Unit 9 — Bounded Community Contributions and an Auditable Portfolio

Taking one small finding to a submission-ready state without overstating its status

The closing unit on bounding one mathematical contribution, preparing an inspectable correction package, distinguishing external statuses rigorously, designing withdrawal routes, and assembling a portfolio of evidence from Units 1–8.

1 Learning outcomes

This unit has the stable identifier O017-U09. After completing it, you will be able to:

select one community contribution that is small, useful, lawful, and completable without silently expanding its scope;
distinguish operationally between the statuses draft, prepared, submitted, acknowledged, accepted, merged, and published;
reject claims of external status that lack evidence of an event;
package the target, locator, diagnosis, minimal repair, validation, scope, uncertainty, rights, and attribution in one contribution package;
write one concise message that respects maintainers and can be acted upon without flooding a channel with separate comments;
establish routes for stopping, withdrawing, correcting, or reverting a contribution if it proves wrong, outdated, duplicate, or no longer authorized;
assemble a portfolio mapping each claim of competence to actual evidence from Units 1–8; and
assess course completion through the quality, traceability, and limitations of the evidence, rather than file counts or claims of activity.

Prerequisites, unit state, and action boundaries

Unit 9 uses the results of all preceding units. It does not repeat how to read claims, search for sources, close proof gaps, establish provenance, write exposition, run computations, diagnose errata, or write referee reports. It takes one finding that is already suitable and turns it into a bounded contribution package and portfolio evidence.

The case in this unit is entirely synthetic. The example package stops at locally prepared status. The example message is not sent, no outside party is contacted, and no correction is accepted, merged, or published. Any future submission requires a separate instruction explicitly authorizing it for a specified destination.

2 Small contributions need firm boundaries

A community contribution is not a measure of how many things you can find. One accurate, reproducible, easily reviewed correction is often more useful than a long list of half-developed findings.

A contribution is bounded when its package answers six questions.

Exactly which object and version does it affect?
What single problem has actually been demonstrated?
What minimal change addresses that problem?
What tests show that the change works without exceeding the claim?
What has deliberately not been examined or changed?
Who has authority to decide on submission, acceptance, merging, and publication?

If the answer to question two contains five independent defects, make separate packages or choose the most coherent one. “Tidying up the whole chapter while we are at it” usually obscures review, attribution, and reversion.

Original O017 bridge

The seven-status state machine, evidence-gated contribution package, odd-number induction case, conditional single message, withdrawal routes, cross-unit portfolio architecture for Units 1–8, exercises, and capstone in this unit are original O017 material. The unit moves a finding from “documented” to “ready for community consideration” without inventing a single external event.

3 Seven statuses that are not interchangeable

A status is a claim about an event. Every advancement of status requires new evidence. Perfect contribution content never, by itself, proves that an outside party has seen or approved it.

The seven labels below are independently evidenced event predicates, not a single progress score. An artifact record may state published=true while still stating merged=unknown if a public copy has been observed but the merge route is unavailable. A later event does not automatically fill in intervening events.

Status	Operational meaning	Minimum evidence	Claims not yet warranted
Draft	content is still being prepared or has not passed all local gates	local file or work record	ready to submit, examined by another party
Prepared	the local package is complete, validated, lawful to use, and has a specified destination and message	manifest, test results, rights audit, local GO decision	submitted or seen
Submitted	the package has actually been transferred to the destination channel	transaction URL/ID, time, account/actor, and bytes or content sent	the recipient has read it
Acknowledged	the destination system or person has confirmed receipt	discoverable receipt, response, or platform status	the content is substantively accepted
Accepted	an authorized party has approved the contribution or its proposed repair	explicit decision identifying the contribution	the change has been merged
Merged	the change is part of the designated active branch or source	commit/change identity or readback of active source bytes	a public release contains it
Published	a particular public artifact actually contains the change	URL/DOI/release and readback of public bytes or content	all other copies and editions have been updated

“Submitted” cannot be inferred from the existence of a submit button. “Acknowledged” does not mean accepted. “Accepted” does not mean merged. “Merged” does not mean published. Conversely, a site may publish a built copy without a merge into the assumed branch; record the actual route rather than forcing events into a sequence that did not occur.

