D6 · Forming: the retelling

Retell the whole model with its forms inside it

D6 brings the population model and the branch accounts into one explanation. A reader should be able to move from a broad idea to an organizational form, then to a community, and see exactly what each level adds.

Where this stage starts

C5 supplied the earlier whole-model account. The return to communities tested and revised that model in D1 to D3. D4 proposed shared forms, and D5 wrote their accounts. D6 starts with those records, including unresolved cases and proposals left by the branch reading.

What the aggregator does

The aggregator works through the broad dimensions of the model. For each part, it explains what each form does: adds local detail, restricts the claim to certain conditions, replaces a way of doing the work, or leaves it unchanged. The explanation names evidence and communities, rather than merely saying that a form “differs”.

The best examples are accompanied by a weak fit. If members of one form disagree, the aggregator considers whether this is ordinary local variation, unresolved evidence or a possible further distinction. It cannot simply invent a subform to remove the awkward case.

The whole account is then rewritten with those differences inside it. The figure should show where forms differ within the model, not place a separate list of clusters beside an unchanged diagram. The account also explains communities that required no adaptation and those the model could not explain, without assuming that either set has one common history.

Evidence about coordination continuing or ending comes from the records, with silence acknowledged. Canvas information can show an outcome, but any role it plays in interpretation must be declared.

There is a firm limit: the D3 model and D4 tree stay fixed. A new insight discovered while writing goes into a separate proposal. It has not inherited the earlier model’s validation. Applying it would require a further supported return to affected communities.

What the next stage receives

E1 receives the integrated retelling, its evidence links and a separate list of unvalidated proposals. It can audit what the completed run supports without mistaking ideas generated at the end for claims already tested during the return.