More of a question for myself to understand the principles of what drives CirclesUBI.
Sarah Friend wrote a clear explanation on how the CirclesUBI process is kept going using xDai and a little amount of ETH. And that last part has now really gotten my worries as the price of ETH has almost trippled in the last month. Does this high price of ETH influence the feasibility and success of the CirclesUBI project? Why is ETH necessary for CirclesUBI? Are there alternatives?
This is the great thing of CirclesUBI, they got away as far as possible from Ethereum, but stayed close enough to benefit from most the the qualities Ethereum has. xDai in itself would probably not work if there was no DAI-to-xDai-bridge. DAI is secured by the big brother of xDai, Ethereum. It takes a lot of money for transactions on its chain, but they are secure enough that a >1 billion market (that of DAI) can run on it without any issues (except high prices).
So the ETH necessary to run Circles is really comparably low, it is just used to transfer DAI to xDai. If this will not be possible or too expensive, Ethereum as a Blockchain would cease to exist as there would be no justification for usage anymore.
However, if something like this seems likely, Circles could still search for an other Blockchain (with the values still intact as we are … on a Blockchain), but already made a bold move with choosing xDai over Ethereum. It’s a really interesting choice with regards to the pro/cons, but most people would agree that it is more on the cheap/unsecure side than of the expensive/secure (Ethereum) one.
It’s necessary to lean on something like xDai for now because the compute and storage for Circles have to come from somewhere. Whatever that source is, it must be reliable and resilient to attacks. Circles is too small to have powerful enemies right now, but if it’s successful we can expect it to attract the attention of people who would harm it.
Traditional economic incentives are currently the best way to get access to such resources: We pay for our compute and storage, and in return the people who keep xDai up and running make sure that we’re hard to attack. It’s very expensive to attack Ethereum, and therefore it is very expensive to attack Circles. (I’ve been meaning to look into how xDai achieves this at such a low cost, but for now I just trust that it does).
At least in my community, the circles web of trust doesn’t mean too much. It’ll take some practice, strategy, and determination to make it into something that I trust to provide me with important things like food. But imagine a world where circles trust was at least as reliable as bank-issued money for ensuring that your needs are met. With such a resilient trust network available, it should be possible to turn circles into a layer 1 protocol (i.e. it would not depend on any other token). This would require a certain saturation of trusted individuals who are willing to run nodes for their communities, and communities that are willing to support that activity.
Proof-of-work (like Bitcoin) and proof-of-stake (like xDai) are only necessary in a trustless context. If you’re already trusting your neighbor the farmer to grow the food that you’re going to eat, I don’t think that it’s so much of a leap to trust your neighbor the techie to run a node that supports the currency that you’re paying to the farmer.
Probably you would want to trust multiple techies that don’t trust each other, that way there’s a reduced likelihood that several of them conspire to cheat. (Luckily, it’s easy to find two techies that don’t trust each other. Just pick an emacs user and a vi user 
)
I just recently wrote a small comparison of xDai/Ethereum, maybe its helpful: https://gist.github.com/ice09/4d0f819c2475ac0292dd31769c1e8671
I love the idea of having Circles as a L1 with CRC as native currency, that’s definitely the ultimate goal, however I think that’s really a long way to go. Though I think “Proof-of-work (like Bitcoin) and proof-of-stake (like xDai) are only necessary in a trustless context” is true, even in a trusted context it helps, as it’s not only about trustlessness, but also incentives and those should be high enough. This again means that Circles must have some value to the validators… So for now and the near future I think xDai is good (enough) and without it Circles would actually not be possible (at least I don’t know how).
Thanks for the writeup @alculexum, that fills in some gaps for me.
I agree with all of you points.
When I initially hinted that (in the future) we don’t need POS, I hadn’t considered that relying on Circles that you don’t have yet to exchange for things you expect to need counts as having a stake in the system. One would trust the people running a node to not form a conspiracy partially because they’re good people, but also because if they got caught, other people would revoke trust in them and then their circles wouldn’t spend as well. I think it’s still POS.
Despite it being very far off, I’m excited about the prospect of L1 because it has some really interesting design opportunities. For instance, I’d love to see if we can make it work without an internet connection. Transactions would gossip from device to device, like how SSB works. In addition to some people running nodes, others would run access points (with connections to those nodes) in public places to help the gossip converge quicker. It would be vulnerable to certain kinds of attacks, but with incentives like those in the previous paragraph, those could be prevented.
For now I’m less worried about bad actors within a community and more worried about external third parties (like a certain censorship-friendly nation state) blocking access to xDai and breaking circles for communities that are only using them for local exchange anyhow.