Abstract
Goal 5, 11g. The upgrades proposal (18 October) needs two traces before acceptance and seven days of public discussion. Trace 1, supply counted with nothing bought, needs no rule change: a new ignored test pays each team the proposal's rates from outside six bot-only matches on naval.map. Teams earn 48.8 supply a minute, at the top of the proposal's guess; every team has the first price of 60 by a median of 97 s; depots give 60% of it and kills 81% of the first two minutes; the leading side's supply runs away. The proposal gains a section 7 with the measurement and suggested changes for its author. The post for the discussion is drafted for Paul to post; this machine has no GitHub credentials beyond git over SSH, and posting is his. Trace 2 needs a prototype.
Index terms: upgrades, supply, proposal, trace, governance.
I. Introduction
A. Intent
Paul chose the next step: "Goal 5: upgrades proposal" (post the proposal for its seven days and write the traces of its section 6).
B. The prompt
ok. nice. the game is hard. continue.
Then, asked for the next step among a bot-skill scalar, the territory read and the upgrades proposal: Goal 5: upgrades proposal.
II. Decision Log
| # | Cause | Decision | Not taken, and why |
|---|---|---|---|
| D1 | "the game is hard"; no queued step | Asked Paul; offered a skill scalar, noting ROOKIES=n | Building a skill scalar unasked |
| D2 | Paul's choice | Goal 5, 11g | — |
| D3 | Trace 1 changes no rule | Supply paid from outside the world in an ignored test | A sim prototype first: trace 1 does not need one |
| D4 | The census's settings (LR 54 on) | The default brain in all sixteen seats, naval.map, 3 respawns, 30 min at most, six seeds, a thread each | No respawns: matches of 8 to 16 min would not show the late game |
| D5 | GOVERNANCE.md section 5: the author revises during discussion | Findings and suggested numbers in a new section 7; section 1.1's numbers left as Paul wrote them | Rewriting 1.1: the author's call |
| D6 | No gh, no token; posting is outward | The post drafted in build/supply/discussion-post.md for Paul | Posting by other means |
| D7 | Trace 2 needs buying | Named as needing a prototype behind Scenario::upgrades | Building it now: a rules change before the discussion |
| D8 | Three of my sentences overstated the data | Corrected before commit (15-minute total, "most matches", the halving's effect) | — |
III. Methods
fleet::tests::supply_trace (ignored): for each of SUPPLY_SEEDS (6) seeds a thread with a 256 MB stack; the census's world (Scenario with the fleet's teams, naval.map, SUPPLY_RESPAWNS 3, no time limit) run to SUPPLY_SECS (1,800) or the end; every seat fleet::brain. Every 10 s each side earns 1 a live pillbox and 2 a depot; every 20 s 1 a powered structure; 10 for a TankDestroyed of another team's tank by its own seat, 5 for a PillboxCaptured from another team. Printed: by side, supply by source, the seconds at 60 and 120, every 5 minutes; in all, supply a team-minute, the shares, the median second at 60, the match length. Log: build/supply/trace1.log.
IV. Results
As PROPOSAL-UPGRADES.md section 7: 48.8 supply a team-minute (7.3 to 92.5 by team); 60 by 18 of 18 teams, median 97 s; depots 60%, pillboxes 21%, kills 19%, captures under 1%, structures 0%; matches 18.0 min on average (16 to 23). Tests: brain 84 passed and 12 ignored; client 32; net 14; proto 3; sim 50.
V. Discussion
- The economy is richer than guessed. The first upgrade is in reach in a minute and a half, and a winning side could own the whole base list by the 20th minute.
- Supply follows the winner. The open question of a catch-up price now has evidence.
For Paul: two things are yours. (1) Post build/supply/discussion- post.md in GitHub Discussions under Proposals; the seven days start then. (2) Section 7's suggested numbers (halved rates, a bounty of 5) are suggestions; section 1.1 is unchanged until you say. Commits are local; nothing was pushed.