Clay Shirky's In Praise of Evolvable Systems is an excellent explanation of why weak "gluing" protocols that offer integration (think HTTP) invariably succeed over strong and tight protocols that do not (think ), even if the latter have enormous amount of money behind them. The reason: tremendous pressure for integration and freedom to innovate comes from end users. The argument generalizes to communication systems in general, and from there to scholarly communication systems: see the excellent text The Future of the Electronic Scientific Literature.
