IMHO, the 'A' and 'B' groups are not 100% mutually exclusive, and the frontier between the two is fuzzy. This is why it is so hard to tell if "rule XYZ" should belong to "standard/CLL/whatever Lojban", or be a fork.
Still IMHO, I may recap a few pros/cons:
- Looking forward for improvements may:
- ... be a never ending task and create new forks "forever". It may be thus considered as an experiment (or a group of experiments), not a "spokable" language.
- ... make people drop learning it, as no one could ever learn it in an "almost stable" way.
- ... split the community, thus lowering the interest for the experiment/language.
- Keeping backward compatibility may:
- ... prevent from fixing unforeseen issues.
- ... also make people drop learning it, as some (potentially important) flaws would not be fixed.
So what? As I previously said, it seems quite obvious that neither extreme choice is possible. I'm closer than A (backward-compatibility / CLL, as gejyspa and Thimothy) than B. But I just can't believe no breaking changes will happen. As far as I know, even the widely-accepted "xorlo" is somewhat breaking, as it redefines a few things. I guess we might find a "90% A, 10% B"-kind of profile. But that will not be up to me to decide! Actually I think that no one could impose his/her point of view to others, despite great efforts and involvement of some.
Furthermore (and still as previously stated), the most important thing is that we need to make the number of Lojbanists increase, and to avoid splitting/schisms that would blast off the efforts of all.
Therefore, we need consensus. This is why I'm still putting efforts on the "submission platform" I told before. It takes much more time than expected (I naively thought I could find out-the-box modules for my usage...), unfortunately. But I'm still on it.
My 2 cents...
la .sykyndyr.