Gavin Joins The "51% Honesty Club"

[psztorc]

http://gavintech.blogspot.com/2014/06/bit-thereum.html

It must be mining.

Mining must be the origin of this 51% number. After all, the phrase "51% attack" has been in the public discourse frequently and recently. That must be it.

For some reason, many Bitcoin 2.0ers have been assuming that something called "honesty" exists and that 51% of people/oracles are it. Even the Great Skeptic, Gavin Andresen, has yielded to temptation. The sad truth is that "honesty" is a function of 1: "What do I gain from being dishonest?", and 2: "Can I get away with it?". So there will always be some juicy, irresistible carrot which presents itself at some crucial time, and forces people to realize that they are dishonest.

We even have sayings for this "Every man has his price", "Opportunity makes the thief", etc. Of course, we pretend to experience honesty, because it makes us seem more trustworthy ("all of the people I hang out with are honest" "honesty is normal for our group" "that guy rambling about dishonesty is some untrustworthy fringe character"). Even our thoughts about honesty are expressed dishonestly!

Bitcoin assumes that all miners will act in their own self-interest. Separately, it says that to attack the network you would need to, at some cost to yourself, in a moment of pure spite, use a huge percentage of highly-coordinated hashrate to disturb the block-ordering. I assume similarly for Truthcoin.

I think Gavin's idea would be vastly improved if you could easily prove that an oracle signed a "bad" outcome (to then punish him), ie, if there were a cost to signing a bad outcome. Reliance on signer-oracles reminds me of the "nothing at stake" problem in this way. Why not at least try to scam someone?

[zack]

His suggestions to make oracles lower-cost are interesting.
Instead of having english directions: "Check who won the basketball game", it could be a piece of code that automatically checks several locations on the internet.
This would allow the votecoin holders to verify decisions much more efficiently, if they do not have to read them.
It also protects the votecoin holders from risk, because everyone's software will give the same result. They will always have 100% consensus.

Perhaps 2 types of decisions should be available.
low cost decisions written in a forth with web-scrapping libraries built in, and verified by software.
high cost decisions written in english, and verified by humans.

I am very excited to be working on a project that even Gavin cannot comprehend.

[psztorc]

This would allow the votecoin holders to verify decisions much more efficiently, if they do not have to read them.

I think that Voters would still glance at them, to make sure their software was working correctly.

Your comments lead me to believe that you might find the conversation at the beginning of this thread, between Edmund Edgar (proprietor of very cool service RealityKeys) and gwern (the gwern), to be interesting: https://bitcointalk.org/index.php?topic=475054.0

I am very excited to be working on a project that even Gavin cannot comprehend.

I'd hardly say that. He is probably too busy to actually read most of the projects out there. Which I think is very wise because most are a waste of quality reading time. I'm sure he would comprehend it if he tried, and I think he'll try sometime soon.

[delulo]

What effectively is the difference between trusting third party oracles and pool operators? There is a difference in what you trust them with / what they can do if they collude.... But no one can say that Bitcoin is trustless...

[zack]

An oracle has much less inventive to be honest than the truthcoin voters.
If an oracles lied, everyone would fall for it, and the oracle could use this situation to earn money.
If a truthcoin voter lied, no one would fall for it, and the voter loses reputation for nothing.

[psztorc]

What effectively is the difference between trusting third party oracles and pool operators? There is a difference in what you trust them with / what they can do if they collude.... But no one can say that Bitcoin is trustless...

I can. We are using 'trustless' to mean "expect them only to act in their economic self-interest".

Its sophistry to say that "we are trusting that the laws of physics will continue to operate" or "we are trusting that we exist" or something stupid like that. Just because you can work the word "trust" into a sentence, doesn't mean that that sentence actually contains any ideas. Bitcoin miners are different from a mint or currency-issuing bank because the bank would make money (in this case, literally) if they broke their rules and fired up those printing presses.

When people say "trust me", they aren't saying "trust that I'll act selfishly and stab you in the back as soon as convenient", they're saying "I know I have every reason to stab you in the back, but please trust that I won't because I also care about X". This fine on a case-by-case basis, but is systemically unreliable because X's influence and context can change over time.

Oracles can make an above-market economic return by betraying user's trust, while pool operators cannot. Oracle-funds are stolen permanently, whereas a 51% is extraordinarily expensive and can barely do anything. Reverse a few 8 confirmation transactions, in exchange for losing your entire revenue stream? Is that a joke? The attack chain might not even succeed in becoming the longest, or the community could coordinate an emergency checkpoint. Theft-by-oracle would leave the oracle with still-valuable Bitcoins. 51% would leave the attackers with devalued Bitcoins. It doesn't even come close.

[delulo]

What effectively is the difference between trusting third party oracles and pool operators? There is a difference in what you trust them with / what they can do if they collude.... But no one can say that Bitcoin is trustless...

I can. We are using 'trustless' to mean "expect them only to act in their economic self-interest".

Its sophistry to say that "we are trusting that the laws of physics will continue to operate" or "we are trusting that we exist" or something stupid like that. Just because you can work the word "trust" into a sentence, doesn't mean that that sentence actually contains any ideas. Bitcoin miners are different from a mint or currency-issuing bank because the bank would make money (in this case, literally) if they broke their rules and fired up those printing presses.

When people say "trust me", they aren't saying "trust that I'll act selfishly and stab you in the back as soon as convenient", they're saying "I know I have every reason to stab you in the back, but please trust that I won't because I also care about X". This fine on a case-by-case basis, but is systemically unreliable because X's influence and context can change over time.

Oracles can make an above-market economic return by betraying user's trust, while pool operators cannot. Oracle-funds are stolen permanently, whereas a 51% is extraordinarily expensive and can barely do anything. Reverse a few 8 confirmation transactions, in exchange for losing your entire revenue stream? Is that a joke? The attack chain might not even succeed in becoming the longest, or the community could coordinate an emergency checkpoint. Theft-by-oracle would leave the oracle with still-valuable Bitcoins. 51% would leave the attackers with devalued Bitcoins. It doesn't even come close.

I agree that there are worse things you can do as an oracle but it might still make economic sense for the ghash pool operator (around 51% atm) to take the cash offered to him by a third party that either is a government that doesn't like Bitcoin or a third party that does some very heavy double spending.