Balance findings
Before any real players exist, the project runs the game itself against scripted players and reads what comes out. The simulator opens the real prototype in a headless browser, gives each of five bots its own page, advances a stubbed clock in twenty-minute steps, and shares mill yards and business counts between the pages so the bots affect one another. Each strategy uses only the actions a human has. This page records what three five-day runs showed, how far to trust them, and what question they leave.
The five bots
Section titled “The five bots”The logger cuts the best stand it can, sells to the best bid and proves up. The forester cuts only mature timber, builds a road to the back of the claim and replants white pine. The miller saves for a sawmill near the landing, sets a bid to buy neighbours’ logs and upgrades. The entrepreneur opens businesses in underserved markets. The casual bot follows the logger’s rules but acts only every eight game hours. The score is net worth: lumber in hand, buildings counted at cost, and any logs counted at 80 board feet each.
The results
Section titled “The results”| Bot | Run 1, early rules | Run 2, after balance fixes | Run 3, current rules (v16) |
|---|---|---|---|
| Forester | 111,654 | 78,184 | 127,561 |
| Entrepreneur | 128,000 | 76,179 | 58,579 |
| Miller | 62,371 | 72,618 | 50,523 |
| Casual | 25,857 | 28,009 | 27,182 |
| Logger | 27,661 | 27,925 | 18,331 |
| Leader over casual | 4.95 times | 2.79 times | 4.69 times |
| Leader over weakest of top three | 2.05 times | 1.08 times | 2.52 times |
Runs 1 and 2 used an older version of the game: an 80-minute crew rest, wagon trips of four to eight minutes, a deed that required residence time, and no Forest Reserve, farming or fair. Run 3 uses the current rules, with a ten-minute crew rest, a production-based deed and the new systems present but not exercised by the bots.
What each run showed
Section titled “What each run showed”Run 1 found three problems. The entrepreneur won easily because opening a business was too cheap. The logger clear-cut its claim by the end of the first day and then had nothing to do. The deed took under two hours. The casual bot trailed the leader by about five times. The fixes were business costs, higher upkeep, a cost of goods and prerequisites, a gauge warning when the forest runs low, a residence requirement for the deed, and changes to how often neighbours sell logs to your mill.
Run 2 looked good. The three leading strategies finished within roughly eight percent of one another, the miller had gone from buying 159 neighbour logs to 1,096, and the casual gap shrank below three times. That was the picture at the time.
Run 3 does not hold it. The forester ran away with 127,561, more than twice the entrepreneur and the miller, and the casual gap is back to about 4.7 times. Two things in the data stand out. The forester replanted 315 times against 183 in run 2, and the working explanation is that replanting and the faster growth timers compound: a forester who keeps planting has stands coming ready again and again, and the shorter crew rest lets it keep working them. That is an interpretation of the numbers, not something the run isolated. The entrepreneur’s cedar shake business finished at minus 5,881 board feet, which cost it roughly a tenth of its total. The logger sits flat at 18,331 from hour 24 onward, because it has cut every stand and will not replant beyond the two parcels the deed requires.
A one-day run on the current version adds a warning about reading short runs. At hour 24 the miller led with 30,880 and the forester stood at 22,905, behind the miller and only just ahead of the entrepreneur at 22,076. The forester’s lead is entirely a matter of compounding, so a test that stopped after a day would have picked the wrong winner. The casual bot at that point had not yet proved up.
What the casual gap is made of
Section titled “What the casual gap is made of”The casual bot is a logger that looks in every eight hours. The crew is three people and each is back after ten minutes, so the crew refills in half an hour and cannot bank more than three days of work. A player who is away for eight hours gets three cuts when they return, the same as if they had been away for thirty minutes. Time away costs little only if there is nothing to catch up on; here the leader is acting continuously and the casual player is capped. That reading comes from the rules, not from a separate experiment, but it matches the numbers.
How far to trust any of this
Section titled “How far to trust any of this”The simulations are a warning light, not a verdict. The bots are scripted and do not adapt. There is one seed, so no measure of variance exists. The bots do not farm, do not buy Reserve stands and do not enter the fair, so the sinks and incomes those systems add are missing from the score. Net worth counts holdings, which means that a pile of unspent lumber scores as well as a well-run town, and in run 3 the forester’s 127,561 includes 119,041 board feet of lumber in hand. Whether that signals strength or a sink shortfall is discussed in Economy, sinks and one-way money. Net worth also says nothing about whether the player had a good time, and the simulator’s own notes are clear that bots can test economy stress, exploits, onboarding clarity and bugs but cannot say whether the game is fun.
The bots also share one board that human players will not share in the same way: each bot is its own town, with the others standing in as competitors, which is a simplification of the market in Pick a lane.