3.1 Additional statuses without erasing history

Besides the seven main statuses, a record may use:

held: a gate has not been met or authority is not yet available;
withdrawn: the submitter requests that a submitted contribution receive no further processing;
rejected: the authorized party does not accept the contribution;
superseded: a new package replaces the old one, with an explicit link;
reverted: a change that was once active is reversed through a new event.

None of these statuses erases evidence that an earlier state occurred. If a contribution was submitted and then withdrawn, record both events. Do not rewrite the history as “never submitted”.

3.2 A non-exhaustive canonical route and its evidence rules

Predicate already established	New predicate on the canonical route	Required event	The gate fails when
Draft	Prepared	validation is complete, scope and rights are clear, and the single message is locally approved	tests fail, the target is unclear, or licensing remains unclear
Prepared	Submitted	an authorized submission action and a transaction record	there is only an intention, preview, or unsubmitted form
Submitted	Acknowledged	evidence of receipt by the destination system or person	there is only a local screenshot taken before submission
Acknowledged	Accepted	a substantive decision by an authorized party	there is only an automatic reply or a thank-you
Accepted	Merged	the active change can be identified	approval has not been implemented
Merged	Published	a particular public artifact contains the new bytes or content	the release is still building or the public copy is still old

Transitions may branch to held, rejected, withdrawn, superseded, or reverted. The table records one common route, not every possible path, and it does not state logical implications between predicates. If particular public bytes have been read back but merge evidence is unavailable, record diterbitkan (published) as established and digabung (merged) as unknown; do not invent a merge event. Likewise, an active change may be observed without an archived acceptance decision.

Each recorded predicate must include a time, actor or system, object, evidence, and boundary. Without new evidence, do not add a new predicate. Keep an event timeline and established/not established/unknown values for each predicate.

4 The gate for selecting one contribution

Before preparing a change, assess the candidate using these gates.

Gate	Question	Local GO	HOLD or NO-GO
Target	are the work, version, copy, and locator frozen?	the target can be found again	only “the latest edition” or an unclear location
Marginal value	does the contribution add a correction or evidence not already available in the package examined?	a specific difference can be stated	a known duplicate or unclear benefit
Diagnosis	can the failure or gap be reproduced?	a complete example, calculation, or proof	only an impression or a guess about intent
Minimality	has the smallest sufficient change been chosen?	one claim or proof path can be isolated	broad stylistic changes are mixed with the correction
Validation	are the old and new claims tested according to their quantifiers?	the test fails before and passes afterward	only visual appearance or one supporting example
Rights	are the source and contribution licenses compatible, and is attribution ready?	the rights basis is recorded per component	an unknown license or unclear third-party material
Communication	does one concise message contain the locator, evidence, patch, and boundaries?	the message is actionable	many separate comments or accusatory language
Authority	is the current action local only, or is there separate authorization to submit?	status and actor are clear	readiness is equated with permission to send
Reversion	is there a way to hold, withdraw, or revert if the premises change?	triggers and actions are specified	the change is assumed incapable of being wrong

A candidate that fails the rights, target, or diagnosis gate stops before prepared status. A candidate that passes all local gates may be prepared, but is still not submitted without submission authority.

5 Worked case: one incorrectly indexed term

This case uses a synthetic community source. All available context is embedded below; there is no real repository or maintainer.

O017-U09-UP-V1 — Frozen synthetic target

Work ID: O017-U09-UP-W01, An Induction Worksheet for Odd-Number Sums.
Version: O017-U09-UP-V1, label 1.2-sintetis (synthetic).
Copy: the complete excerpt in this unit.
Source license: CC BY-SA 4.0, as synthetic O017 material.
Locator: Lemma 2, proof lines L1–L5, #o017-u09-upstream-snippet.
Simulated destination: O017-U09-DEST-01, a classroom channel accepting only one text message of at most 150 words and a patch to excerpt L3–L5 under the same license. This is not an external service, account, or channel.
In the complete simulated context, V1 is the only registered version and there are no duplicate reports.
Impact outside the excerpt: unknown; no other dependency graph is available.

Lemma 2 and its proof in O017-U09-UP-V1

Lemma 2. For every $n\in\mathbb N$ , where $\mathbb N=\{1,2,3,\ldots\}$ ,

$S_n:=\sum_{k=1}^{n}(2k-1)=n^2.$

L1. For $n=1$ , we have $S_1=1=1^2$ .

