Eh I'm not going to go through your list, but putting libre/open office on there is disingenuous. Open office was run into the ground by Oracle, not because of the forking.

The first three all have the same story though, where Oracle acquired the project, forced changes that drove the original authors out, who then forked the project and resumed development there. So Hudson/OpenOfice/MySQL essentially lost the key developers that had all the institutional knowledge, killing them.

I do think there would be a lot to gain to merge vim/neovim, but they're in a very different position. I don't know enough about the other two forks to draw parallels to them.

LibreOffice is the mainline, not a fork in anything but the name.

There are also cases where the fork is caused by differences in philosophy and result in divergent codebases that are perfectly fine doing their own things, like NetBSD and OpenBSD.

The ones that "strangle off the original" are the successful cases. It means that users were given a choice, and they chose to use the forks, which were (for them) "better".

Without the forks, there would be no competition. Would the improvements have been included in the original project if there was no fork? Maybe. But usually forks happen because the intended improvements were either rejected or otherwise not accepted. If there was no fork, and the forked project did not "strangle" the original one, users would not have the chance/option to use the "better" one.

I think all you showed was that competition is good. I don't see how you would come to the conclusion that merging is better.

In a sense this is mostly a philosophical question of the Ship of Theseus kind. What happens is that when one side gains traction because of some specific feature, the other side either gets obsolete quickly, or they adopt the same features. So in the end everything converges in feature (and often even code), only not necessarily in name/leadership.

MariaDB didn't "strangle off" MySQL. The latter is still significantly more popular (probably 10 times as much, if not more) if you look outside our HN/Linux bubble. It also had one of the best releases ever under Oracle's guidance back in 2018, many years after the split, when very few of the original developers were still at Oracle. Meanwhile, MariaDB development slowed down significantly over the past few years as they're having financial problems.

(You might counter that by saying that the last MySQL release was in 2018, but it's not really comparable: Oracle does all of its development behind closed doors and then does major code dumps when it's ready. MariaDB prefer shipping small releases every 6 months or so.)

Vim was developed by Bram commiting and controlling every change. Neovim very much already unlocked a greater than one situation by adopting a different development model.