Well, it turns out that some people aren't joking in endorsing this suggestion:

Instead of creating a new word to represent someone who is receiving guidance under a mentor as a 'mentee', couldn't someone (not certain of who is responsible for adding/changing definitions to the official dictionaries) simply add an additional definition to the word protege to allow for further meaning?

Recall the context: On the "mentee" thread, some commenters suggested that "protege" was an adequate substitute for "mentee." Others pointed out that "protege" tends to have a different meaning (a protege gets patronage and support, while mentees tend to just get advice). Then came the above quote.

Wow. So here we have a word ("mentee") that has been around for several decades, that is common enough to be listed in several leading dictionaries, and that is easily understandable (especially in context), both because it's not uncommon and because it fits a common pattern of English word formation. Now I'm not saying comprehensibility is enough; "udnerstadnable" is probably understandable, too, but I'm not advising you to use it. But surely comprehensibility is pretty important.

But some people disapprove of "mentee," whether because it's a back formation that doesn't correspond to an actual verb "to ment," or because they just think it's ugly. So instead, the suggestion is to add a definition to the dictionary.

What exactly do people think will happen when this definition is added? Will it, like domain name registry entries, get automatically propagated over the following 48 hours through the minds of English speakers? Will there be public service announcements on TV telling people, "Please remember that starting 2 am Sunday, the word 'protege' will also mean someone who gets merely advice and not patronage or support"?

No: The same people who today hear "protege" and think not just someone who is being mentored, but rather someone who is being politically backed, will keep on thinking this. Perhaps over time, some of them will look up the word in the dictionary, learn the new definition, and slowly spread the definition to listeners (in contexts in which the listeners will recognize the new definition, rather than just being confused). Perhaps, and only over many years. In the meantime, the extra definition in the dictionary will in no way affect what the word actually communicates to listeners. The new meaning will not be properly comprehended.

On top of that, imagine what would happen even if the suggestion worked: We'd take a word that usually has a moderately crisp definition ("a person under the patronage, protection, or care of someone interested in his or her career or welfare"), and add to it another, materially different definition (a person who is simply receiving advice) -- thus making the word ambiguous (or at least more ambiguous). Now sometimes words do acquire new meanings that make them ambiguous, and often there's not much to be done about that. But do we really want to deliberately create extra ambiguity? Is an ambiguous "protege" really better than an unambiguous "protege" plus an unambiguous (even if ugly-sounding to some, or etymologically impure to some) "mentee," to the point that we should deliberately choose making "protege" ambiguous?

Third, and this returns in some measure to the comprehensibility point, authors of dictionary have a certain professional responsibility to readers. They should inform the readers of what a word in fact means when English speakers generally use it. They could, if they want to, inform the readers of what meanings are socially condemned, or even what meanings the dictionary authors think are in some sense "the best." But can it be right for them just to add a meaning that English speakers don't generally use, simply on the theory that English would somehow be a more elegant language if English speakers did use such a meaning?

I think that if we read a dictionary and learned a definition that, it turned out, was just the authors' own pet project rather than an actual current meaning of the word, we'd rightly feel duped. The dictionary would have made us less likely to communicate effectively rather than more.

Finally, I think all this illustrates a broader point about words and dictionaries. Dictionaries are not the language; they are useful snapshots of the language. Prescriptivists might argue that dictionaries should condemn certain aspects of the language. And indeed dictionaries do in some measure mold the language. But no-one, prescriptivist or descriptivist, should assume that the language will change just because a definition is added to the dictionary, and no-one should accept a dictionary that simply invents new definitions that the authors think might be useful. Drawing a new street on a map won't actually change the city. Adding a new definition to the dictionary won't actually change the language.