IMHO the fact that "mocha" got the highest engagement also shows that it's the most easy word to type, as opposed to, say, "macchiato"
Also, didn't anyone at CSAIL write a simple bot to try all the coffee types (eg the list here - http://en.wikipedia.org/wiki/List_of_coffee_drinks)? Or perhaps the most prolific link follower did but just disguised it well??
In order to rule out _simple_ guesses you could have used a code after the coffee name - "/mocha64", "/espresso42", etc.. Without that any analysis of the geographic preference is immediately muddied. Could have chalked the entire board as a giant QR code .. obviously you'll need to design a chalk board plotter first.
It all boils down to the fact the you "didn’t really have any concrete hypotheses".
However it's interesting. It would be great to find out the conditions in which people will be most engaged in looking for the next info. But there would be sooo many variables to control ...
There are so many variables to control, things that I hadn't thought about before writing the URLs. The guessable URL part was perhaps one that I should have caught, but there were so many other things. I know that people did things like taking pictures of the boards and sending them to social group chats that they were part of, which is something I never thought of before seeing/hearing about it happening on Friday.
Even the most prolific link follower looked at only ~150 links, and looking at timestamp information and the content of the guesses themselves, it doesn't seem like it was a bot.