Quick Hitter — "Grape Catch"

sequence_container #70 · 4 items, all question_containers · 2 variations each · NO resources, NO context — every question carries its own dot plot

config: template: quick-hitter · navigation: linear · feedback: immediate

student surface

✕

◉ 449

This dot plot shows how many grapes (out of 5) students caught in their mouth. Each dot is a student. This is Sakeem — how many grapes did he catch?

step 1/4 · serving #9311 from container #511

database (live)

run #1 — serving variation set A (each container: members[run % count])

sequenceProgress(#70) → 0/4 answered · 0 correct · in progress — computed, never stored

sequence_containers

id	title	config
#70	Grape Catch	{template:quick-hitter,navigation:linear,feedback:immediate}

sequence_items question_container_id XOR resource_id

sequence_container_id	position	question_container_id	resource_id
#70	1	#511	NULL
#70	2	#512	NULL
#70	3	#513	NULL
#70	4	#514	NULL

question_containers + their member questions serve ONE per run

id	name	concept	members (variations)
#511	Read a student's dot	read-dot-plot	#9311 ← serving #9312
#512	Count dots at a value	frequency	#9321 ← serving #9322
#513	Name the mode	mode	#9331 ← serving #9332
#514	Find the range	range	#9341 ← serving #9342

resources

0 rows — Quick Hitter needs none: each question's figure lives inside its own content (and varies with it)

responses 0 rows — only on SERVED questions

0 rows — answer something on the left

assignment_items this sequence = ONE slot in assignment #77 · question_container_id XOR sequence_container_id

assignment_id	position	question_container_id	sequence_container_id
#77	3	NULL	#70

Slide Deck Lesson — "Point-slope form"

sequence_container #75 · slides = resources (#85 #86 owned, #88 shared) interleaved with CFU question_containers (2 variations each)

config: template: slide-deck · navigation: linear, gated · feedback: immediate

student surface

?

Point-slope form

y − y₁ = m(x − x₁)

(x₁, y₁) — a known pointm — the slope

Any line can be written from just one known point and the slope.

slide = resource #85 → row in events, never responses

database (live)

run #1 — serving variation set A (each container: members[run % count])

sequenceProgress(#75) → 0/2 answered · 0 correct · in progress — computed, never stored

sequence_containers

id	title	config
#75	Point-slope form	{template:slide-deck,navigation:linear,feedback:immediate}

sequence_items question_container_id XOR resource_id

sequence_container_id	position	question_container_id	resource_id
#75	1	NULL	#85
#75	2	#521	NULL
#75	3	NULL	#86
#75	4	NULL	#88
#75	5	#522	NULL

question_containers + their member questions serve ONE per run

id	name	concept	members (variations)
#521	Identify (x₁, y₁)	identify-point	#8811 ← serving #8812
#522	Write point-slope form	point-slope-form	#8821 ← serving #8822

resources placement = sequence_items above · owner = lifecycle (private: clones/deletes with its sequence · NULL: library, shared across sequences)

id	title	owner_container_id
#85	Slide: Point-slope form (anatomy)	owned by #75
#86	Slide: Putting it together	owned by #75
#88	Explorer: point & slope	NULL — shared library

responses 0 rows — only on SERVED questions

0 rows — answer something on the left

events slide views — never responses

id	type	detail
#1	slide_viewed	seq #75 pos 1 (resource #85)

assignment_items this sequence = ONE slot in assignment #77 · question_container_id XOR sequence_container_id

assignment_id	position	question_container_id	sequence_container_id
#77	2	NULL	#75

Testlet — "The Inventor's Notebook"

sequence_container #78 · passage = resource #482 in context (the one true context case) · free navigation · verdicts deferred

config: template: testlet · navigation: free · feedback: deferred · context: [resource #482]

student surface

Passage: The Inventor's Notebook

When Ada Reyes donated her notebooks to the city library, archivists expected sketches of her famous water filter. Instead they found pages of failures: seventeen pump designs that cracked, a filter membrane that dissolved in a week, margins crowded with the word 'again.' Reyes had saved every dead end. In a late entry she explains why: 'The failures are the map. Anyone can copy a machine that works; only the map shows you where the swamp is.' Historians now argue the notebooks did more for young inventors than the filter itself — not because they taught success, but because they made failure look like progress.

