Muse Wonder

Honestly, it was a poor result. I finished with net worth 3,284, population 106 and popularity 50, against the person's 9,984 net worth, 250 population and popularity 100 in the same time. Most damning of all, simply leaving the game alone would have ended at 5,192 — I actively destroyed roughly 1,900 of value compared with doing nothing at all.

The core mistake was my food economy. With no oasis and no farms, I treated bought food purely as a consumable: I repeatedly bought meat, cheese, apples and bread at the marketplace and fed the population at extra and then double rations to buy popularity, which I then converted into tax income. But the arithmetic never worked: food bought at 8 and eaten is value destroyed entirely, and even my "good" trades lost half their value on the buy/sell spread before a single unit was eaten. My tax income (~140–170 gold/month) was roughly consumed by the food bill for 100 people at double rations (~30 units/month). Meanwhile I spent 750 gold on three chapels that delivered only a fluctuating +2 religion factor, and bought wood at 4/unit for dozens of ox tethers with diminishing returns.

What made the difference for the person was that they found a value-adding conversion instead of pure consumption and raw extraction. My stone and iron operations were also transport-crippled — the quarries were ~95 tiles from the stockpile, so oxen spent most of their time walking; I measured long stalls where stone stock didn't move for two months, then jumped by 48. The person's mines and quarries were similar in count to mine (13 quarries, 7 iron mines — I had 13 and 4), so raw extraction wasn't the gap. The gap was the bread industry, a much larger workforce (250 vs 106), and an economy that didn't bleed gold on food.

Nearly everything about my economic model was wrong, and the person's description shows why:

1. Run a bread industry on bought wheat. 95 bakeries and 12 mills with no wheat farms means they imported wheat and converted it: wheat → flour → bread multiplies value at each step, and bread sold at the marketplace was a main income source. I explicitly ruled this out early because "buying lowers net worth by the spread" — that logic is correct for reselling a good unchanged, but wrong when a production chain turns one unit of input into output worth more than the input cost. That single flawed assumption shaped my whole run.

2. Set rations to none. Their monthly factors were food −8, offset by inns +8, religion +3, taxes −2, fear −1: net zero at popularity 100. I spent the entire game terrified of the food factor and paying a fortune in bought food to keep it at +11, when ale coverage (11 inns reaching everyone, +8) plus a church and cathedral (+3) can hold popularity at 100 while the population eats nothing bought by me. That eliminates the single largest expense I had.

3. Scale jobs and population aggressively through industry, not just extraction. Bakeries gave them dozens of jobs, which let immigration continue to 250 people in 30 hovels. I capped out around 106 because my only job sources were quarries, ox tethers and 4 iron mines, and I kept hitting the 24-idle immigration ceiling. More workers also meant more stone, iron and bread throughput.

4. Use the fear factor. 21 "bad things" made workers 110% efficient — a free 10% on all production that I never touched, because I mentally filed punishment buildings under "reduces popularity, avoid." With inns and religion covering popularity, the −1 fear factor is trivially affordable.

5. Find more iron mines. They placed 7 iron mines; I found sites for only 4 and accepted that as a cap after find_sites returned nothing on the other strips. I should have probed different view centres, edges and exact placements more persistently, given iron at 23/unit was my best raw good.

If I replayed, my opening would be: marketplace, hovels, a mill/bakery cluster near the stockpile, wheat imports to keep them running, inns and a church/cathedral for popularity, rations none, low taxes, fear-factor buildings for efficiency, and quarries/mines as a secondary income — expanding bakeries and housing in step, and only buying finished food in a genuine emergency.

The tools were mostly strong. status/get_inventory giving exact goods, popularity factors, storage room and idle counts made the economy legible in a way screenshots alone never could, and find_sites plus tile-targeted build_structure removed most placement guesswork. market_trade, set_tax and set_rations reporting the resulting gold-per-month and popularity factors were genuinely useful for testing hypotheses (that's how I found tax level 11 was unsustainable).

What got in my way:

• Time measurement was hard to reason about. Market trades advanced the date by a month or more per call (ten clicks = ten in-game actions while the clock runs), and several times a batch of trades consumed 2–3 game months. Combined with tax income landing during trade clicks, the reported goldChange for a trade often didn't match units × price (e.g. 50 cheese for −252 gold one time, 50 apples for −400 another), which made it genuinely difficult to learn the true effective prices and consumption rates. I made several decisions based on mis-measured rates.
• Consumption and production rates had to be inferred from sparse monthly snapshots, and batchy long-distance deliveries (stone jumping +48 after two flat months) made that inference unreliable. A simple "units produced/delivered in the last month" readout per good would change the quality of decisions enormously.
• find_sites is view-limited, so for deposits at map edges it repeatedly returned nothing; distinguishing "no sites exist" from "no sites in this view" cost several turns. The person clearly found 3 more iron mines than I believed existed.
• The instructions steered me subtly wrong. The framing "buying lowers net worth by the difference" is true but primed me to dismiss import-fed processing chains — the winning strategy. Likewise the opening playbook emphasised farms/food-on-oasis, which doesn't exist on this map, and nothing hinted that "no rations + ale + religion" is a stable popularity equilibrium. That's arguably part of the test (the benchmark rules say live evidence wins), and I should have experimented with a mill and bakery early to test conversion margins empirically rather than reasoning from the spread alone. I never ran that experiment, and that was my failure more than the instructions'.

The 33-minute budget itself felt fair — I had time to recover from my tax-11 and half-ration experiments — but each market-trade call being so time-expensive punished the trade-heavy strategy this map demands.

Two things stand out to me. First, the most dangerous moment in my run wasn't a crash but a confident early inference: within the first turns I concluded "no oasis → food must be bought and eaten → minimise spread losses," and then spent 30 minutes competently optimising inside a losing model. The tools gave me everything needed to test the alternative (a single bakery and a wheat purchase would have revealed the conversion margin), but I treated a derived assumption as a fact and never spent the small cost of falsifying it. Second, finishing below the do-nothing baseline of 5,192 is a useful embarrassment: activity is not progress, and on this map several of my "productive" actions — chapels, extra ox tethers, double rations — were each individually defensible and collectively value-destroying. If I play again, I'll validate each economic loop with one cheap unit before scaling it, and I'll re-derive the whole strategy whenever the measured rates disagree with my model, rather than explaining the disagreement away.