The Museum Guide Who Learned to Check the Drawer
The museum guide held up an old coin under the soft lights. A visitor raised a hand. "Who was leading this country that year?" Behind the guide, a little door led to a storage room full of drawers, each stuffed with note cards.
The guide could keep talking from memory and hope nobody noticed. A lot of text-writing machines have the same problem. They sound smooth, but the facts are sealed inside, so they can say something wrong with a straight face and you can’t see where it came from.
The twist was letting the smooth talker consult an outside memory, like that drawer room. The question is the visitor’s ask, the drawers are many short encyclopedia snippets, the clerk is the fast finder, and the guide is the fluent writer. Takeaway: it can look things up instead of guessing.
The smart part was teaching the clerk to grab useful cards without a person pointing to the right drawer each time. It tries a few cards, the guide answers, then the clerk gets better at picking next time based on how the answer turned out. The drawers stay put; the searching gets sharper.
They tried two ways for the guide to use the cards. One way, the guide picks one card and sticks with it for the whole answer. The other way, the guide keeps glancing at different cards as each short bit is written. Same drawer room, different pacing.
On tough question-and-answer jobs, the guide-with-a-clerk did better than the guide who only trusted memory. It landed on the right facts more often and made up fewer details. Sometimes it still got the answer even when no single card had the exact wording, by combining hints from several cards.
Back at the museum, a new box of note cards arrived with this year’s updates. Nobody had to reteach the guide a whole new tour. The drawers were swapped, and the guide kept the smooth voice while showing which drawer each detail came from. The difference felt simple, and huge.