Well there is “exploitable in theory” and “exploitable in practice”. For example we can see that both Bitcoin and Ethereum are vulnerable to cartel formation.
- In Bitcoin you can do a 51% attack by rejecting all blocks outside of the cartel.
- In Ethereum you can do a 66% attack by not including attestations of validators outside of the cartel.
Why don’t we see that in practice:
- Majority is actually honest (thing which is easy to forget in game theory).
- Coordination at this scale is hard. Particularly in Ethereum where validators need other people ETH, trying to start a cartel basically tells the whole world “I am a dishonest actor”, which would likely lead to stakers changing operator.
- Cartels may not be stable, in a world of cartelisation, you are not guaranteed to be part of the cartel.
- If cartelisation were to happen, this would be seen as an attack, greatly tanking the price.
- There is always the social layer which can fork out the cartel (very true for Ethereum, a bit more complicated for Bitcoin).
In addition in this proposal there is an upper limit (10%) of what can be redirected, so a cartel could at maximum increase its yield by 10%.
- This means that the upside is very limited compared to all the previously cited downsides.
- The gains in term of ETH price increase with proper funding should be higher than the stolen yield. Indeed, if we had a cartel capable of 51% attacking the system, we would actually end up with a group of coordinated actors which would have an incentive to direct part of the stolen yield to ecosystem funding.
I do concede that with proposal we would go from:
- With 51%, we can break the system (by messing up with the gas limit).
- With 66%, we can steal other people yield and principal.
To
- With 51%, we can break the system (by messing up with the gas limit).
- With 51%, we can steal 10% of other people yield.
- With 66%, we can steal other people yield and principal.
Because we already have possibilities or cartelisation in theory, but never saw those in practice, I estimate this risk to be mostly theoretical (<1% and I am ready to take such a position in a conditional prediction market).