The open question
Section titled “The open question”The central open question after these runs was whether the casual player needs a catch-up mechanic, along with farming against timber income parity, the width of the Reserve price band, and business profitability after the third milestone. The data show a casual player ending at about the same level in every run, near 27,000 or 28,000, while the leaders vary a great deal. That suggests the floor is steady and the ceiling is what moved, which can be read two ways. It can mean the casual player is protected by outside buyers and the passive growth of timber, so the gap is a reward for engagement. It can also mean that the gap is wide enough that the lead builds past anything a casual player could close. No catch-up design had been settled at that point, and a mechanism would collide with the pillar that the studio does not steer outcomes. The later sections of this page test some options.
Rerun before rebalancing
Section titled “Rerun before rebalancing”The project’s guardrail says to run the simulator before rebalancing and to compare the result with the files under sim/. The raw results are results-before-balance-fixes.json, results-after-balance-fixes.json and results-v16-5day.json. The one-day file confirms that the simulator runs on the current version and should not be used for balance on its own.
Lane experiments, 2026-10-08
Section titled “Lane experiments, 2026-10-08”The first v16 run left two questions open: whether replanting really explains the forester’s lead, and whether the Reserve and farming are competitive routes. Three new bots were added to the simulator (a forester that never replants, a Reserve buyer and a farmer) and run on the same five-day clock.
| Bot | Net worth after 5 days | What happened |
|---|---|---|
| Forester | 128,118 | Replants and keeps cutting |
| Farmer, 12 fields | 37,544 | Hay and wheat, sold at the Mercantile |
| Casual | 27,182 | Logger rules, checks in every 8 hours |
| Forester, no replanting | 28,464 | Nothing left standing, flat after day 1 |
| Reserve buyer | 17,574 | Bought 4 stands, then clear-cut and stalled |
Replanting is the whole story. Take it away and the forester finishes level with the casual player. The forester’s runs differ by about 1% from one run to the next, so a gap of 100,000 is not noise. The practical reading is that this version of the game rewards one behaviour, keeping land in timber, far above everything else, and that every other lane is a way to earn while the forest grows or a way to avoid the work.
The Reserve did not turn out to be the loophole it looked like on paper. Early on, with most stands ready, they cost about 63% of log value and resell above that. But simulated settlers buy up to 2.4 stands an hour once the town is full, so the Reserve is mostly sold out within hours and the price rises toward full value. The opening window is a windfall of a few stands and then it closes. Whether that window is intended is an open question.
Farming pays, but per crew-day it earns well under timber. A harvest uses a crew-day and returns roughly 280 to 400 bd ft for the crops in the game, where a mature timber parcel returns on the order of 1,000. The farmer needed 12 fields to beat the casual player and still finished at under a third of the forester. This is one bot with a simple rotation, so it is a floor, not a verdict.
These are all single strategies in a five-bot world, and the farmer and Reserve buyer are first drafts. The results show where the pressure is, and they do not yet show what a human would do.
Era pacing and a catch-up for infrequent players, 2026-10-08
Section titled “Era pacing and a catch-up for infrequent players, 2026-10-08”Two questions were left open when the era layer was built. The first was how fast a shared town should move through the eras in real days, because the first table of thresholds was a clock in disguise: settlers arrive on a fixed schedule, so a settler count of 30 was met at 14.7 hours whatever anyone did. The second was whether a player who checks in every few hours can be helped without helping the player who is always there. Both were tested with the same simulator, with a few additions that matter for how to read the numbers below.
What changed in the simulator
Section titled “What changed in the simulator”The bots can now share a town for the era needs (TOWN=1), so lumber produced is averaged per bot and businesses and roads are summed, which is how a shared server would count them. Any strategy can take a check-in interval (forester_4h, miller_12h), the step can be halved to ten minutes (STEP_MIN=10), and variants can be composed from a small library, including a fast mode that squeezes the era clock into five days. A run with the same field of bots is deterministic: repeating a run gave identical net worths every time, so there is no run-to-run noise to subtract. The roughly 1 percent variation the forester shows across runs comes from changing which other bots share the market, and every comparison below keeps the field fixed. Where two fields are compared the text says so.
One caution about step size. At twenty-minute steps an active bot visits once per step, which undercounts a crew that rests under twenty minutes, so any rule that enlarges the bank looks like a gift to active bots. At ten-minute steps the active bots see the true flow. The tables say which step they used.
Pacing the eras
Section titled “Pacing the eras”The calendar floors stay at 14, 28 and 42 days after the previous era, which puts the latest arrivals at day 14, day 42 and day 84 of a town. The needs are the reward for a town that moves faster than that. The simulator’s foresters are close to the best a player can do, so their pace is an upper bound, and a real town is likely to arrive nearer its floors. The aim was that the first change never comes in the first two days, that a very effective town reaches each era at roughly a fifth to a third of its floor, and that a town can never be left behind forever.
