But in this case, the software works as it did, provided that you give it the surroundings for which it was made. The change needed isn't one of the four classes you point out, but rather a fifth:
- support for new networking technology
For the N95, a new web stack, presumably including xhr, webrtc and the other goodies so many apps use nowadays. For the Playstation, the ability to connect to new WLANs.
I suspect that new hardware would be needed in both of these cases. Running webrtc on a CPU from 2007 sounds dubious, and didn't 802.11foo change something extremely low-level wrt. hardware?
In those cases, you might truly want no updates of any kind, with which i agree - updates should always be compatible with the "opt out" approach, or even do "opt in" (like Debian has an unattended upgrades package, for example: https://wiki.debian.org/UnattendedUpgrades).
That said, due to the world that we live in, it's risky to forgo any sorts of security updates, especially in the context of networked devices and systems!
So the premise that we could remove security issues but not touch functionality seems utopian.
It's probably a bit chaotic, but some of the arguments that i made were that "updates" is too broad a term to be meaningful, since you might be interested in security updates without any other sorts of breaking changes being forced upon you:
Essentially, you might want stable and secure software and not much more, since if it was enough for you at a certain point in time, then it might be the same way in 10 years, instead of needing new features that break your logic numerous times over that period.But perhaps the current situation that we're in is simply a reflection of the current industry and market conditions, where companies try to have their products remain competitive.
You can't sell stable software (or hardware) that has a new feature once every 10 years, when there are products out there that release new stuff monthly - at least when it comes to the layperson, otherwise we wouldn't have enterprises like Google or Facebook make significant UI/UX changes, since none of those make sense from a purely engineering oriented point of view.
Of course, that's just my interpretation after applying Occam's razor - for some reason we don't hear much about extremely stable products as success stories.