The way I see it:

* tail emission of a fixed number of bitcoin trends towards 0 inflation and is therefore not different in the long run from a fixed supply.
* tail emission as a fixed percent of bitcoin is forever inflation. This does have an effect, but as noted by many in the "Bitcoin covenants are inevitable" email thread on the dev mailing list (which went off the rails as a discussion of blockchain security and targeting/funding of that security), doing this does increase the amount of security (all other things kept equal). However, choosing an arbitrary forever-inflation-rate does not ensure that blockchain security will always be sufficiently funded, unless we can find some upper bound on required blockchain security (in terms of percentage of total coin supply).
* Some have mentioned that people lose coins. I don't think this fact is worth considering - ie its not significant. The future rate of lost coins will decrease towards 0, and is likely already nearly 0 today.

If for some reason we determine that we *can't* sufficiently fund miners with fees alone (which I think is a very real, but unlikely possibility), then adding a percentage-based tail emission would be basically our only option. However, I agree with most that until we find that to be the case, its likely a very dangerous thing to attempt to do to bitcoin. Even tho there are reasons of economic efficiency for holders to pay some of the cost of bitcoin's security (via inflation) in addition to transactors (via fees).