I was told that actuarial mathematics emerged in the 17th Century to calculate mortality so that insurance premiums could be calculated, and that the underlying idea was the “race to death.” In other words if you look at mortality as a race to death, some finish the race earlier than others, but undoubtedly everybody eventually finishes the race.
It occurred to me that we can think of the life of computer systems in a similar way-some die before they are even born (not approved for instance), but once approved mortality tends to be low-but not negligible (costs get out of hand, scope creep, new management) and once implemented computer systems, if properly supported, last for a while. Eventually new business practices, new technologies and even new organization types emerge that cause the system to be replaced, typically after 15 years in my experience.
This can probably be applied to many types of projects with long expected lifecycles.