"Or can it be other reasons to use older kernels?"
Yes, I suppose there could be any number of reasons. One I can think of is
the radeon kernel module. On newer kernel versions the module is known to
fall over and die when the proprietary junk is not present, resulting in the
computer using VESA and getting a lower screen resolution. I know of a few
people that have stuck with older versions in order to avoid this.
This could also be an example of a regression, and could also be a reason
that someone would use an LTS or ETS kernel because the version that worked
for them (3.2) was an LTS kernel. People that stayed on the older LTS version
never experienced that problem, while those that moved on to ever-newer
versions were bitten by the bug, which still exists to this day as far as I
know.
"is it logical to say that it's recommended to use the newest one then?"
Not necessarily. What might be "recommended" depends on someone's priorities.
This circles back to the features vs. stability problem. This could be
compared to using the Testing version of some GNU/Linux distribution versus
the Stable version. Testing is newer. Stable has the version numbers of the
packages frozen on release and so will get older with time. It gets bug and
security fixes only. Which is better? Which is "recommended"? That's a
decision someone will have to make on their own about what they find to be
more important to them.
