The new Compare experience puts nine approaches to work on the same fictional broken footbridge. One exposes agreement between pairs that cannot hold jointly; another shows how equal progress can leave different futures; another asks whether a feasible design can actually be reached. Each perspective has an executable example. The “best for” judgements are provisional and specific to this comparison.
The Causal Lab now lets readers challenge whether team capacity influences automation adoption, and see how that assumption changes the required adjustment. It checks the consequences of a supplied causal story, not whether the story is true.
Solway now connects intended use to declared handover conditions and required work, then explores keeping work breakdowns and dependencies coherent as work changes. Releasing the next crew after Unit A’s acceptance, rather than after the whole inspection package, permits 16 action orders instead of 12. A coarse boundary can impose unnecessary order. These are finite action orders, not timed or resource-feasible schedules.
The drawing that forks moves a drawing revision before or after copies are issued. The wall and roof can receive conflicting revisions even when the diagrams have the same outer types.
Gimmer’s open-net reference shows what assembling work packages leaves to specify: initial resources, selected work and repeatability. Separately, the Island Reading Room check agrees with a second mathematical implementation on 48 exterior inputs, after explicitly translating wire order. These checks make assumptions visible; they do not establish physical fit or a project plan.
Corrections also matter. The currying example now compares whole functions, fixing an earlier counting error. The flight-estimate example now explicitly describes its daily-maximum reconciliation policy, rather than claiming formal sheafification.
Nodes and edges compares two ways to make relationships into objects. One follows dependencies but loses isolated tasks and some structural distinctions. The other retains tasks and relationship records, useful when a dependency needs its own reason, owner or evidence. Neither turns the network into a work breakdown structure.
The refreshed reading map distinguishes where sources were filed from how particular readings inform an enquiry, with contributions and limits stated explicitly.
Also useful: Working views has expanded and been curated, offering visual routes into existing examples. The Idea notebook gained manual JSON backup and restore for ideas, comments and categories. Records remain in the user’s browser.
One workable choice finds that a dependable route for the museum whale can remove the need to resolve which other passage is open. Related extensions test when resource changes invalidate a summary, when the same exact rule remains adequate, and when knowing the process changes a deadline claim. Choosing a route remains distinct from completing the move.
These are bounded, inspectable models; improved outcomes on real projects remain unproven.
Coverage expanded on 10 October: the initial version focused too narrowly on that day’s related experiments.
]]>