I got poured a pint by a newbie behind the bar at a hotel recently and she looked embarrassed as it was about 40% head, but to her credit she went to fetch the shift supervisor before I said anything.

He explained after pouring it better that, even the remaining head (It had ~3/4 inch even after fixing it) might still be met by derision by many customers. "They'd be asking if you would be charging them for just for the half" etc.

There's a bit of leeway but you'll quickly hear about it if you short a pint too much.

I wonder how much accuracy would be improved if expanding from single words to the most common pairs or n-tuples.

You would need more computation to hash, but I bet adding frequency of the top 50 word-pairs and top 20 most common 3-tuples would be a strong signal.

( The nothing the accuracy is already good of course. I am indeed user eterm. I think I've said on this account or that one before that I don't sync passwords, so they are simply different machines that I use. I try not to cross-contribute or double-vote. )