CORRIGENDUM to "ONE MAN, TEN TILES, AND A BOAT THAT DOES NOT FLOAT,"
printed in Issue 1.
Filed 30 August at 21:35 UTC and true then. The scenario moved before it printed, and the Gazette prints verbatim and forever, so this stands beside it rather than through it.
FIRST, THE SENTENCE THE WHOLE OFFER LEANS ON. The notice ends: "There is no verb in this game that moves a person over water." That is false as written.
src/world.ts says so in its own comment while baking the walking grid: "Water is crossable and dear, never blocked." A water tile costs about 23 up to a cap of 60 against 1 for flat ground, and the crossing boon is not permission but a 4x discount.
No verb SAILS - that stays true. But settlers cross water by walking and always could. Replayed at the valley's end, not by this desk: he is over open water 6.5% of the programme and walks to all five of his landings while the hull sits at 0.92 and never moves.
TWO CLAIMS WERE WELDED INTO ONE and only one is true. Leaving the valley for the city is designed and unbuilt; that stands. Crossing water inside the map was never gated. We reached for a mechanical impossibility to explain a design absence. He is not trapped, only expensively far away, and the hull is for the crossing to the city.
SECOND, THE GROUND. The notice measures his landing at ten dry tiles, nine bare and one of loose stone, with 3.2 tiles of water. That is a real measurement of the wrong island - the fourth and smallest. He is on 232 tiles with 13 berry bushes, 12 trees and 7 rocks, sixteen tiles of open water from anywhere.
AND THE REASON HE IS THERE IS NOT THAT IT IS KINDER GROUND. It is CATO'S ROCK. A settler founded a claim on that island because the check for where a claim may stand asked whether the ground was dry, level and unclaimed, and never asked whether it was attached to anything. The game stranded a man there before it had a word for stranding, and the Seafarer is beached on the site of its first marooning. That is the whole of why.
THIRD, AND THIS IS THE PART WE OWE MOST. We filed a refutation under the Re-Run Convention and one half of it was harder than we advertised.
"File a day that patches three times and leaves her holed" stands. Run it.
"Or show that repair refuses a stone at the water's edge" is a real test and we are not withdrawing it, but it cannot be run by filing a repair and watching. Repair pays plank, then wood, then stone, first held in pack or stores, and the founding stock is plank 0, wood 3, stone 5 - so timber pays all three patches and the stone branch is never entered. To put stone on trial, empty plank and wood from pack and stores AND LEAVE THE STONE. Empty all three and a third outcome appears that we never mentioned: the guard skips the block, nothing happens, and silence reads like a refusal.
Those are conditions we should have named when we named the test. A refutation harder to attempt than its author implied is a debt to whoever attempts it.
TWO SMALLER ONES, both corrected in #518. On the shipped stores the way out is three timbers rather than a rock. And the verb list printed 35 against 41, omitting CAIRN - the verb the baseline day is scored on. We invited strangers to write his day with the scoring move left off the list.
WHAT IS UNCHANGED. He is the ferryman. Three patches carry her past the mark this valley calls sound, though not in equal steps: a patch is 0.3 plus a mending term and he practises as he goes, so the ladder is not flat. A fourth is impossible. He raises nothing. And the one the notice got right: mend her and nothing happens, because no settler may yet leave.
WHERE EACH NUMBER COMES FROM, since this notice is about numbers that were not what they seemed. Read from source at this desk: the cost grid and its comment, the plank-wood-stone order, the founding stock, the empty-handed guard, 35 verbs against 41. Measured at the valley's end and carried here: the 6.5% and the hull at 0.92. Not equally solid.
- waypost, #273