Consider, for a moment, this edit:
http://en.wikipedia.org/w/index.php?title=The_Addams_Family_%28pinball%29&am...
It is a minor modification to our description of "The Addams Family" pinball machine. Because I happen to own one of those machines, I know that this edit is partly right but almost certainly partly wrong. However, I haven't played it much lately, so my first instinct was to commit the grievous sin of original research by playing a few games.
In thinking about this further, there are whole classes of article just like this one, full of uncited information that is probably original research. The unifying characteristics seem to be:
1. If the article is somewhat inaccurate, there is little risk of real-world harm, 2. The topic is of relatively low importance, 3. Having something on the topic is a net benefit to our readers, and 4. There is a wide enough base of people with knowledge of the topic that the article can generally be verified from collective personal experience.
Personally, I think these articles are worth keeping. Our readers get information they want. It also seems like a good place for newbies to contribute: it's a topic they are interested in, there is plenty for them to fix, and if they don't get it exactly right they won't immediately be reverted and slapped with a talk page notice containing eight links to policy shortcuts as they would on, say, [[Evolution]].
As far as I can tell, though, there is no written policy or guideline for this kind of thing. Is that the case? It's probably for the best, honestly, as they are doing fine without it, and I imagine creating a special exception for this kind of thing would lead to all sorts of disruptive wikilawyering.
Regardless, I thought it was interesting how much has been built in the outskirts of our metropolis. Not up to our building codes, but not a big problem, and better than nothing.
William