Tezos conversation

[Clay Shentrup]

I just had a pretty long conversation with Arthur Breitman, whose leading the Tezos project.

Tezos uses Approval Voting and I wanted to suggest Arthur consider the Bayesian Regret framework and perhaps go even further. He was aware of Score Voting and Condorcet, but says that his deeper interest is actually in the game theory of the decision making process at a more system level.

For instance, he cited the game Nomic.

https://en.wikipedia.org/wiki/Nomic

It starts with simple rules can be amended. The ideal is that edits to the system should stay in the best interest of the group over time, which isn't just about the voting method but also about how incentives are created. He posited the situation:

Proposal

- Arthur gets a million dollars.

- All the money from all the money who bet against the rule is distributed equally to the people who voted against the rule.

So the decision could be bad because it just causes inflation, but individuals still want to vote for it because it helps them individually.

One proposal (I couldn't understand the name he gave for it) is to take bets on whether a proposal will be good or bad for the "economy". E.g. say we had a bill to reduce CO2 emissions, and people could place a bet that pays out if it really does reduce CO2. Then you pass bills if the bids are correct.

Thing is, who gets to decide whether the bill passes if people think it'll lower CO2 vs. increase? Maybe Arthur has a plan for that too. I'm skeptical that there's much you can do other than just use a good voting system. But I did want to run this by Warren, Andy, and Jameson at least. 

[Warren D Smith]

sounds like "thomson insurance."


--
Warren D. Smith
http://RangeVoting.org <-- add your endorsement (by clicking
"endorse" as 1st step)

[Clay Shentrup]

I couldn't find a definition of Thomson insurance. Can you describe it or provide more information so I can research it?

[Warren D Smith]

T.N.Tideman's book: Collective decisions and Voting, Ashgate 2006
chapter 16 discusses Thompson insurance among other things of that
ilk.

google found
http://mason.gmu.edu/~rhanson/futarchy.pdf

tell us more

[Jameson Quinn]

Thanks for letting us know about this, Clay.

Obviously, Tezos (and specifically Breitman) have a broad array of governance issues to deal with, and only some of them relate directly to social choice theory as most of us understand it. Questions like "what if people set up referenda that intrinsically bribe voters" or "what if a majority votes to steal from a minority" are not voting theory (and the old meatspace answers of "secret ballots" and "constitutional protections" are decent starting places).

But voting theory itself is definitely still a factor. Less of one than you might think; many decisions are independent and binary, so that any voting method works, and as for representation issues, delegation works; so for many things, you need neither single-winner nor multiwinner voting methods besides the most trivial ones. But sooner or later, you'll have something with more than two natural options; and you may have need for multiwinner methods.

I think we should approach these people. Clay, could you provide an introduction?

Thanks,

Jameson

--
You received this message because you are subscribed to the Google Groups "The Center for Election Science" group.
To unsubscribe from this group and stop receiving emails from it, send an email to electionscience+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Clay Shentrup]

So while this is incredibly simple, I think the real key here is to have a secret ballot, though I suppose that's not possible with any kind of system where you're not forced to vote alone.

If you can't prove you voted a certain way, then you can't reap "external" incentives. So whatever actions you take are generally incentivized to be in the best interest of the protocol and other users.

But okay, if that's not possible, what are some other options?

[Clay Shentrup]

Jameson, Warren et al.

I really would like to provide value to these guys, as I think Tezos is really promising. And unlike our efforts to change political electoral systems, it's viable.

Having said that, I really suspect that the only major areas of improvement here are:

1) Use an even better system than Approval Voting, such as Score, SRV, or 321.

2) Make voting anonymous (which may not be possible)

Warren — I'm surprised you're not more interested in this given the interesting mathematical challenges of these Bitcoin-like protocols, as well as the potential to significantly impact the real world with better voting methods. The Bitcoin market cap is about 71B for example.