Sneaky algorithm

Well, (and take this for what its worth because I don't use SQL much at work) I do think that a query that omits "hidden" results would be easy, just as you say. Just leave them out of the query. But what was mentioned earlier in the thread was associating each item with its own unique weight to ensure that some things come up less often than others.  I suppose you could query for maybe twice as many items as you need and then roll the weighted dice for each result in the list (backfilling with items 25+ as needed when an item doesn't make the cut), but that's not particularly efficient because you're getting way more data than you need. Not noticeably slower. Just less efficient.

Whatever methods may be out there to do it, my own opinion is that it's not happening. And believe me -- I've had moments where I've hit that refresh button for the 20th time and NOT A SINGLE CHEESECAKE and my resolve starts to fade.  But then I'll be badly in need of patio furniture and I'll see it pop up on 3 out of 4 refreshes. 🙂

Every other point you've made, I agree with.  And that's interesting about how quickly items are added to the store -- I have to wonder, with that many going up for sale, how many people are there to buy?  I'd imagine in the thousands browsing at any given time.