And the Results (Patreon)

OK, the majority of you were "right" -- the map thumbnail actually won pretty impressively -- the 7-point spread is actually huge for the first hour when most of the viewers are die-hards who will click regardless. (YouTube will end the test and declare a winner as soon as it hits a statistically significant result, which I assume is 95% confidence interval.) I actually agreed with the majority here, which is why I used the "map" thumbnail in my social media promos a few hours ago.

Anyway, I put "right" in quotations, because the people I REALLY want to run the A/B test on are the people who aren't watching my channel yet, and that isn't the population that's seeing the thumbnail in the first hour after upload. The problem is, that's a much smaller sample size, and I can't tell when THOSE people are seeing my thumbnail...it just depends on the hour-to-hour machinations of the algo, which are opaque to me for the most part.

So there's actually lots of nuance to this! Also, people see the thumbnails at all kinds of different sizes because people increasingly consume YouTube on phones as opposed to desktop...but also, consumption via the YouTube app on smart TVs is growing. You really have to account for all this somehow, with just one thumbnail. It isn't easy! (I have an app that shows the thumbnails at all sizes so I can run a check.)

Anyway, this stuff is fascinating to me (and critical to the health of my channel), so I'm always happy to engage conversation about it, or other YouTube inside baseball-type stuff. Thanks for the feedback!