On the necessity of REP - Couldn't ETH be used to buy registration tokens?

Question

Apologies for asking a very similar vein of question to the one discussed here: What makes REP tokens a critical component of Augur?

However, I am still slightly confused about the necessity of REP coin to make Augur function. For example, I could imagine a design as follows:

1. ETH is used to buy the registration token every reporting cycle.
2. The reporters then report on a subset of markets.
3. If an event is challenged, then a much wider number of reporters (potentially all of them) are asked to report on this market.
4. The market then resolves itself to the majority decision, paying ETH back to the reporters who made the 'correct' decision, and I guess say redistributing ETH from the 'wrong' reporters to the 'correct' reporters.

To me the fallback of a hardfork in REP seems like a slightly overkill option to solve this problem. Presumably under the current design you would only hope that a hardfork of REP happens rarely (> every 2 months?) anyway?

Clearly the option of a hardfork adds some utility to the system; I am just concerned that this benefit does not outweigh the complexity of adding a whole new token.

Please tell me where my logic is wrong!

Answer 1

Registration tokens are actually being removed. The dev team has never been particularly fond of the registration requirement, and just the other day (late last week/this past weekend) they came up with a novel solution to the set of problems the registration token was aiming to solve. This change significantly simplifies the system for end-users I believe.

Regarding the forking system, the hope/design is such that forks should never occur. That being said, they probably will occur at some point but they should be as frequent as Ethereum/Bitcoin contentious forks (e.g., ETC, Bitcoin Cash, etc.).

The reason we can't remove the forking system is because we have no way of authoritatively saying "X is the truth" in the case of an attack. The forking system allows us to let the market decide what universe it wants to live in, and therefore assign value to REP in that universe and assign no value to REP in all other universes. This is what allows us to punish a 51% attacker such that even if they do acquire 51% of all REP and resolve markets in a lie, they are punished so severely that they can't profit from the attack.

If you just assert that the majority of reporters are always right, then that results in a Keynesian Beauty Contest where "Truth" doesn't actually matter, only consensus. Since we want Augur to be an Oracle for truths, not consensus, we can't rely on the majority necessarily being right. Also, a Keynesian Beauty Contest would not allow us to punish the attacker more severely than they can profit from the attack, making 51% attacks profitable.

You are correct that the forking system adds a lot of complexity. It is probably the most complex piece of Augur. However, without the forking system we would either have a trusted Oracle (like Gnosis) or we would have a Keynesian Beauty Contest that is susceptible to profitable attacks.