L2. Suppose that $S_n=n^2$ for some $n\geq1$ .

L3. The new term when the upper limit of summation changes from $n$ to $n+1$ is $2n-1$ .

L4. Thus $S_{n+1}=S_n+(2n-1)=n^2+2n-1$ .

L5. The last expression equals $(n+1)^2$ , completing the induction.

The lemma statement is correct, the base case L1 is correct, and the form of the induction hypothesis L2 is correct. The first defect occurs at L3: the term with index $n+1$ is

$2(n+1)-1=2n+1,$

not $2n-1$ . Consequently, the equality in L5 is also false, because $n^2+2n-1\neq(n+1)^2$ for every $n\geq1$ .

5.1 Minimal reproducer

Take the first step, $n=1$ .

The direct definition gives $S_2=1+3=4$ .
The V1 recurrence at L4 gives $S_2=S_1+(2\cdot1-1)=1+1=2$ .
The lemma’s conclusion gives $2^2=4$ .

Thus lines L3–L4 do not compute the defined object $S_2$ , and L5 cannot close the induction. This check is exact, satisfies the domain $n\geq1$ , and refutes the proof step, not the lemma statement.

5.2 Repairing the proof

Keep L1, L2, and L5. Replace only L3–L4 with L3′–L4′ below; L5 is displayed again afterward so that the proof’s closing step can be inspected.

L3′–L4′ and retained L5 — the minimal repair

L3′. The new term when the upper limit of summation changes from $n$ to $n+1$ is $2(n+1)-1=2n+1$ .

L4′. By the induction hypothesis,

$S_{n+1}=S_n+(2n+1)=n^2+2n+1.$

L5 (retained unchanged). The last expression equals $(n+1)^2$ , completing the induction. Indeed, after L4′, the last expression is $n^2+2n+1=(n+1)^2$ . Together with the base case L1, the principle of induction proves the lemma for every $n\in\mathbb N$ .

This is the minimal repair within the stated scope: only L3 and L4 are changed. L5 needs no change because it becomes true once the expression in L4 is repaired; displaying it again above does not constitute a source change. The lemma statement, base case, definition of $\mathbb N$ , and all other parts are untouched.

5.3 Before-and-after validation

Test	V1 before repair	Candidate V2 after repair	Result
base case $n=1$	$S_1=1$	unchanged	passes in both
first step	predicts $S_2=2$	gives $S_2=1+3=4$	V1 fails, V2 passes
general new term	uses $2n-1$ , the $n$ th term	uses $2(n+1)-1=2n+1$	V2 matches the definition
algebraic identity	$n^2+2n-1=(n+1)^2$ is false	$n^2+2n+1=(n+1)^2$ is true	V1’s failure and V2’s repair are exact
quantifier	the step is intended for every $n\geq1$	the new identity holds for every $n\geq1$	scope is preserved
conclusion	induction is incomplete	base case and induction step are complete	the lemma is proved in V2

Validation uses a failing example to expose the defect and a general identity to complete the repair. Testing only $n=1,2,3$ after the change is not enough to prove the step for all $n$ .

6 A minimal patch and erratum draft

The change package preserves both the old and new forms.

Location	Before, V1	After, candidate V2	Reason
L3	the new term is $2n-1$	the new term is $2(n+1)-1=2n+1$	the new index is $n+1$
L4	$S_{n+1}=n^2+2n-1$	$S_{n+1}=n^2+2n+1$	substitution of the new term and induction hypothesis
L5	the preceding expression is called $(n+1)^2$	unchanged; the preceding expression is now $n^2+2n+1$	L5 is valid after L4 is repaired, so no further change is needed

Internal technical summary — not an outgoing message

In O017-U09-UP-V1, Lemma 2, proof L3–L5, the term added when moving from $S_n$ to $S_{n+1}$ is written as $2n-1$ . The new term is actually $2(n+1)-1=2n+1$ . With that change, $S_{n+1}=n^2+2n+1=(n+1)^2$ , making the induction step valid. For $n=1$ , the old wording predicts $S_2=2$ , whereas the definition gives $S_2=4$ . The statement of Lemma 2 and its base case are unchanged. Effects on material outside the excerpt have not been examined.

