> Aside: How is the randomisation for KWOTD done? It seems to be
> disproportionately "newer words", rather than uniformly random.

I noticed that too. It may be a technical error, but on the other hand,
TKD+KGT has roughly 2000 words, and after that we got about 1000 new
words, so maybe even when doing randomization, every third word is "new".

It pulls words randomly from those that have not been posted this cycle, so it has to show every word before it goes back to the start.

We had gone through a big chunk of the wordlist when we received a ton of new words two years in a row. That means that the system will have to go through all the new ones before it will be done.

If we keep adding 360+ words a year we might never get back to the base TKD words... ;)

qurgh