The conclusion is slightly defeatist, but ultimately correct. At time 49:23, Casey says "But we have to do them right now, because we haven't figured out how to do it better."
[excessive modularity] is the worst form of [software development] – except for all the others that have been tried.
Perhaps it's a desperate attempt to defer the problem of what the organization should look like. The more nodes you have in your graph, the easier it will be to collapse it into something that maps to the organization you want?
Sometimes it feels like a desperate attempt to defer working on some part of the problem a team doesn't know how to solve. Usually some domain specific thing no one wants to think about. So onion layers get put in wrapped around that bit without actually solving it, just creating an abstraction by which the rest of the system will use the "magic". But often the abstraction overconstrains the domain specific magic through ignorance, so it all has to change when someone gets around to adding it. Seen this over and over.
The conclusion is slightly defeatist, but ultimately correct. At time 49:23, Casey says "But we have to do them right now, because we haven't figured out how to do it better."
[excessive modularity] is the worst form of [software development] – except for all the others that have been tried.