Knowing enough to act together · Essai 03
What should the surveyor check?
The museum whale is ready to move. The surveyor can inspect which passages are open. The handler knows where the exhibit fits. What would let the handler choose a passage?
Try the handover
Six reports from four passages
There are fifteen nonempty sets of open passages. If both people know a common route exists, six different report meanings suffice. A surveyor can produce those six meanings after three inspections. Checking the same three passages every time needs eight report meanings. Choosing the third inspection from the first two answers brings that down to six. If the handler must also recognise blockage, all fifteen sets must remain distinguishable in a one-way report.
The four-passage handover
What reaches the handler?
In the starting example the building permits A and C. The exhibit fits B and C. The surveyor checks A, B and D. Can the handler justify taking C even though no one checked it? Change either menu to see when the answer changes.
Delivery and authority
The inspection
The report
The handler’s choices
Which building layouts can the handler still believe?
Compare arrangements for this same move
| Arrangement | Checks | Questions sent | Replies sent | Decision |
|---|
One question names one passage. One reply may contain several readings. These counts describe this move. They do not measure time or the length of each message.
The calculation could not be loaded. Open the worked argument.
A two-passage test
Can the person questioned know the answer?
Someone proposes asking the building surveyor, “Can this exhibit use East?” Hold the building open on both sides. Switch which passage the exhibit fits. The right answer changes while the surveyor sees the same building.
Checking the exhibit as well as the building makes the joint answer available at its source. There is a cheaper change in this two-passage case. Let the surveyor answer only “Is East open?” and let the handler combine that answer with their own fit list. Given the promise of a common route, that suffices in all seven admitted cases. A known usable service passage in the four-route experiment changes the physical arrangement instead. Choosing a passage settles neither completion of the move nor permission to reopen the museum.
What the calculation buys
When blockage is possible, the direction of the conversation matters. Across all 225 building–exhibit pairs, a handler who chooses the questions can settle every case within three inspections. A three-inspection one-way survey settles 187. Try a case where two questions settle it. Compare the work of the five arrangements.
With four passages, six and eight messages both fit in three fixed-length bits. With nine passages, an adaptive survey reaches the optimum of 126 message meanings in seven bits. The fixed survey has 256 meanings and needs eight bits. Explore the nine-passage construction.
A checklist can hand over the full building survey. The finite calculation tests which inspections can be omitted, when two results may share a report, and whether the recipient can act across every layout still possible to them. The number of inspections, the number of possible report meanings and the number of exchanges measure different kinds of work. The museum has supplied the possible layouts and the rules of communication. Actual briefing time and practical usefulness remain open.