I don't agree that it really affects need for system mastery--unless you are talking about the rule as it currently stands as well as the adjustment I favor. If that's the case, then yes, I agree that using ability scores for multiclassing prerequisites at all isn't the best solution. My suggestion was a bandaid. I've considered the point buy concerns, but I don't think it's really that big of a deal*. If the DM is handing out such high point buy that ability score prereqs are meaningless, then he is unlikely to be concerned about multiclassing being overpowered. It is true however that the phenomenon you described could happen.
Tangentially, I think the adventurer's midlife crisis has always been exceedingly rare--yes, so rare that it doesn't need to be overly considered. Back in AD&D the ability prerequisites for dual-classing were so high (15 in prime req of current class, 17 in prime req of next class--this with randomly rolled stats and no normal stat advancement) that if you rolled high enough stats to have a character capable of dual-classing you were aware of it from the start. In 3e you tended to plan ahead. In either version if you really wanted to play a different character, you generally just made a new character. Especially in 3e, you'd generally be gimping yourself to make an unplanned class change.
* - Funny how we decide these things. I can find myself on exactly the other side of this sort of disagreement, claiming that something of similar practical effect does matter very much, depending on the issue.