Four candidate sets were run. Each combines a lumber need per player, a count of businesses and a count of roads, and the first keeps the 30-settler milestone. The earlier sets ran on the old crew rules.
| Candidate | Lumber per player | Businesses | Roads | Seven bots acting every step, era 1 | Bots checking every 4 hours, 14 days |
|---|---|---|---|---|---|
| A | 25,000, 80,000, 220,000 | 2, 4, 6 | 6, 10 for eras 2 and 3 | 44 h | era 1 at 52.7 h, no era 2 |
| B | 40,000, 120,000, 300,000 | 2, 4, 6 | 6, 10 | 96.7 h | era 1 at 96.7 h, no era 2 |
| C | 20,000, 60,000, 160,000 | 2, 3, 5 | 5, 8 | 31 h | era 1 at 40.7 h, era 2 at 208.7 h |
| D1 | 35,000, 120,000, 350,000 | 2, 4, 5 | 6, 10 | 79 h (era 2 at 296.7 h in a 28-day run) | not run |
D1 looked right until the catch-up change went in. With the bigger bank and the rested bonus, the same table moved a town of foresters to era 1 at day 1.5, era 2 at day 5.5 and era 3 at day 19.5, and a mixed town to day 1.7 and day 11.7. The catch-up speeds up output, so the lumber needs had to rise with it. The applied set, D2, is 60,000, 180,000 and 500,000 bd ft per player, with 2, 4 and 5 businesses and 6 and 10 roads for the last two eras. Run on the applied crew rules for 28 to 38 days:
| Town of seven | Era 1 | Era 2 | Era 3 |
|---|---|---|---|
| Five foresters, an entrepreneur and a miller, all checking in every 4 hours | day 2.5 | day 8.5 | day 28.3 |
| Forester, entrepreneur, miller, farmer, casual logger (4 h), casual forester (8 h), casual logger (8 h) | day 4.0 | day 20.2 | not by day 28 |
| Six foresters and an entrepreneur, all checking in every 8 hours | day 3.0 | day 8.4 | not by day 28 |
| Seven foresters checking in every 8 hours, nobody opens a business | day 14 (the floor) | day 42 (floor, not reached in the 28-day run) | day 84 (floor) |
| Seven bots acting every step, 5 days | none | none | none |
Two things stand out. A town of foresters who check in every 8 hours is almost as fast as one that checks in every 4, with an average of 144,000 against 148,000 bd ft per player at day 7 (the two fields are not identical, but close). Forester output is limited by how fast trees regrow, not by how often the crew can work, and the larger bank means an 8-hour visit now spends the whole crew. The second is that the businesses need does real work: a town in which nobody opens a business never meets it and arrives at each era on its floor. That is intended, since an era of steam and rails with no trade in town is a strange thing to announce, but it is a rule that has to be explained to players.
For the 5-day simulator, a fast mode compresses the floors by a factor of 20 and the lumber needs by 0.22, 0.145 and 0.08 (the town’s output is front-loaded, because the starting forest is cut in the first day, so one factor cannot space the changes out), and drops the settler need because compressing the settler clock would distort the markets. It puts the eras at 8.7, 32.7 and 72.3 hours with the applied crew rules.
The catch-up options
Section titled “The catch-up options”All options below were tried first as variants with eras off and a seven-bot field (logger, forester, casual logger, farmer, entrepreneur, miller, casual forester), then with fast eras on. The leader is always the forester. The casual logger is the bot the targets name; the casual forester is the same strategy as the leader at one-eighth of the visits, which is the fairer comparison.