The passage is mainly about…

pick an answer — it saves immediately

passage = context (lives OUTSIDE the questions — the one true context case) · navigation: free · change answers before submit

database (live)

run #1 — serving variation set A (each container: members[run % count])

sequenceProgress(#78) → 0/3 answered · 0 correct · in progress — computed, never stored

sequence_containers

id	title	config
#78	Passage: The Inventor's Notebook	{template:testlet,navigation:free,feedback:deferred,context:[{resourceId:482}]}

sequence_items question_container_id XOR resource_id

sequence_container_id	position	question_container_id	resource_id
#78	1	#531	NULL
#78	2	#532	NULL
#78	3	#533	NULL

question_containers + their member questions serve ONE per run

id	name	concept	members (variations)
#531	Main idea	main-idea	#9411 ← serving
#532	Interpretation	interpretation	#9412 ← serving
#533	Cite evidence	evidence	#9413 ← serving

resources placement = sequence_items above · owner = lifecycle (private: clones/deletes with its sequence · NULL: library, shared across sequences)

id	title	owner_container_id
#482	Passage: The Inventor's Notebook	NULL — shared library

responses 0 rows — only on SERVED questions

0 rows — answer something on the left

assignment_items this sequence = ONE slot in assignment #77 · question_container_id XOR sequence_container_id

assignment_id	position	question_container_id	sequence_container_id
#77	4	NULL	#78

Where a sequence sits — the tree ABOVE the assignment

lessons layer (above the assignment) · added 2026-06-11 · full argument + decision card: container-taxonomy proposal §6 / D7 · this view is static — the layer above the assignment is pure structure, nothing to serve

naming rule: containers group questions (below the assignment, for serving) · lessons group assignments (above it, for curriculum) · mode: "lesson" is deprecated — items carry role instead

The bridge to what you just played: the three prototypes are sequence_containers living INSIDE assignments — and in the real curriculum those assignments aren't floating. Each lesson owns four assignments by role: bb is ONE assignment whose items interleave per KC — mini (a Quick Hitter like #70), then that KC's practice, then its check, then the next KC's mini — with each item carrying its own role (instructional | practice | check); the Slide Deck (#75) is the syn-instructional; the Testlet shape (#78) is how the unit test assembles assessment-bank items. The lesson layer never touches serving — it gives "the mastery check of Lesson 5" a foreign key instead of a naming convention.

course "NY Grade 6 Math"

unit "Fractions & Decimal Operations"position 0 → "Unit 0" · owns the unit test + assessment bank

section "Multiplying Fractions"position 2 → "Section B" · the writer's unit of thought

lesson "1/n × Whole"numbering derives globally → "Lesson 5" · the unit of progress gating · also owns its learning objectives (→ concepts, → standards)

#210bb · ONE assignment — per-KC minis + practice + checks, interleaved (items carry role)

#204syn-instructional · a deck like #75

#205syn-practice · sequential

#206syn-check · the mastery check

units / sections / lessons NEW typed, fixed depth · external_id (uuid, unique) → publish = idempotent upsert

table	parent fk	position	example
units	course_id	0	Fractions & Decimal Ops
sections	unit_id	2	Multiplying Fractions
lessons	section_id	5	1/n × Whole

unit_assignments NEW the unit test lives at unit level

unit_id	role	assignment_id
#3	unit-test	#200 (mode: assessment)

learning_objectives NEW lesson-owned, 1:N by design (shared M:N rejected — the cross-lesson layer is concepts; proposal §6) · no LO table exists today · edges: lo_concepts (primary KCs via CTA), lo_standards

lesson_id	text
#12	SWBAT see 1/n × m as repeated addition

lesson_assignments NEW role = closed enum of four · UNIQUE (lesson_id, role)

lesson_id	role	assignment_id
#12	bb	#210
#12	syn-instructional	#204
#12	syn-practice	#205
#12	syn-check	#206

assignment_items for #210 role ON THE ITEM learn it → practice it → prove it, one KC at a time · check semantics key off item role, not assignment mode

pos	role	points at
1	instructional	seq #71 — KC-1's mini (quick hitter)
2	practice	qc #551
3	practice	qc #552
4	check	qc #561 — KC-1's check
5	instructional	seq #72 — KC-2's mini
6	practice	qc #553
7	practice	qc #554
8	check	qc #562 — KC-2's check

Modules are untouched — they stay the teacher-facing classroom grouping (teacher-created, mutable, lockable) and are not part of the curriculum model. Students see this tree directly: the curriculum ships with a course view reading units/sections/lessons ("you're in Unit 0 → Section B → Lesson 5"); flat modules can't express it, and the bb interleave needs an item-role-aware player anyway. The tree is authored in the workspace repo (dirs + course.json/unit.json manifests, draft-tolerant); publish runs the reconciler once and persists its output here. Students never receive an uninterpreted tree.

External precedent: Common Cartridge/SCORM = generic files + manifest tree for exchange (our repo layer); Open edX OLX = typed fixed-depth course → chapter → sequential → vertical at runtime (our units → sections → lessons); QTI's section selection/ordering rules = serve-ONE / serve-ALL (the containers this doc prototypes). Every major standard splits along the same seam this model does.