When I said “a number of uncommon test forms from June 2018 still have not seen a major administration,” I meant that College Board has a cache of test forms first tried out last June that have not yet been given to nearly all test takers on a given test date. I will try to explain - I do not have the proper, technical terms. This is all information I have read about elsewhere, just by paying attention, with a little putting together of puzzle pieces. (To be clear for others, I am just a parent.)
Back in June 2018, there were a large number (>20?) of different sets of questions, or test forms, given in addition to the main, most common one that had the famously harsh scale. This is evident from red dit chatter; entirely different sets of reading passage subjects are the big clue.
These many different test forms were given to small portions of test takers. For example, uncommon test forms W, X, Y, Z (making up these labels) may be given to only a handful of kids in a room while everyone else receives test form A, the most common one. This is College Board trying out full test forms, a sort of final level of experimentation.
Each test form gets its own scoring scale unique to it. The scale is based on difficulty level, of course, such that tests of similar difficulty end up with similar scales. College Board chooses the difficulty level intentionally. (NB the difficulty can be tracked, as each question gets a hard, medium, or easy difficulty designation which is reported in the SAS and QAS. Numbers of hard, medium, and easy can be added by section and compared to other test forms. Example: Nov 2017 had 28 hard math questions while Oct 2018 had only 15.)
College Board then takes the June uncommon test form results to help finalize its scoring scales for each those forms, so that they are ready for a later major administration; thus it takes six weeks to get scores for June rather than the normal two weeks (personally I don’t buy the excuse of AP causing the delay). Once the scale is set, it should remain the same for each future use of that test form, unless College Board decides to replace a few unscorable questions with new ones (that appears to happen from time to time). My rough understanding is that College Board is not supposed to switch out, say, an entire math section, as that may render the entire test’s scale inaccurate, but that is an interesting question for the psychometric experts if any are around here.
Later, College Board might use our hypothetical test form X, uncommon from June, for the October US administration, given to almost everyone for the Oct date. It is a “new” test form because it will be the first time test form X has ever been given to almost all testers for a particular test date and there would be virtually zero prior internet discussion of the test form beyond one or two random forum posts asking hey, did anyone else have the test form with the elephants and the daisies (I am making up those reading subjects). (As a separate matter, other groups/times of testers, such as Sunday administration, accommodation classrooms, etc. might be getting test form Z for our hypothetical Oct test date. College Board might also throw in a separate test form for a couple of people per room, though the extent of that is unclear. And as always, international testers get a different test form than US.)
Uncommon test forms from June 2018 (per red dit discussion at the time) appeared for US Oct, Nov, Dec, both March 2019, one of the April school day (if I recall correctly, while the other was a major reuse from a couple of years ago), and both May 2019 tests. I have not paid much attention to the discussion of test forms given internationally over the past school year, except that international takers in May were surprised by the harshness of the scale that is similar to June 2018 US. (Those who pay attention to this kind of thing were not surprised at all and very glad that College Board finally stopped its prior pattern of reusing the April school day for May international.)