At http://lists.linuxfoundation.org/pipermail/bitcoin-dev/2015-September/011309.html Gregory wrote: "Security requires a bit of vigilance, inherently." and
[A non-upgraded miner will end up] "> producing invalid blocks forever until the owner shuts it down and upgrades. This is the outcome guaranteed for absentee miners with a hard fork, but it is not guaranteed for a soft fork."

It seems that the main benefit of a soft-fork is that it allows participants on the network to keep participating even if they aren't vigilant enough to notice and upgrade when that is safest.  Are there other reasons that might entice me if that one by itself is not enough?

Gregory provided two more: [Using soft-forks] "radically lowers (in most of our experience and
opinion) the cost of deployment; again-- making them different. They prevent a industry wide flag day, and tight release synchronization  which is harmful to decentralization promoting software diversity."

I understand these benefits.  The cost in complexity is still too high for me, and I think most of the pain in "cost of deployment", "industry-wide flag days," and "tight release synchronization," as well as the centralizing effect of those things can be minimized with waiting periods.  The promotion of software diversity offered by soft-forks is pretty cool, but that gets close to messing with fungibility.