Addressable softforks - trader pseudo-veto

[turpin]

Rather than going  all or nothing on the integrity of the network with each wager - using checks and balances of voters, audits, and miners - I suggest using supply and demand. Say there is a successful attack. Say even that it's proven ahead of time that it will be successful so to avoid being penalized voters switch over, and only some tiny minority of shareholder voters stay honest - say 1%. No audit is triggered. The miners are asleep at the switch and don't veto. That decision gets made wrong, the people who should lose win, and those who should win lose. Is this the end of the PM market, or can it instead be the beginning? Say that any non-unanimous decision results in an addressable softfork. A future trader can post an opening wager that indicates it will be decided as if that past decision had gone the other way. This allows the traders to decide which subset of shareholders to trust. Thus any conflicting decisions will split the market into two markets - a liar's market and an honest market. Each market is supported only by the wage makers who use it. Both markets are mined on the same chain. Only unused markets are eventually trimmed by miners and hardforked out of existence when potential future fees from that fork don't justify its existence. Now the honest shareholders will get most of the business from anyone seeking an honest prediction market, and will address future wagers to the 1% and trade only those wagers.  People who want a corrupt market are free to use the main fork and get their money stolen. Over time the honest 1% will prevaill through supply and demand, and will regain any temporary losses from betrayal through perseverance, while the 99% only get their profits from the open wagers when their exit scam became public knowledge. Now the PM can survive a 99% attack from liars. Now that is good Byzantine fault tolerance

[turpin]

Because blokchains are used, all the information needed to calculate the softfork (let's call it a voteshare fork) is in the blockchains. All that is needed is for the opening wager to specify which decision he disagrees with, and voteshares can be recalculated for deciding this prediction. In case of multiple such bad decisions, the last few bad decisions can be referred to, and any bad decisions they presume are inherited, so the added transaction fees from longer message size discourage arbitrary voteshare forks but don't unduly burden well motivated voteshare forks. When a particular generates more fees than the main fork, miners can then vote to make that the main fork.  By default miners vote for the fork that is generating the most revenu in wager activity - they only need take active interest if they wish to vote for the less active fork. After two weeks (or equivalent number of blocks) of miners voting 2/3 to change forks, the change goes into effect and the burden of higher fees switches to the less used fork.

Incidentally this votershare fork also could be used to address moral issues. Voters who truthfully answer a question from a PM that was obviously used for bribery such as Bell's AM could find themselves softforked in protest.

[zack]

I posted the same idea the other day.

http://forum.bitcoinhivemind.com/index.php/topic,259.msg1251.html#msg1251
February 02, 2016, 01:01:28 am

I think that technically this is a valid oracle, but it seems difficult to program.
I don't like the idea of people voting the wrong way for moral reasons. I am worried that a politically charged event could result in a blockchain fork. If people on opposite sides of the fork have contradictory morals.

[psztorc]

It is an interesting concept.

What if the 99% vote in highly random and confusing ways, drowning the "true 1%" in an ocean of similar-but-wrong ballots?

Fortunately, SVD can help a great deal...I even wrote code (for the "Audit Ballots") which greatly simplifies the problem, reducing Ballots to 5 maximally-representative teams. Traders could specify their "team", but I'm not sure how this would actually be coded in practice.

A second, more fundamental problem, is that it seems to reduce (as almost everything does) to "buying VTC"...instead of needing to buy 51%, I can just buy 1%, and then spend all of that remaining money (what would have been spent on the 50%) on transaction fees immediately following the fork. An attacker will strike at the weakest link...so if this is cheaper you have actually decreased security. However, it is possible that this will not be cheaper, and furthermore it is possible that this change will make the 51% route more expensive for the attacker (because it is now 51% + surviving this second hurdle).

It is hard to tell, isn't it? The transaction-fee-differential is really just a proof of work, as well, as one can trivially make trades that are non-representative (buying 1 of each state risks no money, but allows one to pay a transaction / trading fee...hence it just becomes a wealth contest that has nothing to do with "trades"...I proposed this awhile ago but have since abandoned it as being not worth it).

Also, miners can simply censor transactions which they do not like, so your way would allow a 51% miner coalition to buy 1% of all VTC and then control the outcomes (in my way, they can simply veto and force the voting to be redone).

In general, a principled approach will be most robust to attacks, vs a whack-a-mole security patches. For example, in the whitepaper I attempt to explain, why I feel that a second token can represent reputation (a kind of "InTrade stock price"), and why RBCR should happen using SVD. These considerations are based on my real-life experiences with trying to uncover the truth amid a room full of potential liars.

Nonetheless, nice try.