Do designers know how good their games are?

How well do designers understand how good their games are, or how well they will sell (which is two somewhat different things)?

Reiner Knizia (over 200 games published) recently said that his game Pickomino has quite surprised him. Pickomino is a dice game with scoring tiles that's fairly simple, but has become a big seller.

In my own case, 25 or so years ago, after I had had some games published, I had four that I felt were "my best". One of those was Britannia, and it *has* turned out to be a "classic". The other three have not been published. I've revised one quite a bit, that a new publisher appears to be interested in, another has been played a few times recently but I haven't submitted it anywhere, and the fourth is "old fashioned" in a way that might not work well nowadays, but it's only been played once or twice in 25 years. I don't work on the three much, because I have several games designed recently that I feel are better. But do I really know? No. Only repeated playing by lots of groups can tell you how good the game is, and even that won't tell you how well it will sell, as there are many factors in selling other than the quality of gameplay of the game.