Good question. I think if you look at where Livepeer’s protocol design is today in this early stage, it’s likely that you would see the same sort of cartel based behavior. Here are some characteristics that work against this however…
- Live video encoding distribution benefits from location awareness — it is important that nodes are physically located near broadcasters/viewers.
- Capabilities vary significantly across video encoding requests — do you want high end deterministic CPU encoding for a media platform, or do you want affordable GPU encoding for your garage security camera? Providers have the ability to compete on capability, as well as price.
I think in the situations you describe that are more susceptible to the cartel attacks, the work being performed is generic — any server anywhere can perform the same work, so it’s easy to gain a position in the protocol with enough stake and just sit on it. We would hope to reward those who are actually performing the work — via fee collection — and this will be a funciton for the demand based on price and services provided…allowing any upstart who can compete to earn their way in.
There are certainly protocol iterations to consider to enable this — offchain job negotiation based upon capability, shift from stake based assignment to more a masternode license-to-work model, etc. Livepeer has a long path to walk towards decentralization and a useful, scaled system, so we welcome ideas and participation!