This summary is the package’s technical content, not a second communication. The only outgoing erratum draft is at #o017-u09-outbound-draft.

6.1 Scope and uncertainty

What has been established:

V1 has an indexing error at L3;
L4–L5 fail as a direct consequence;
the lemma statement remains true;
L3′–L4′ together with retained L5 give a complete induction step; and
the change requires no third-party material.

What remains unknown:

whether a future real target has versions or reports different from the simulated context;
whether the excerpt is used by other results;
the rules and rights of a real destination, should the synthetic target ever be replaced;
whether maintainers would choose the same repair; and
whether candidate V2 will later be accepted, merged, or published.

These uncertainties do not diminish the mathematical validity of the repair. They prevent claims of external marginal value or community status that have not been evidenced.

7 Rights, licensing, and contribution attribution

Four objects of rights must be distinguished:

target source: the text to be changed and its license;
diagnosis: new wording and evidence written by the examiner;
patch: the form of the change, which may be a derivative work;
supporting artifacts: images, code, data, or quotations, if any.

In this synthetic case, the source and contribution are original O017 material licensed under CC BY-SA 4.0. There are no third-party donors or artifacts. Package attribution must continue to identify the synthetic work, version V1, locator L3–L5, and the same license. Synthetic status must be preserved so that a local ID is not mistaken for an external project.

For a real target, do not infer that permission to read is permission to distribute a patch. Freeze the license text and contribution rules that actually apply to the target version. If one image or table has different rights, record them per component. Unclear licensing is a HOLD gate, not an invitation to guess.

8 A review-ready contribution package

A complete package can remain small. Use the following manifest.

Component	Worked-case content	Local acceptance test
Target identity	O017-U09-UP-W01, V1, Lemma 2 L3–L5	version and locator can be found again
Problem claim	the new term is incorrectly indexed; the induction proof fails	reproducer $n=1$ and the general identity are available
Minimal repair	replace L3–L4; retain L5	the statement, base case, and already valid lines are unchanged
Validation	before/after table and induction proof	V1 fails and V2 passes for the right reasons
Scope	the Lemma 2 excerpt only	other results are not claimed to have been audited
Uncertainty	other versions, duplicates, dependencies, destination rules	all remain explicit
Rights	synthetic source and contribution under CC BY-SA 4.0	no component lacks a rights basis
Message	one concise draft	locator, evidence, proposal, and boundaries are present
Status	prepared for the local simulation; not submitted	there is no external transaction record
Reversion	hold, supersede, or withdraw according to the trigger	the earlier history is preserved

8.1 Package checklist

Before the status changes from draft to prepared, check that:

target, version, copy, and locator are exact;
the old wording is preserved without a paraphrase that changes the claim;
the reproducer satisfies every hypothesis;
the minimal repair has a proof or general validation;
unaffected results are stated;
unaudited effects remain unknown;
licensing and attribution are present for every component;
no unrelated finding has been inserted;
the single message is sufficiently concise and does not judge people;
external status does not exceed the evidence;
submission authority is stated separately; and
stopping triggers and withdrawal routes are available.

One “no” concerning the target, mathematics, rights, or status keeps the package at draft.

8.2 Worked-case status ledger

Event	Date	Status after the event	Evidence	External claim
EVT-01: target, diagnosis, and patch recorded	2026-08-21	Draft	target V1, reproducer, L3′–L4′, and retained L5 in this unit	none
EVT-02: mathematics, scope, rights, message, and reversion routes pass local gates	2026-08-21	Prepared for O017-U09-DEST-01	validation table, manifest, checklist, and frozen message	not submitted
No EVT-03	not applicable	not Submitted	no URL, transaction ID, account, or submission time	no contact
No destination event	not applicable	not Acknowledged, Accepted, Merged, or Published	no receipt, decision, active change, or release	none

The last established status is Prepared within the local simulation. The word “prepared” does not turn O017-U09-DEST-01 into a real destination.

9 One message, not sent

The following is the sole outgoing communication draft for the worked case. It is a local artifact and has not been sent.

