The argument that difficulty will tend toward zero in the absence of a block reward is something I've brought up in the past.
One thing that I had not considered at the time was the effect of merged mining, since I was spouting off this line before merged mining ever existed.
I think that to understand this issue, it's significant to consider that if Bitcoin really does end up in a situation where its difficulty tends to zero due to a lack of economic incentive to mine, something somewhere else is likely to not have that problem (inflatacoin? litecoin?). Someone's always going to be mining something. Bitcoin can be secured with merged mining the same way Namecoin benefits from Bitcoin mining.
Of course, it would be a huge OUCH to Bitcoin (specifically, the legitimacy of the main chain as we know it) for it to have to take a back seat to some other alt chain just to survive. But at least it would remain secure.
I can't imagine Bitcoin becoming a second fiddle to an alt chain far into the future ...
I think the process of discovering whether Bitcoin should be the world's virtual currency, or whether some alt chain is better equipped for this, will take a handful of years, not 40. We will discover "the best" crypto currency, and just go with that.
One of the key selling points of Bitcoins, at least these days, is its fixed money supply.
If the best argument we can find about why Bitcoin can survive well into the future is merged mining, then this undermines this argument, and makes investing into Bitcoin right now now viable. (Why would people mine the alt chain ... it will too become dependent upon yet another alt chain later on). This is sort of equivalent to removing the limit of 21,000,000 coins (which might not be such a terrible idea, when you stop to think about it). If this is the case, a Bitcoin2 without this limit can be started right away, and if it's really a better idea than the market (miners + users + shop owners) will flock there ... if it's not needed, it will dwindle like all the other alts.
Nah, I think other methods can be devised to keep Bitcoin alive, perhaps with some protocol modifications as suggested by Mike Hearn.