Thursday 25 February 2021

A bad run


A bad run on the Scrabble board, in the sense that I have lost two days running. BH went out on both occasions, winning on both occasions without needing my penalty points.

I thought perhaps some error in the tile frequencies was giving her some obscure, systematic advantage, so decided that it was time to check. And, I had been getting far too many 'I's, so perhaps there were too many of them. 

Upstairs, Bing turns up a Wikipedia article which includes the left hand part of the table above, offering two distributions. One column based on a large sample of text, one based on root words in dictionaries - although I could not find any description of either how the table was constructed, when or by whom. Very tiresome.

The two frequency distributions are fairly similar, with the exception of H and W, both much commoner in text than in dictionary, but I decided that dictionary frequency was probably most relevant to Scrabble - although one could clearly argue about the treatment of root words. And one might try and work some word length angle into the analysis, with word length in Scrabble games almost certainly being on the low side compared with either text or dictionary. And if one was bored, one might try and develop some theory about letter frequencies in Scrabble games, given the rules. Perhaps some Scrabble whiz has built a database of senior level games which one could use as input.

All that aside, the big story from the dictionary is that the Scrabble people have halved the number of 'S's, perhaps because the ease with which one can tag an 'S' onto the end of a word, meant that one got a better game by cutting down the number of 'S's available. The little story is that J, Q, X and Z all get one tile apiece, although if frequency ruled they would get none. K also gets one tile, but this one is deserved.

So no excuse for my bad run here.

By way of consolation prize, I knocked up some cheese scones, just about month since BH last made them, an unusual event noticed at reference 2. 14 instead of the usual 12 scones, of which we did 11 in the first shift. That said, BH was grazing within the hour. Will there be any left in the morning?

PS: the foregoing was made easier than I expected by the table in Wikipedia pasting into Excel more or less unscathed. No need for retyping or tedious messing about converting text to data.

Reference 1: https://en.wikipedia.org/wiki/Letter_frequency.

Reference 2: https://psmv4.blogspot.com/2021/01/more-patisserie.html.

No comments:

Post a Comment