There seem to be some projects that actively resent the idea of version numbers having meaning. Maybe they don't want to be "constrained" by what they can and can't do based on major/minor version updates, or maybe they just don't want to keep track of it all - I'm not really sure.
To me, version numbers are a way to communicate something about a project to users. "Oh, it's a patch release? That should be relatively small fixes and fairly safe. A minor release? Cool, some new features or significant changes to look at. A major release? The file format changed and the GUI got rewritten - going to need some eval/retraining on this one."
There are reasons for this rational, structured approach - when you have users in the Real World, they need time to prepare for major changes. Your software is probably doing Real Work, and cannot be simply yanked and upgraded without first ensuring that it will continue to do what it needs to do.
Developers may resent this, but it is an utterly inescapable reality. Critical tools cannot be casually changed - there MUST be a testing and validation period. Patch releases with minor/security fixes allow for relatively quick and simple deployment of truly essential changes without the major upheaval of EVERYTHING changing. Eventually you do need to make the jump to the next major upgrade, but surely Debian stable is proof positive that users need controlled, gradual change? Relativly stable periods between major changes are ESSENTIAL for a controlled computer environment, and it's hard to blame those who are saying the new Firefox approach is automatically disqualifying it from their networks.
Of course, there's also the point that developers with limited resources don't want to have to keep backporting code to older versions of browsers and work around issues that should have long since been restructured away (and have, in newer versions.) This is actually one of the better cases for commercial support of FLOSS - a company being paid to ensure older versions work can do the grunt work that the open source volunteers aren't going to want to spend time on.
Perhaps Mozilla could think about offering a paid service that maintains and supports particularl version numbers of browsers and have their "unversioned" open source browser with all the latest changes be the default, if this is a resource constraint issue?