the open/closed dichotomy is not necessarily the leading factor in why a project succeeds or fails. As the former community manager of Diaspora, I would say from experience that the leading causes involved lack of an API, lack of available apps, a detached core team whose popularity and expectations exceeded the possible rate of development, a fixed amount of one-time funding and (later) a community-led development that moved so slowly and was so change-resistant that many long-time contributors and users moved on. The original core team got so wrapped up in trying to find funding and appeal to a wider audience that certain important things never got finished.

Diaspora was quite popular in its heyday, and its federation design is ostensibly good. It had a healthy community for a long time, and the privacy scopes were better than what we have now.

I think you're right: Success is the exception and it's worthwhile to investigate those rare cases.

I have a different approach, I believe each success is unique because the combination of factors that produced it, so studying successes will at most enable you to reproduce the same successes with a negligible probability. I believe studying failures can yield more actionable information because the corpus is bigger, and it's often easier to identify the few reasons for failures and summarize a list of common pitfalls to avoid.

This enables you to keep an original vision that doesn't depend on past successes that you will have trouble replicating anyway because many success factors are time-based. You can easily be too early or too late, all other factors being equal.