Re: Not quite 31 years
The stable trees as they currently exist have been a feature for quite a while, Wikipedia suggests since 2005...
There's a fine line to be trodden when considering updates to a "stable" kernel, or probably any other software to be honest. Too few updates and it becomes a security hole riddled liability. Too many and it's no longer something software written when an earlier point release was current can run on.
The rules for what can be included in a stable kernel are well defined as it's the process for having changes considered:
My gut feeling is that part of the increase in patches being applied to the stable kernel is due to there being more focus on using the stable kernels now than there may have been in the past (which shows it's achieving its goals I suppose and more effort made available to work on them as a result) and there's probably an element of the seemingly never ending increase in kernel contributions resulting in more potential for both bugs and the generation of fixes.