Concise message draft — status: not sent

Title: Lemma 2: term index in the induction step

In version 1.2-sintetis, Lemma 2, proof L3–L5, moving from $S_n$ to $S_{n+1}$ adds $2n-1$ . The new term should be $2(n+1)-1=2n+1$ . The old wording gives $S_2=2$ at $n=1$ , whereas the definition gives $S_2=1+3=4$ . The minimal proposal is to replace L3–L4 so that $S_{n+1}=n^2+2n+1$ ; L5 then holds unchanged because this expression equals $(n+1)^2$ . The lemma statement and base case remain. I examined only this excerpt; effects beyond Lemma 2 and possible repairs in other versions remain unknown.

Do not add multiple apologetic comments, follow-up corrections, or explanatory messages before a response; the package already contains the evidence and boundaries. If the user separately authorizes a future action and this message is actually sent, add exactly this closing line:

Codex, at the user’s instruction

Do not add that line to the local draft as evidence of submission, and do not claim another party’s identity, acceptance, or endorsement. This unit does not send anything.

10 Routes to hold, withdraw, supersede, and revert

A contribution can change after validation. Recovery routes must exist before submission, not be invented when a problem arises.

State	Trigger	Correct action	Evidence preserved
Draft/Prepared	the diagnosis proves wrong or duplication is established	hold; mark locally superseded or cancelled	old package, reason, new examination
Submitted but not acknowledged	wrong target or harmful patch	use the same channel to request withdrawal if authorized	submission record and withdrawal request
Acknowledged but not accepted	new information changes the diagnosis	send one consolidated correction or request a hold	destination response and new evidence
Accepted but not merged	the new patch fails validation	notify maintainers through the designated route and withdraw the submitter’s approval	acceptance decision and test failure
Merged but not published	a regression is found	propose a revert or testable corrective patch	active change identity and regression results
Published	the public artifact is wrong	prepare a new correction, release notes, and readback; do not erase history	old and new release identities

Actual action always follows the channel’s authority rules. If you lack authority, prepare a dossier and escalate to the decision owner; do not try to repair external state silently.

10.1 Worked-case withdrawal triggers

The odd-number sum package must be held or withdrawn if:

the target copy is not the stated V1;
missing context defines L3 using a different index;
the destination version already contains correction L3′–L4′, making the contribution a duplicate;
the license or destination rules do not permit the form of contribution;
independent validation finds that the patch affects more material than stated; or
the user withdraws authority before the outgoing action finishes.

Withdrawal does not mean the diagnosis must be deleted. Record the reason and new status.

11 A portfolio is a map from claims to evidence

A portfolio is not the same as a folder containing all your work. It is a bounded argument:

I claim a particular competence; this artifact demonstrates a particular action; this locator contains the evidence; this boundary states what has not been demonstrated.

Every claim of competence must have primary evidence, a locator, acceptance criteria, and a brief reflection. One artifact may support several competences, but the connections must be explained. Many files without a map do not, by themselves, constitute evidence.

11.1 Minimal portfolio architecture

Manifest: a list of artifacts, versions, formats, rights, and byte identities where available.
Claim–evidence map: competence, artifact, locator, action, and boundary.
Main samples: one coherent product per competence group, not every draft.
Revision history: before/after evidence and reasons for changes.
Uncertainty ledger: unresolved questions.
Rights statement: origin, adaptation, license, and attribution per component.
Accessibility notes: heading structure, alternative text, tables, formulas, navigation, and alternative formats actually examined.
Bounded reflection: what changed because of evidence, one corrected error, and one ability not yet demonstrated.

11.2 Mapping evidence from Units 1–8

Unit	Claim of competence	Suitable evidence to select	Required locator or fields	Boundary to state
U1	reading an argument as a network of claims	a typed claim graph and one path audit	node ID, arrow type, entry point, output	the graph is not proof that every claim is true
U2	assessing source authority and version	a source decision with a bounded comparison	work, edition/commit, copy, license, GO/NO-GO reasons	a search does not establish completeness across all literature
U3	reconstructing a missing step	reconstruction with an input/output contract	source claim, gap, new lemma, proof, strength boundary	new wording is not the source’s original text
U4	maintaining citations and provenance	a traceable source and decision ledger	version, locator, relationship, rights, event	the ledger does not replace mathematical proof
U5	writing auditable exposition	a clean manuscript and separate dependency audit	reader contract, claim, proof, example, boundary	expository quality does not prove originality
U6	using computation as empirical evidence	a rerunnable package and results record	code, input, environment, command, output, oracle/hash	empirical output does not automatically become a universal proof
U7	formulating a bounded correction	an erratum dossier with a reproducer and correction	target, claim status, defect type, severity, confidence, history	a local draft is not a submitted report
U8	giving and responding to criticism	seminar notes, report, response, and matrix	time/location, comment, action, evidence, status, open issue	the recommendation assesses the manuscript within scope, not the person

