Hey @j-cook!
Awesome read! Love the onion idea! And I particularly like that it aligns pretty well with @tjayrush concerns. It leaves a room for tweaking and … friction (A cool helpful term that I adopted while talking with Phillip Sheldrake. Totally recommend reading his essay. Bends mind in proper ways). Also thanks for describing airdrop farming and “exploits” vs “hacks” in DeFi. Helpful too!
Wanna insert though a short note on Cost of forgery and plutocracy. It seems to me that the “Plutocratic tendencies” section relates to “staked identity” which requires that a person stakes money on theirselves or peers (described in this post). Anyway it makes no sense in case of Price of forgery (PoF) or Cost of forgery (AFAIK the term got into Gitcoin from this paper and then transformed a bit, but ”Price of forgery” is how we still call it in Upala, the reason and the difference is also described in the methodology). Explaining my point further.
“rational actors are disincentivized from attacking a system”
The methodology of Price of forgery measurement does recommend a setup of an environment where conditions slowly approach the point of disincentivizing the rational actor from attacking the system (to be specific attacking matching funds). But disincentivizing is not the goal, the goal is measurement.
It may happen that we would nudge exiting Fraudsters to sell their bots instead of using them to extract matching funds. But chances are they will be outperformed by other Bot farmers who would sell their armies faster and for lower price. The measurement would happen long before the above described condition. Moreover we could be satisfied with just a single measurement (single bot sale event). We don’t need to keep the environment (and funds at risk) forever. Just rerun the measurement periodically to stay up to date.
So In the context of fraud detection the benefit of using Upala protocol and figuring out PoF is just knowing. Otherwise it would require enormous amount of money and does not make sense.
“One way to manipulate the environment is simply to raise the cost of attack to some unobtainable level, but without excluding honest participants.”
Accordingly, we cannot set the Price of forgery. After we measure it we can only require a certain PoF or alternatively we can set trust bonus in relation to PoF (just an example, still learning how Fraud detection works in Gitcoin - would be happy to brainstorm btw).
As for plutocracy. For any human verification method there’s price of forgery anyway whether we measure it or not. Learning PoF does not affect this fact. We could require 6 stamps or require a sum of approved PoFs behind those stamps - it would be the same thing. If we wanna raise PoF, we have to offer more stamps (and measure their PoFs). There’s no other way.
Also there are no money involved for the users. There’s no deposit or stake. They don’t even have to know that they got PoF calculated/measured. It is bots who do all the job. No wealth-gating occurs.
PoF makes plutocratic tendencies neither stronger, nor weaker. It is a measurement tool. Which I believe could be a very useful one in the fraud detection pipeline. And which could be easily integrated into the fraud detection 
.
If you wanna dive deeper check out this Price of forgery measurement campaign grant for details. Also check out Upala dashboard and article on EthResearch on price of forgery. I’m working hard to deliver an up-to-date documentation. And best of all let’s chat (my twitter).