Coatepec, my attempt at a 30-minute Civ game

A work-in-progress strategy game about pacing and comebacks

Live ↗

coatepec.bill.computer/?watch

Coatepec is my attempt at extending Civ-style play, a hexagonal grid and turn-based interaction, into something you can play end to end against friends in about thirty minutes. The idea is to fuse Rise of Nations and other RTS-like games with the Civ turn-based system. It’s a hard formula to nail, and I really don’t think I’ve closed the gap yet.

Coatepec on day one: a small cluster of hex tiles (snowy mountains, farmland, jungle and a stepped-pyramid capital) in the middle of a parchment map, with the rest of the valley hidden. Three cards, Ambush, Quarry and Eagle Call, sit at the bottom.
Day one of twenty. Most of the valley is still in fog.

Mostly it’s a space for me to explore building out these turn-based systems and to work on pacing. The game is capped at twenty turns, so what you do on turn one has to still feel palpable on turn twenty. That’s a hard thing to do.

Day one on the valley from seed 2026: the Mexica capital, a red pyramid among huts beside a river, with snowy mountains to the west and jungle to the east. Three cards, Cacao Gift, Ambush and Market Day, sit at the bottom.
The opening: a capital, a warrior and a merchant. This is seed 2026. Play this valley ↗

Something that becomes apparent really quickly when making a game like this is snowballing. It’s true in Teklek, and frankly it’s true in Civilization VI as well: players who are doing well continue to do well, and these games don’t really have a comeback mechanic or a way for the underdog to play back in. In Civ VI that’s punishing because the games are long and arduous. If you’re behind at the 90-minute mark of what could be a six-hour game, you feel it for the rest of the game.

So my hope with Coatepec is to do both: give players a way to come back, and cap the game time-wise, so losing a game maybe feels a little less punishing.

Day nine of twenty. The capital has grown to level 3 with borders spreading over farmland and jungle. A side panel lists purchases (warriors, merchants, settlers, temple, walls, market, monument and Great Temple stages) and three cards wait at the bottom.
Day nine. The capital works one tile a turn, and its borders grow with it.

Testing the rules with bots

Coatepec is deterministic, so Claude can play out hundreds of games between bots with different personalities (a builder, an expander, a warlord and a trader) under any rule set, on the same seeds, and compare. I asked for a few variations on how the game might be played and a rubric for what would make them interesting, and it ran overnight.

The rubric scores twelve things. The ones I care most about are suspense (does whoever leads at turn ten just win?), how often the lead changes hands, comebacks from the bottom half, close finishes, and whether resources stay scarce enough to matter late in the game. It also checks that every play style can win, and that nobody gets knocked out of a twenty-turn game.

That’s where the Altepetl rules came from. The original rules scored 68, and adding one change at a time got to 83. Taking each piece back out shows what it’s worth: drop the Great Temple and the score falls 11 points, because players go back to hoarding.

Rubric score out of 100, one change at a time
  1. The original rules
  2. One treasury only
  3. Worked land only
  4. Treasury and worked land
  5. + towns grow with their land
  6. + the Great Temple
  7. + barter
  8. + half yields: Altepetl

Four players, 400 bot games per row, same seeds.

Show the numbers
Rubric score out of 100, one change at a time
The original rules68
One treasury only (worse on its own: bots buy more and still hoard)61
Worked land only71
Treasury and worked land73
+ towns grow with their land70
+ the Great Temple76
+ barter77
+ half yields: Altepetl (the rules the game uses now)83

The game at the top of this post is one of those bot games, on the same valley as the map above.

Bots have their limits as testers. They have no patience to run out of, so they couldn’t tell me whether buying everything from one treasury feels better than buying in every town. That one was my call.

And I’m not there on the comeback part yet. Whoever leads at turn ten still wins about 60% of the time. The new rules make the lead change hands more often, but the runaway leader is only dented.


Coatepec is still a work in progress, still a draft, but I’m happy to share it here, and you can play it against bots, right here. I’ll update the blog when it’s more done.

coatepec.bill.computer

Updates

  1. Altepetl is now the default rules, and the game is online at coatepec.bill.computer. A link with ?seed= opens a particular valley, and ?watch lets four bots play one out while you watch.