Thanks for your comment, and I think those are valid points. I won’t speak for the company, but I can give you my interpretation based on the whitepaper. Note that in order to keep my post concise, I simplified parts of their framework. I’ll expand a bit on it here:
- Reputation “points” and colony tokens are both taken into account when evaluating a contributor’s reputation for interacting with the colony and other users. They are awarded on an individual task basis. As the colony’s objectives change over time, the tasks and projects will change as well, meaning people will only be rewarded with reputation by completing tasks that contribute to the current goals. Over time reputation is automatically depleted by the CLNY miner calculations, which keeps people who contributed in the past but have been dormant for a while from returning to throw their weight around and change current objectives. Colonies also have a task proposal queue, where all tasks created by managers are “funded” with tokens and rep. rewards. This is built as a permissive system where simple tasks will be automatically funded unless a colony member raises a dispute. Larger proposals must be voted on. This system makes sure that everyone is aligned on the colony’s direction. Tasks & projects that people don’t agree with won’t be funded. I think the combination of these factors will help to make sure meritocratic rewards match changing objectives.
- I agree that merit is not a unidimensional variable. The Colony team has addressed part of this with the domain system. Let’s say a contributor is completing tasks in the Marketing domain. She will gain reputation points + tokens (referred to from here as rep. to keep it simple) for completing those tasks. That rep. is valid in the Marketing domain and all parent domains. This means she has more meritocratic “power” in the Marketing domain and the top level colony domain. However, she doesn’t have that same level of power in the Accounting domain, which is parallel to Marketing. This could get even more granular by splitting the Marketing domain into different types of Marketing, so a person working in the Social Media sub-domain would not get rep. gain in the parallel Print Ads sub-domain. This helps keep rep. relative to the context in which a person contributes.
But I disagree that merit/rep. is not a good predictor of future action. In a well-constructed colony, a person will only gain a significant amount of rep. by contributing regularly over an extended period of time. At this point, the person has proven repeatedly that they are a reliable employee. They are not only invested emotionally due to the personal time put into a project, they are invested economically due to the tokens they’ve received and the reputation points (influence) they’ve gained. Reliable history of contribution + emotional investment + economic incentive should be enough to assume that a rational person will continue to contribute. Therefore it’s reasonable to give them influence in the colony, which is the whole reason you’d want to worry about predictive validity in the first place.
Let me know your thoughts on that. I’m also tagging Jack du Rose to correct or expand upon anything I’ve said here.