As Pterry says, the natural size of a coven is one. Any project with two or more people involved will have at least as many opinions, and probably more where these are reasonably intelligent people and the subject is complex, as software projects often are. Operating system design is one of these.
It's not that mutual self-interest can't bring people together effectively. For instance, it's very easy to forget that for international economies – although some companies take a "we'll charge what we can get away with and mark down for certain markets" view – are still often a case of the choice being piracy or free software. This has a positive effect on and is a big driver for sharing code and open source movements.
But the broader factor for many is control – people like having their computer do what they want and tell it to, rather than just getting what's forced by Microsoft, Apple, or others. Linux and to an extent Android make it easier to get away from some of that loss of control, but there can still be controversies when a new piece of software becomes a big player and starts "taking over" from the old in most distributions, or has options taken away people were relying on to control their computing experience.
A minor example is the default update notifier in Ubuntu and Xubuntu losing its tray icon and working by popping up a window over whatever the user is working on at the time, with certain types claiming straightfacedly that this is better for usability.
Of course, this wouldn't be an issue had the option for the icon been retained for systems that have a notification tray, and patches offered to do that been repeatedly ignored. That's one of the more noticeable downsides open source shares with closed development – unless someone is in a position to fork, maintain parallel development and in a best case scenario supplant the bad practice, things can still definitely get worse as well as better.
(Xfce actually has a preference under Settings > Window Manager Tweaks > Focus called "Activate focus stealing prevention" that I've turned on to see if helps).
The current bête noire of the Linux world in this respect is Systemd, which replaces an older sysvinit system and essentially acts as a core process through which other software starts. It therefore touches everything, so design or security issues have a big impact. Most distros use it by this point, particularly those branching from Debian.
As a simple example, running processes in parallel during system startup has both pros and cons: it can speed things up but if done in a way that doesn't make sure process B – that relies on process A having done something – starts after, can lead to failures.
For a more general overview, you could wander around here:
It particularly ought to be kept in focus that systemd development is backed by Red Hat. if they disagreed with its direction or wanted to put more resource behind testing and bug-fixing they have a lot of clout. If they weren't okay with the lead developer being a dick in responding to concerns, they're in a position to help reign that in.
Authoritarian project control isn't necessarily a bad thing but it does require consistently sound judgement, and unless you're in a position to be king or bask in the reflected glory of one is better used to create meritocracies (even if they're shark pits) than fiefdoms.
Some recent issues have included:
All of which still isn't a reason to jump ship for the average user – nowhere is perfect, after all – but it is a regular reminder that although levels of control are far more advanced when you can replace a desktop environment, file browser, etc simply, there are still ego battles over the direction of core technologies that aren't easy to avoid getting caught up in.