In Principles, he describes some high-level decision making processes. Bridgewater is split into groups of about 300 people. Within those groups, smaller groups are formed to discuss particular strategies. Everybody has "baseball cards" of their coworkers with stats showing peer-rated talent in various sub-domains. When a decision is made, there are two rounds of voting: a raw vote, and a vote weighted by stats (e.g. somebody whose peers have rated them as knowledgeable about agricultural trades would have more votes in this phase). If the two rounds of voting don't align, they try to keep discussing the issue until they do align (though there are tie-breaker rules if deadlock is reached). I'm not saying this system is perfect, but it's interesting to me as a meta-strategy for evaluating potential strategies.