The BoLD paper is awesome! Permissionlessness is crucial for validation. I was wondering how many stakes the defenders can be forced to submit in the multi-level setting.
From the paper:
3.4.5 Staking. A party who creates a level-zero edge must submit a stake S which they should expect to lose if their edge is incorrect. If there are L level-zero edges, the total stakes will be LS.
When a lower challenge is started, any party can create a level-zero edge in the lower challenge
Within each challenge or sub-challenge, the basic protocol is executed as described in the previous section.
I understand this as meaning that for each sub-challenge a party wishes to create a top-level edge, they’d need to submit another stake. This makes sense to me, otherwise I see the risk of a Sybil attack.
With this in mind, considering a two-level challenge, what would happen if the attackers create X different top-level edges at the first level? Would it force X level-two sub-challenges, in which defenders would have to submit another stake for each sub-challenges, for a total of X new stakes?
Here’s a step by step of what I’m trying to describe.
Defender: creates the honest top-level edge in level-one challenge, submitting 1 stake.
Attacker: creates X dishonest top-level edges in level-one challenge, submitting X stakes.
Defenders and attackers: play the standard game by creating further child edges, “chasing” each other in parallel. My understanding is that both attacker and defender would end up submitting X * log2 (steps) edges. At this point, I believe there’ll be a total of X one-step forks in the challenge, starting X sub-challenges.
Is the next step for the defender to create X honest top-level edges, one in each new sub-challenge, submitting one new stake for each?