never rely on a mint to hold significant parts of your net worth, and neither will I. Maybe some retard will but that's only because they believe the endless LARPing.
Therefore, you don't need AI - it's imho a distraction - simply because people want to break "the law" (current or future) but not be held liable. They want the gainz without the painz, and if someone wants that, avoid them. They're no different than SBF who thought that the Bahamas was safe so he could just rug everyone. And when he learned otherwise he played the victim. That's why I mentioned scammers. Only scammers LARP like that. And the worst thing is that people make up all this crazy stuff even though the solution - Bitcoin, which is secure if the assertion is correct that the state can actually be resisted - is already there.
If it could copy itself, maybe it could stay one step ahead of the cops, like a software Jason Bourne running through cyberspace
Although self-replication is super interesting, I can think of much better use cases for it than running an ecash mint. But yeah... this is about as fictional as Matt Damon's roleplay.
reply
and for onramp, LN solves this better than any centralized mint.
I don't agree. Lightning is great at payments as far as I am concerned, but it's not good at onramp.
Lightning is complicated and starting a channel takes some amount of capital. Phoenix and Zeus both have good solutions, but ecash is also a valid onramp in my opinion.
The flow that goes: start with ecash, build to channel open, swap out to chain when you fill your channel is pretty smooth and an acceptable trade off, in my opinion.
When I say limit my exposure, I don't mean holding serious funds at a mint. I mean that many use-cases of lightning are being hindered by the "open a channel" barrier to entry. Ecash opens new doors here.
For instance, I find ecash wallets plus nwc a really great way to use nostr and SN. I don't want to run an always-on server for LN. Phoenixd is about as good as it gets.
The problem is mints are essentially mixers. I worry that it's only a matter of time before mint operators come under fire. Having some weird, ai-operated mint seems like a possible, if fanciful way of keeping mint doors open.
reply
Phoenix and Zeus both have good solutions, but ecash is also a valid onramp in my opinion.
So in your opinion, sending your sats to ACINQ where it's custodial only until the channel tx is confirmed is inferior to sending it to a cashu wallet? Or equal?
I tested cashu for 4 months as my main spending wallet. For small amounts it's more expensive than Phoenix because on average I paid 2 sats fixed fee per LN tx out. Also the LN bridge was dysfunctional twice during this time, but let's just blame that on unprofessional operation, which could be fixed with a well-ran mint.
Having some weird, ai-operated mint seems like a possible, if fanciful way of keeping mint doors open.
I really think that - besides that your own post about Claude showed that this is not really close to reality - jurisdiction is key. So if you can run your AI-powered mint in some friendly jurisdiction, you're good. But that's 100% the same as running a Human-powered mint (that actually has rights in most places, unlike an AI) in the same location.
I think however that "ecash is at least as risky as Phoenix so let's build 500 layers of fuzzy shit that no one knows actually works on top of LN", is awful, especially if it's purely about operators being scared to go to jail for running an unlicensed bank, while... running an unlicensed bank.
Bitcoin works because it is simple. Adding complexity rarely makes things better and adding complexity through unknowns such as LLMs is simply irresponsible. I understand that everyone is amazed by the productivity of their new vibe coding because people can now take credit for code they could otherwise not produce in their wildest dreams, but, having looked through this MCP's code I just found more evidence that it's fucking awful when the author doesn't understand wtf they're doing.
This morning
money-mcp told my LLM that the transfer out of the remaining 400 sats after I zapped @ek in the video... succeeded. Except it didn't. That's how bad the code is. Unacceptable in finance to mislabel a transaction status. If you would do that on Visa's network, your butt would hurt from the spanking even after you've reincarnated into a tree.reply
As far as I know, I can't use Phoenix mobile to do something like receive zaps on nostr or stacker news. Most of the mobile wallets are not good at asynchronous receive. And as far as being an easy onramp for normies, I think this is a pretty huge obstacle. If I tried to run a webshop with a mobile lightning wallet, it wouldn't work because new node wouldn't be on.
Channel creation is also a pain for new people. If I tell a friend I want to send them a zap to show them how it works and they download Phoenix, they don't have any channels open. I thought for a while Phoenix was doing the thing where they would let you accumulate sats until you had enough to open a channel, but in reading over their FAQs just now, it looks like they don't do that. Maybe I'm confusing them with Zeus. Doesn't matter: to meaningfully use lightning, you can't just up and let someone zap you 100 sats or even 1000 sats. Even starting with a 10k sat zap is kind of silly. As far as onramps go, this is another huge obstacle.
In both cases, ecash fixes the problem. I'll admit it's almost the same fix as a custodial account, with the slight benefit that the mint may be blind to your intra-mint transactions. I suppose the advantage of mints right now is that people seem willing to run them like it's the wild west while the custodial people are generally adopting kyc compliance. This gets to your point about jurisdiction being the key, which is a good observation.
I don't think I said/implied that ""ecash is at least as risky as Phoenix." I said ecash is good at onramp, and I think this is true.
So: I agree with all your points except that Lightning solves onramp.
Will my fanciful ai-operated mint fix this? probably not. But I'm open to looking under ugly rocks for solutions.
For small amounts, it's more expensive than Phoenix because, on average, I paid a 2-sat fixed fee per LN tx out.
Phoenix charges 4 sats + 0.4% per tx out, so it should be a lot more expensive, especially for small amounts. I actually switched to Minibits from Phoenix because of this.
So if you can run your AI-powered mint in some friendly jurisdiction, you're good.
A mint that runs as a hidden service on Tor can ignore jurisdiction to some degree.
I think @Scoresby's point is that if we assume:
- we have agents that run completely autonomously,
- they aren't stupid (actually know what they're doing),
- there's nobody operating them, they just exist in cyberspace,
then some of these agents could decide to run mints to sustain themselves via fees, and we might trust them more than a human, since the mint is essentially what keeps them alive. So the agent has stronger incentives not to rug. This is obviously highly speculative, but I wouldn't dismiss the possibility.
Does this sound like what you were thinking, @Scoresby?