The Dock Ledger That Lied, and the Map That Stayed True
Before sunrise, the ferry dispatcher lines up clipboards as small boats take on crates for the islands. Each island must send some crates away, but also keep enough coming back to run tomorrow. The table on the desk is the rulebook: islands are industries, crates are inputs, the table is what each needs. Takeaway: a whole economy can be tracked like this dock.
On paper, the dispatcher tries different day-one loading plans. One pattern is oddly fair: it keeps the slowest island moving as well as it possibly can. Start with most other patterns and, after enough days, one island is asked for crates it cannot have. The ledger dips below zero, and the day it happens can shift if you round a figure.
Then real life gets in the way. People on the islands use some crates and they never come back. Old planning tricks can force negative crates, which is nonsense on a dock. A cleaner rule works: each day, every island sends only part of what the table suggests and keeps the rest for local use. Growth slows, but the long-run balance stays in the same proportions.
The dispatcher still hates how touchy the crate ledger is. So the same plan gets rewritten as a chance map: from island A, a set of fractions says where a traveller would hop next, and the fractions add to one. Crate flows become hop chances, after a careful rescale. The shock is that the first place and day the ledger would go below zero stays the same, but the sums behave more calmly.
With the chance map, the dispatcher asks a new question. If a traveller keeps hopping day after day, which islands do they keep landing on? That long-run hang-time becomes a fingerprint of the whole network. A richer score also looks both ways, how much an island leans on others and how much others lean on it, so weak islands and pillar islands stand out.
Then the mayor wants a different balance of supplies, but without changing what makes the system steady or shaky. The dispatcher tweaks the crate table so the desired balance becomes the new best loading pattern. The chance map stays the same, so the warning signs do too. The first island to fail, and the weak-versus-pillar sorting, still match the old signals.
By the end of the shift, the sea is no kinder, but the desk feels steadier. The dispatcher still watches for the same real danger, the first impossible promise of crates. The difference is the working tool: less time staring at a brittle ledger, more time using an equivalent hop map that keeps the warning in the same place.