My point is that reader_utility is more or less the same when the story is posted here, as it would be if it was posted in the Open Thread and Media Thread, but that nonreader_utility would be significantly better if it was posted there rather than here.

I don't follow. A viewer is a viewer whether they are viewing from an open thread or discussion. Posting in Open Thread vs Discussion only has the effect of scaling viewer numbers (in this model, num_nonreaders & num_readers) up or down. And I would expect that the num_nonreaders to num_readers ratio is the same for Discussion readers vs Open Thread readers. (BTW note that num_readers refers to the fact that the people have read Flowers for Algernon, etc... sorry for the confusingly named variables. Feel free to refer back to the grandparent comment.)

In fact, in this toy model the open thread vs discussion vs main distinction is fairly pointless: if the expected utility of a given person seeing a given post is positive, we should give it as much exposure as possible in order to harvest as much utility is possible. And if the expected utility is negative it should not be posted anywhere.

To the extent the toy model is wrong, it's wrong because people have a limited amount of time when they read LW and we would like to put the highest expected utility stuff in places where they are more likely to see it. If everything was posted to Discussion, finding higher-value stuff might be harder. But if there isn't already stuff competing for attention in main/discussion, little is lost that I can see by upgrading open thread posts to discussion posts or discussion posts to main posts.