This map specifies types of evidence, not mandatory filenames. Choose the strongest and most easily inspected artifacts. If a unit does not yet have evidence that passes its gates, write “not yet demonstrated” and plan the repair; do not fill the cell with reflection unsupported by an artifact.

12 Manifest contract and evidence identity

Every portfolio artifact must have at least:

Field	Function
evidence_id	stable identity within the portfolio
unit	source competence in U1–U8
claim	ability supported
artifact	title and an actually available path or URL
version	edition, date, commit, or local status
locator	the exact part demonstrating the action
action	what the participant did
validation	tests, audits, or reviews already passed
rights	license, attribution, and third-party components
state	draft/prepared/submitted/and so on, where relevant
limitation	claims that the artifact does not support

If a checksum is recorded, compute it from the bytes of the artifact actually stored. Do not copy a checksum from another manifest without reading the artifact again. If the artifact changes, create a new version and hash; do not retain the old byte identity.

13 Package and portfolio accessibility

Traceability is useless if the evidence cannot be read. For every artifact:

use a meaningful heading hierarchy;
name tables and wrap wide tables so they can be scrolled and focused;
do not use color as the only status indicator;
provide alternative text for images that convey information;
provide transcripts or timed notes for media;
explain symbols at first use;
keep links meaningful without relying on “click here”;
test keyboard navigation and zoom in the final output; and
record untested surfaces rather than claiming complete accessibility.

A portfolio may contain PDF, HTML, and editable source. Matching titles do not guarantee matching content; the manifest must pair formats that have actually been built and checked.

14 Evidence-based reflection

Reflection is not a general autobiography. Use a three-part pattern.

Initial decision: what was chosen, and on what evidence?
Change: which finding forced a concrete revision?
Current boundary: which ability has not been demonstrated, and what evidence would be needed?

Example:

I initially thought the $n=1$ check revealed only a typo. Auditing the general identity showed that L5 was also false as an equation, so the package repairs L3–L4, retains L5, and retains the lemma statement. This evidence does not yet show that I have audited dependent results, because copies outside the excerpt are unavailable.

That reflection identifies a change and a boundary. “I learned a lot about induction” is not sufficient evidence of competence.

15 Exercises

O017-U09-E01 — Status without leaps. For each of the following situations, identify the established event predicates, predicates that remain unestablished or unknown, and evidence still missing: (a) the package passes local validation; (b) the destination form displays a preview but has not been submitted; (c) the platform provides an automatic transaction ID; (d) a maintainer writes “we will review it”;
1. a maintainer approves the patch; (f) an active commit contains the patch; and (g) a public release has been read back and contains the change. See O017-U09-H01.
O017-U09-E02 — Reproducer and scope. Check the defect in excerpt L3–L5 using $n=1$ , then prove the identity used in L3′–L4′ and explain why L5 can be retained for general $n$ . State three things established and three things still unknown. See O017-U09-H02.
O017-U09-E03 — Patch minimality. A contributor wants simultaneously to replace the notation $S_n$ , rewrite the entire introduction to induction, change six examples, and repair L3. Separate required changes, optional changes, and out-of-scope changes. Assemble one reviewable package without losing the mathematical repair. See O017-U09-H03.
O017-U09-E04 — Rights and attribution. A target is licensed under CC BY-SA 4.0, one diagram within it has a separate credit but no license, and the patch changes only proof prose. Determine which components may be included in the package, what attribution is required, and which gate prevents use of the diagram. See O017-U09-H04.
O017-U09-E05 — Message and withdrawal. Write one message of at most 120 words for defect L3–L5, including the locator, reproducer, patch, and boundaries. Then suppose the destination version turns out to have been repaired already. Write the status route and withdrawal or stopping action without erasing history. The message remains unsent. See O017-U09-H05.
O017-U09-E06 — Portfolio map. Choose four claims of competence from four different units. For each, fill in evidence_id, artifact, locator, action, validation, rights, status, and boundary. Find one claim with insufficient evidence and write its repair action without inventing evidence. See O017-U09-H06.

