Personally, I think "too little, too late" has a ring of truth, depending on your perspective. But "better late than never" has a lot of truth too and you have to bear in mind that the political and legal climate must be considered. IIRC Castle have previously stated clearly that they were not willing to release source code, but times change and it would be madness to ignore this and cling on to some tired mantra in ignorance of the world around you. Of course, some people have complained that Castle have changed their mind on the open (strictly, shared) source issue; just as some were complaining when they didn't change their minds. That's life.
As for RISC OS turning into multiple little distros, well, that might happen. We're unlikely to ever have the number of developers that something like Linux has, but even with a smaller number, divergent releases are a possibility. As a community, we have to consider:
* Are multiple distros in fact a healthy sign of a varied developer community? Bear in mind the RISC OS philosophy of forwards and backwards compatible APIs and its modular nature. A "distro" containing one set of modules by default might just mean a set of "*RMensure" lines in an application and softloading what you need on a distro containing another set of modules by default. Not much different from loading shared libraries on any other OS really. Application developers seem to cope.
* Can the community, if changing APIs, can maintain the high quality (on average, with obvious exceptions!) of the Acorn-designed APIs in the existing OS? Well designed interfaces with graceful fallback (or fall-forward) can help mitigate application development problems. Many RISC OS developers share the design approach/philosophy that underpins parts of RISC OS so they do produce modules with well conceived interfaces. If the developer community expands, it will be up to existing developers to provide new developers with support and documentation to help them achieve what we believe is a robust and considered design, while recognising that everybody learns from everybody else all the time. If we instead choose to turn inwards and ignore one another then the results are likely to be grim.
* Does the community wish to submit changes and improvements back to ROOL for inclusion in the master tree to try and avoid fragmentation? If people believe that divergent code is the way forward then that's what they'll do. If people believe that convergence is essential, then we've done everything we can to provide the support mechanism through a neutral third party. The community makes its choice.