In circles the agreement is that trust creates a 1:1 value ratio. I value the tokens you mint periodically as equal to my own, and this influences the number of tokens I give you in exchange for a loaf of bread or something. If I value the bread at 6 of my own tokens, that's the price I change you: 6.

But maybe we value each other's contributions to society differently, perhaps we consult the graph and end up with a 2:3 ratio where I trust you more than you trust me. This ratio influences prices. That loaf which I value at 6 of my tokens (times two, divided by three) I offer to you at a price of 4.

Or maybe you're working to cause me trouble, damming the river I drink from or somesuch, so the trust graph gives us a 5:1 trust ratio. In this case I'm going to need 30 of your tokens in exchange for this bread because I'm aware that by feeding you, I'm giving you energy that you'll spend harming me.

After exchange, the tokens get wrapped in a layer that indicates me as well. Since the next person to accept it will be benefiting both you and me by doing so (contributing to a system that supports our various activities), they'll have to consider the trust ratio between themselves and both of us in order to determine whether to value it. This creates a risk on my part: maybe I'll accept your token and be unable to find anybody who will subsequently accept it from me because everybody I associate finds your activities problematic. (These dynamics are all implementation details, humans just scan a QR code and see a price that was determined by the weighted trust graph).

You don't encounter tokens with problematically large stacks of wrappers because demurrage counteracts inflation. We're constantly minting new tokens for ourselves, and the value of existing tokens are constantly degrading so nobody has a token that's 100 years old. That is to say, they have a half life, they decay out of existence eventually, so there's an incentive to continue to be trustworthy and useful, rather than just hoarding enough that you can then opt out of being trustworthy for the rest of your life.

> I (an important person who is very loved and trusted by thousands of people)

The ratio we end up will not be a function of how many people trust you, or how many people trust me, it'll only consider cases where I've trusted somebody who trusts you, or where you've trusted somebody that trusts me. We opt-in to the asymmetry by using variable degrees of trust to enable or prevent the activities of our peers. It restores balance to the "vote with your wallet" situation. Currently, the only way to vote with your wallet is to vote yes or to abstain. This lets you vote no.

For a first pass I'm considering using https://github.com/cblgh/appleseed-metric for the trust graph. But I don't intend to start by making apps like the one I've described here--nothing so politically charged as money. I figure I'll get the protocol working with things that are low stakes and easy to get on board with and try my hand at making something money shaped only once it's performing well for other stuff.

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.

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.

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%...

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.