MK, I had an idea that conchie is also used sometimes to mean simply conscientious, as in, "You finished your essay a week early? You're so conchie!" However, a quick Google suggests this usage is a lot rarer than I had thought. I'm inclined to think the word ought to be reclassified to rare, along with calcine. However, chicane perhaps could remain common.
TRex, I have used such statistics in the past, to identify candidates for reclassification. After the site had been going for a couple of years, I generated a database of all the common words that had appeared in puzzles, with the percentage of times each word had been found. I then reviewed a lot of the words that had the lowest percentages. As a result, quite a few words were changed to rare. I think this might have been before the forum started, so there was no big announcement about it.
Now that a lot more play statistics have been accumulated, it would probably be a very good idea to carry out a similar exercise again. However, when I did this before, I didn't feel that I could completely automate the process, by saying that all words found by fewer than x% of players should be rare, and all others common. One complicating factor I mentioned before, with the example of inhale, is that some well-known words can be hard to think of because of their uncommon form. Of course, thinking of such words is part of the challenge of the game.
On the other side of the ledger, there are words that have become well-known to Chi players, although they may not be common words in general usage. For example, tine is one of the most frequently occurring "common" words in Chi, because its letters are all very common ones. I imagine regular players have become familiar with this word, and play it whenever the letters allow, but how well-known is it to the general public?
Morbius, your idea of more statistics on the Web site is a good one. I'll give some thought to it. It would be interesting, for example, if a player who comes up with an unusual word could see how many other players got that word. (I think something like this has been suggested before. Maybe if the idea comes up often enough I'll get around to doing something about it.) By looking at the number of rosettes on the scoreboard, and the chart below the scoreboard, you can get an idea of how many players got all the common words, and how many reached each level, but not percentages.
Of course, the ideal would be that the percentage reaching each level is about the same for every puzzle. It would be an interesting exercise to see how much variation there actually is between puzzles.