undefined | Better HN

0 pointsinigyou1d ago0 comments

I think all or nothing sounds better. There was a top HN story yesterday about the curse of excessive granularity. Not only does it require excessive attention to maintain the right values - and now you're requiring coin selection as well - but can also lead to outcomes being contradictory to your intention.

Circles trust isn't about how much I actually trust you to deliver the product I bought and not harm me - it's about how much I trust you to not be a Sybil clone. But if Musk is harming me then I'll not trust his coins and people will have to find someone else to pay through.

You also have to make sure your intended outcome is a Nash equilibrium. In your system it sounds like I can set up a relay that pays out 1:1 (or 1+fee:1) without wrappers, which will quickly become a requirement, making the system harder to use, capturing a financial fee from normal system users, and unsolving the problem you wanted to solve.

0 comments

3 comments · 1 top-level

__MatrixMan__1d ago· 2 in thread

It was inspired by this mural I saw when I visited Chile. There was this whole saga of exploitation by outsiders, terminating with a brown person shoving another brown person's hand into a blender and accepting a credit card from a white person (it's right outside the street art tour place in Valparaíso). It was pretty striking to be there, a white tourist with a credit card in hand, and to see how the locals felt about the way foreign money was impacting their society.

All or nothing means that we have to chose between excluding outsiders entirely, or treating them as equals. A rich outsider shouldn't be able to use a pile of money that the locals don't have a say in and act like a king, but on the other hand there should be a gradual path to gaining the trust of the locals which has to do with whether you're helping or harming them.

I'll have to think about Nash equilibria. As for making the system harder to use... I guess there will always be the problem of displaying different prices to different people based on how trusted there are, but I think it's a small price to pay compared with avoiding the exploitation that rich foreigners visit upon poor nations: exploitation mediated by the fungibility of money. As for the other complexities of use, that's just software implementation details.

inigyouOP14h ago

Good news, he doesn't have to because money only travels along reverse trust chains. If you don't trust Elon but Alice trusts Elon and you trust Alice, then it's actually Alice who is paying you directly to shove your friend's hand in a blender. Maybe reconsider your friendship with her.

Why wouldn't it lead to a bid war bringing exchange ratios arbitrarily close to 1? You may take a 70% cut of any transaction does with you, but Alice figures out if she charges a 50% cut, she gets to keep 50% and you keep nothing of that same transaction. Then Bob figures he can undercut Alice by charging only 30%...

__MatrixMan__1h ago

I hear you re: Alice and Elon, Circles does solve that problem in a way that our money does not. But it does so in a way that lacks granularity... Suppose Elon is actually friendly in this situation and it takes a while for the locals to figure it out, all-or-nothing means that you essentially have one traitor, then two, then the pro-Elon faction and the anti-Elon faction, and then eventually everybody realizes that Elon is actually ok.

While this is occurring, anybody who is doing business with Elon is doing so at 1:1 rates, so nobody is being rewarded for taking the risk of getting to know him. And on the flip side of that, suppose we were wrong about him and he's actually a big problem for everybody around him... people have to either fully embrace his economic impact, or insulate against it all at once. It takes what could otherwise be a small mistake and forces you to make instead the big mistake. I think it's important to let people express their half-sure stance within the protocol rather than forcing them to put it all on the line.

As for the bidding war... imagine there's a mining company that has just showed up in town. Two grocery stores exist, one gives the miners a rate of 2:1 and the other gives them a rate of 3:1, because both stores don't trust them, but they have different tolerance for risk.

The 3:1 store has the opportunity to undercut the 2:1 store by switching to 1:1, and then the'll end up with all of the miner tokens. But then they have to spend that money somehow--probably by buying food from local farmers. Well if the mining operation ends up poisoning the aquifer and killing the crops, then the farmers are unlikely to accept that money, so it ends up only being useful for doing business with other miners, which isn't likely to be useful for the grocer because how much yttrium does a grocery store need anyhow.

Elsewhere in this thread somebody was arguing that I didn't need a new accounting system, I just need to talk to the people around me and only do business with the trustworthy ones. By encoding into prices the perceived risk that the other party might be a bad citizen in the future (thereby creating a liquidity problem for whoever holds their tokens) we create an incentive to actually have that conversation an act on what is learned from it. And that creates an incentive for that party to not be a bad citizen in the future.

Come to think of it, it might even enlist locals who have decided to take a risk on feeding the miners to take on a bit of a policing role. They now have a financial incentive to prevent the potential bad behavior. Circles has that dynamic too, but requiring somebody to be all in to even dabble makes it less likely to happen.