OK, so we have two cases here:
1) gnome-software as is currently in F23 and F24
2) gnome-software future releases
For (1), the version of gnome-software in F23 and F24 currently doesn't
have the UI in place to allow choosing which version to upgrade to, so
gnome-software needs to somehow automatically pick the version. My idea
here was patch F23 and F24 gnome-softwares so that they always offer the
latest non-development version, bounding it to N+2.
Example: A user has F23 installed. F24 comes out. gnome-software offers
to upgrade to F24. 6 months later F25 comes out, gnome-software switches
to offering a F23 to F25 upgrade. 3 years later when F23 and F25 are
both EOL, F23 gnome-software still continues to offer the F25 upgrade.
As for (2), I guess we should do the same as (1) but just allow the user
to choose the N+1 release as well in addition to N+2.
This sounds as the best option to me.
If the user hasn't upgraded to N+1 in 6 months, it's pretty clear she does not
want to upgrade too frequently, quite the opposite. So why should we recommend here just
N+1 upgrade, which doesn't decrease the number of needed upgrades at all? In another 6
months she will be forced to upgrade again. Rather, let's suggest N+2 upgrade, which
cuts the number of needed upgrades in half. (As for upgrade safety, this is very hard to
measure, but I believe both methods have a very similar success rate). Of course, having a
option to choose the target release in gnome-software would definitely help.