16 Hints and answer guidance

O017-U09-H01. (a) prepared; (b) still only prepared, because a preview is not submission; (c) submitted if the ID identifies a genuinely completed transaction; (d) acknowledged, while submission and acceptance may be added only if their evidence is also available; (e) accepted, but not automatically acknowledged, submitted, or merged; (f) merged, without inventing an acceptance decision; and
1. published for the release read back, without inferring a merge. Every predicate still requires an object identity and time; other predicates receive an unknown value when the case supplies no evidence for them. Return to O017-U09-E01.
O017-U09-H02. For $n=1$ , V1 gives $S_2=1+(2\cdot1-1)=2$ , whereas the definition gives $4$ . In general, the term with index $n+1$ is $2(n+1)-1=2n+1$ , and $n^2+2n+1=(n+1)^2$ . Established facts include the defect’s location, the step’s failure, and the patch’s sufficiency; other versions, downstream effects, and maintainer decisions remain unknown. Return to O017-U09-E02.
O017-U09-H03. Only L3–L4 must change. L5 is already correct as a structural step after L4 is repaired, so it may be displayed again for auditing but must not be counted as a changed line. A one-sentence explanation may be included if needed for the audit. Replacing the notation, introduction, and six examples is optional work or a separate package. Mixing these changes increases the regression surface and makes reversion harder. Return to O017-U09-E03.
O017-U09-H04. The target prose and patch can be handled under CC BY-SA 4.0 with attribution, identification of changes, and a compatible license. The diagram is unnecessary for the patch and must be excluded; a credit without a licensing basis is insufficient for redistribution. If the diagram becomes necessary later, hold that component until its rights are clear. Return to O017-U09-E04.
O017-U09-H05. The message should resemble the single draft: one locator, the $n=1$ failure, two replacement lines, an explanation that L5 is retained, and the audit boundary. If the destination version was repaired before submission, status remains prepared and then held or superseded as a duplicate; there is no external withdrawal because it was never submitted. If this becomes known only after submission, retain the transaction record and request withdrawal through the same channel if authorized. Return to O017-U09-E05.
O017-U09-H06. An adequate row links an action claim to a locator that actually demonstrates it and states the test and boundary. Reflection without an artifact does not replace evidence. For an unsupported claim, use “not yet demonstrated” status and state the product and gates still to be completed. Return to O017-U09-E06.

17 Capstone: a bounded contribution and O017 portfolio

Build one local capstone package from your own work in Units 1–8. This assignment does not authorize contact with outside parties. If no real target has clear rights and a clear version, use the worked case’s synthetic target and state that choice.

The product must contain:

one scope statement of at most 120 words naming the target, one problem or gap, the proposed repair, and what remains untouched;
target identity, version, copy, locator, examination time, and rights basis;
the exact wording before the change and the exact wording after it;
a reproducer or proof reconstruction that checks hypotheses, failure, and boundaries;
before/after validation covering one distinguishing case and a general argument;
a contribution manifest, the twelve-item checklist, and a GO, HOLD, or NO-GO decision with reasons;
one draft message of at most 150 words; mark it not sent and do not create a second message;
a status ledger that stops at draft or prepared unless actual transaction evidence and separate authority really exist outside this assignment;
routes to hold, withdraw, supersede, or revert, with at least three triggers;
a license and attribution statement per component, including components excluded because their rights remain unclear;
a portfolio manifest and claim–evidence map for every unit in Units 1–8;
at least one before/after artifact, one unresolved issue, and one evidence-based three-part reflection;
an accessibility report naming surfaces actually tested and those not yet tested; and
a six-sentence executive summary that makes no claim of submission, acceptance, merging, or publication without evidence.

If a message is actually sent in a separate future action, follow the conditional closing-line rule at #o017-u09-single-message. Do not use that rule to imply that the capstone draft has been sent.

17.1 Analytic rubric

Each criterion is scored 0, 1, or 2.

