Grade 3: how Olivia's three themes get fixed across Topics 8–42

12 October 2026, midday. Themes 2 and 3 are DONE across Topics 8–42 and live: 95 recall items moved to the recall banks (every one re-read cold, all keys agreed), about twenty quizzes reordered, about 90 drawings added and read by a second reader, the package re-exported to Drive and Aiden and Kirsten emailed. Theme 1 has three questions below; its rewrites start on your taps. Two new cards from the roll-out sit on the Grade 3 backlog: the 44 video moments where something moves with no cause, and whether a fish "needs air".

Her diagnosis, in her words today: the course asks about ideas a child already has, in sentences harder than the idea, because the question has to describe the idea without a word for it. The fix is to give the word, ask with it plainly, and spend fewer questions on what children already know.

3 questions for you below, one tap each; none ruled yet. Everything you have already ruled is applied, not re-asked.

Skill: course-build → review-page → write-like-james @ 14950ce

Status: 12 October 2026, midday. Themes 2 and 3 are DONE across Topics 8–42 and live: 95 recall items moved to the recall banks (every one re-read cold, all keys agreed), about twenty quizzes reordered, about 90 drawings added and read by a second reader, the package re-exported to Drive and Aiden and Kirsten emailed. Theme 1 has three questions below; its rewrites start on your taps. Two new cards from the roll-out sit on the Grade 3 backlog: the 44 video moments where something moves with no cause, and whether a fish "needs air".

Decisions I need from you

1. Should each topic's key word be taught in the first lesson where the idea appears?

The course says: a term is coined in the lesson the register assigns it, which in some topics is two or three lessons after the idea first appears; until then the questions describe the idea in a sentence ("the thing that stays still that you compare against") instead of naming it ("the reference point").

Why it matters: Olivia today: "Definitely have that introduced earlier and then have questions reinforcing the vocab as well as the concepts. I think having the vocabulary will also make it easy to generate questions that are more straightforward." Your 25 September rule keeps the rule stated last within a video; this is about which lesson coins the word, not where in the video.

Cost of changing it: for each topic where the gap is two lessons or more, the coining beat moves to the earlier lesson's script and article (a video edit by id), and the earlier quizzes gain the word; the register's lesson order is unchanged.

My recommendation: yes, where the word is coined two or more lessons after the idea first appears; leave one-lesson gaps alone

2. Should there be fewer questions on the ideas children already have, and more on the new ones?

The course says: every lesson quiz has about ten items and every PP100 bank at least fifty, spread evenly across the topic's atoms, so "did it move?" and "push or pull?" get as many items as "balanced forces".

Why it matters: Olivia today: "fewer questions about whether something moved or not. Also with pull and push, I'm guessing that most kids can get that pretty quickly. So some questions on that but then maybe move on faster." Each extra item on an easy idea is a chance to word it badly and a minute not spent on the hard idea.

Cost of changing it: on atoms the register marks as everyday knowledge, lesson quizzes shrink to about six items and their share of the PP100 bank drops; the bank stays at fifty by adding items on the topic's newer atoms, each blind-read.

My recommendation: yes: about six items for a quiz on an everyday idea, and no more than a sixth of a PP100 bank on such atoms

3. Should these plain-question limits become a standing Grade 3 rule?

The course says: no limit on stem length; the 28 September ban on "Tessa says … is she right?" items covers initial instruction but many remain in topics written before it; the 5 October ban on negative stems likewise.

Why it matters: Olivia today: "The language should not be several times more complicated than the idea." A rule with a checker catches the next one before she does; a reading pass alone fixes only what is there now.

Cost of changing it: a checker sweep (the stem over two sentences or twenty-five words, an option over eight words, a named-student claim, a negative stem, a true/false item) that fails new Grade 3 content and lists hits on live content for the reading pass; the reading pass then rewrites the hits.

My recommendation: yes, as written: two sentences and twenty-five words for a stem, eight words for an option, no claim frames, no negatives, no true/false, in Grade 3 quizzes and banks

What each theme is, and how it gets fixed

Theme 1: questions that trip on their wording (148 of her 606 notes, a quarter). Long set-ups, "so-and-so says … is she right?", true/false traps, two options that both work.

A mechanical shortlist first: 683 stems of three sentences or more than thirty words, 204 named-student claims, 242 negative stems in Topics 8–42 (overlapping; 4 308 items in all).
Then a non-author reader applies her test to every shortlisted item: is the language harder than the idea? If so, rewrite in her plain form: one scene sentence, one direct ask, the taught word used, options parallel and short.
Every changed item is read cold by a second reader for its key. Keys, the science and coined terms are not touched.
On your answers to questions 1 and 2, the same pass moves coining earlier and thins the easy quizzes.

Theme 2: pictures and animation (124 notes, a fifth). No picture where a child must imagine the scene; direction asked in words; hands drawn as sticks; video objects that move with no cause.

158 direction items get a drawn picture with lettered arrows, in the form she asked for on Topic 5. Code draws these, because the arrows must be exact.
23 drawings of hands or fingers in Topics 8–22 become the oblong arm of your 29 September rule (Topics 24–42 were already swept).
A reading pass adds a picture wherever a child must imagine something unfamiliar: code where the drawing must be exact, David's image pack where it must look real. About 250 new pictures in all.
Videos where something moves with no cause are listed for you as edits by Peps id; none is sent without your word.

Theme 3: quiz order (86 notes, a seventh). The first item after a video recalls an earlier lesson, or is harder than the video; an idea is asked before it is recognised.

85 of the 159 lesson quizzes open on a recall item from an earlier lesson: each moves to the topic's recall bank, under your 8 and 9 October rulings (at most two before a lesson, and only for an idea from well back).
Every quiz then opens on the video's own simplest case and climbs from recognition to production. About 35 quizzes are reordered.
Lesson order, splits and merges are not touched; those stay as cards.

Olivia's words today (pasted by James)

> One item that I feel like isn't a huge issue when you just look at an individual question but becomes a big issue in its totality is this: there are a lot of concepts that I think are not really that difficult and that kids have a pretty good intuitive sense of. But then there are questions asked that are kind of linguistically complex and hard to parse. I'm concerned that we're taking things that are easily understood and actually going to make the students doubt if they understand it. > > I'd rather see more straightforward questions and time spent on questions that are newer and more difficult. For example fewer questions about whether something moved or not. Also with pull and push, I'm guessing that most kids can get that pretty quickly. So some questions on that but then maybe move on faster vocabulary. > > I'm not sure how easy it is to just add in the vocabulary but I know we talked about this. Definitely have that introduced earlier and then have questions reinforcing the vocab as well as the concepts. I think having the vocabulary will also make it easy to generate questions that are more straightforward. Right now I think what it's trying to do is generate questions about concepts that are kind of intuitive but a little hard to describe. That's why we're ending up with this language issue.

What happens next

Themes 2 and 3 run now, strand by strand, each topic committed and redeployed as it finishes. Theme 1's shortlist is built now; its rewrites start on your three taps, and your answers also go to the vault as rules in your words.

Skill: course-build → review-page → write-like-james @ 14950ce · built 12 October 2026 by build_tools_rulings_page.py