Ray Saintonge wrote:
A dictionary chronicles the language in both its past and its present. Its past needs to be subject to calls for evidence; if a word is challenged the burden of proof for verifying its legitimacy needs to fall upon the contributor. Otherwise, the rest of us are left with the futile task of proving a negative. Evidence for new words is even more important. It is not enough to say that the word was used in some unspecified episode of a TV series. What amuses the members of today's peanut gallery may be completely forgotten by this time next year when the forces of marketing will have diverted our attention to some new ephemeral fantasy. Web evidence does no better. It is not good to accept any word as valid irregardless (sic!) of where you found it.
Disclaimer: I'm a moderate descriptivist, whereas it seems you might be at least a moderate prescriptivist (see [[en:prescription and description]]).
I think we ought to document neologisms if they have been used by any high-profile source (any major author, for example), or if they are used by any verifiable subculture. I do agree that if I coin a word and my friends use it, that doesn't count, so there has to be a judgment call somewhere. A widespread neologism with specific connotations, like the phrase "teh sukc", ought to be documented, though. Wikipedia's always has as one of its strengths that it gets articles on new concepts before almost anyone else, so it'd be a shame if Wiktionary didn't have similar advantages.
They can of course be discussed neutrally--mention if they're in other major dictionaries or not, who uses them and to what extent, etc. But we're not the language police...
-Mark