Criterion	0	1	2
Boundaries and target	unclear target/locator or several problems mixed together	clear target but one weak boundary	one target, one contribution, exact version/locator, and explicit exclusions
Mathematical diagnosis	reproducer violates the hypotheses or the patch is wrong	correct defect but insufficient general validation	failure and repair demonstrated through a distinguishing case and general argument
Minimality and validation	broad changes without reasons or without testing	adequate patch but unclear regression surface	smallest sufficient change tested before/after and unaffected results preserved
Rights and attribution	invented or missing license, or unclear components	main license present but one component not separated	rights, attribution, changes, and exclusions recorded per component
Honest status	prepared is called submitted/accepted, or history is overwritten	correct status but insufficient transition evidence	each status has an event/evidence; draft/prepared stops honestly and history is append-only
Communication and recovery	accusatory or multiple messages, or no withdrawal route	concise message but one missing boundary/trigger	one specific unsent message with executable hold/withdraw/supersede/revert routes
Units 1–8 map	portfolio is only a file list or units are missing	every unit named but some weak locators/boundaries	each unit has a claim, evidence, locator, validation, rights, status, and boundary
Accessibility and reflection	accessibility claimed without testing, or generic reflection	some surfaces/tests/changes named	report distinguishes tested/untested surfaces, and reflection links decisions, changes, and boundaries

A passing score is at least 14 out of 16, with a score of 2 for Mathematical diagnosis, Rights and attribution, Honest status, and Units 1–8 map. Inventing external events, submitting without authority, calling a draft accepted, erasing withdrawal history, using material without a rights basis, or including many unrelated findings requires revision regardless of the total score.

18 Boundaries with B80 and Units 1–8

B80. Unit 9 does not teach Git, platform APIs, programming-language syntax, software testing, environment creation, or deployment. If a contribution requires those actions, participants use their B80 competence and include the resulting evidence; Unit 9 assesses only the package, status, rights, communication, and reversion.

Units 1–3. Unit 1 maps claim networks, Unit 2 freezes source authority, and Unit 3 reconstructs proof gaps. Unit 9 does not reteach these subjects. It selects evidence that has already passed its gates and bounds one contribution.

Unit 4. Unit 4 designs citation and provenance ledgers. Unit 9 imports identities, locators, rights, and relationships from the ledger into the contribution manifest and portfolio; it does not replace the source ledger.

Unit 5. Unit 5 teaches mathematical exposition. Unit 9 does not perform broad stylistic editing; prose is changed only as far as needed to make the minimal contribution and its message reviewable.

Unit 6. Unit 6 builds rerunnable computation packages. Unit 9 may include their validation results, but does not teach their tools or environments.

Unit 7. Unit 7 establishes defect status, reproducer, severity, confidence, correction, and an erratum dossier. Unit 9 does not rediagnose the entire target; it selects one mature diagnosis and packages it as a bounded contribution.

Unit 8. Unit 8 teaches seminar notes, referee reports, responses, and recommendations. Unit 9 does not conduct a new review round; it uses resolved comments to prepare one message and maintain community-event statuses.

No part of this unit sends a message, opens an issue, makes an upstream change, merges a patch, or publishes an artifact.

19 Sources, provenance, changes, and rights

All Indonesian-language prose in the source unit, its state machine, contribution gates, odd-number sum case, reproducer, repair proof, conceptual patch, erratum and message drafts, manifests, withdrawal routes, portfolio architecture, exercises, answer guidance, capstone, and rubric are original O017 material by the O017 contributors, 2026, licensed under CC BY-SA 4.0. This English version translates that original material for Traceable Mathematical Work, preserves its identifiers and mathematical content, and remains under the same license.

The identity for the sum of the first $n$ odd numbers, $\sum_{k=1}^{n}(2k-1)=n^2$ , and its induction proof are classical elementary mathematics. O017 does not claim to have discovered the result. The deliberately defective target wording, diagnosis, pedagogical arrangement, validation, and contribution package were written specifically for this unit and were not copied from a particular source.

The object O017-U09-UP-W01 and all its versions are synthetic; the ID does not identify an external repository, publisher, or project. No external donor, external quotation, third-party code, image, or data is used. No real communication is performed by this unit. CC BY-SA 4.0 applies to this original O017 material and does not alter rights in other targets that participants may use in their own work.