| Option, eras off | Logger | Forester | Casual | Farmer | Entrepreneur | Miller | Casual forester | Leader over casual | Leader over casual forester |
|---|---|---|---|---|---|---|---|---|---|
| Today: 3 crew-days at 10 min | 18,307 | 127,339 | 27,182 | 36,344 | 58,555 | 50,523 | 27,399 | 4.68 | 4.65 |
| (a) Bank 6 at 20 min | 19,500 | 127,660 | 28,373 | 36,861 | 59,767 | 51,720 | 38,515 | 4.50 | 3.31 |
| (a) Bank 9 at 30 min | 18,369 | 127,397 | 28,373 | 36,234 | 58,683 | 50,065 | 38,424 | 4.49 | 3.32 |
| (a) Bank 12 at 40 min | 19,500 | 127,660 | 28,373 | 36,861 | 59,767 | 51,720 | 38,515 | 4.50 | 3.31 |
| (c) Hired hands, 0.5 an hour, max 6 | 18,691 | 127,542 | 28,373 | 36,351 | 58,966 | 52,052 | 38,515 | 4.50 | 3.31 |
| (c) Hired hands, 0.25 an hour, max 6 | 18,691 | 128,313 | 28,358 | 36,063 | 58,966 | 52,052 | 38,771 | 4.52 | 3.31 |
| (b) Rested ramp 3 to 8 h, 3 cuts, +70%, no bank | 18,307 | 127,511 | 46,090 | 36,344 | 58,984 | 51,477 | 49,791 | 2.77 | 2.56 |
| (b) Rested cliff at 8 h, 3 cuts, +70%, no bank | 18,307 | 127,511 | 46,090 | 36,344 | 58,555 | 51,477 | 49,791 | 2.77 | 2.56 |
| (a+b) Bank 6 + cliff at 6 h, 6 cuts, +50% | 19,500 | 127,660 | 42,861 | 36,866 | 60,196 | 52,401 | 63,615 | 2.98 | 2.01 |
| (a+b) Bank 6 + ramp 1 to 8 h, 6 cuts, +60% | 19,694 | 133,049 | 45,565 | 36,861 | 60,387 | 52,710 | 68,371 | 2.92 | 1.95 |
| (a+b) Bank 6 + ramp 2 to 8 h, 6 cuts, +50% | 19,561 | 129,781 | 42,861 | 36,861 | 60,029 | 52,401 | 63,615 | 3.03 | 2.04 |
| (a+b) Bank 6 + ramp 3 to 8 h, 6 cuts, +65% | 19,500 | 127,660 | 46,828 | 36,861 | 60,196 | 52,537 | 70,920 | 2.73 | 1.80 |
| (b+c) Hired hands + ramp 3 to 8 h, 6 cuts, +70% | 18,691 | 127,542 | 48,485 | 36,351 | 59,395 | 52,870 | 73,220 | 2.63 | 1.74 |
| (a+b) Bank 6 + ramp 3 to 8 h, 6 cuts, +70% (applied) | 19,500 | 127,660 | 48,485 | 36,861 | 60,196 | 52,537 | 73,220 | 2.63 | 1.74 |
| (a+b) Bank 6 + ramp 3 to 8 h, 6 cuts, +80% | 19,500 | 127,660 | 51,215 | 36,861 | 60,160 | 52,674 | 78,034 | 2.49 | 1.64 |
| Option, fast eras | Logger | Forester | Casual | Farmer | Entrepreneur | Miller | Casual forester | Leader over casual | Leader over casual forester |
|---|---|---|---|---|---|---|---|---|---|
| Today | 18,307 | 121,021 | 30,150 | 36,094 | 58,600 | 55,279 | 24,827 | 4.01 | 4.87 |
| (a) Bank 6 at 20 min | 19,525 | 117,566 | 31,782 | 35,642 | 59,812 | 56,150 | 35,834 | 3.70 | 3.28 |
| (c) Hired hands | 18,716 | 118,526 | 31,782 | 35,178 | 59,011 | 56,537 | 35,834 | 3.73 | 3.31 |
| (b) Rested ramp, no bank | 18,307 | 117,142 | 50,453 | 35,072 | 59,020 | 56,319 | 43,867 | 2.32 | 2.67 |
| (a+b) Bank 6 + cliff at 6 h | 19,525 | 114,667 | 48,087 | 35,403 | 60,232 | 56,893 | 56,322 | 2.38 | 2.04 |
| (b+c) Hired hands + ramp | 18,716 | 114,449 | 54,301 | 34,673 | 59,431 | 57,577 | 64,790 | 2.11 | 1.77 |
| (a+b) Applied | 19,525 | 114,100 | 54,222 | 35,349 | 60,232 | 57,190 | 64,790 | 2.10 | 1.76 |
The bank on its own lifts the casual forester by about 40 percent and the casual logger by 4. The logger-rule casual does not gain, most likely because it is not short of crew-days: with three cuts a visit it clears 45 parcels in five days, and the timber is finite, so more cuts per visit only reach the end of it sooner. The runs did not isolate this, and it is the reading that fits. The bank size does not matter above six (12 gives the same numbers as 6, and 9 differs by under 0.3 percent). The rested bonus on its own lifts both casuals, the logger by 70 percent, but the casual forester only reaches 49,791. The two together do more than either: with a bank of six, all six cuts of a visit carry the bonus, and the casual forester reaches 73,220, which is 57 percent of the leader’s score.
Hired hands, which keep banking a fraction of a crew-day while the whole crew sits rested, produced the same numbers as a bank of six in every column that matters (38,515 against 38,515 for the casual forester, 73,220 against 73,220 with the bonus added), and they need accrual state and an extra rule about when accrual stops. The bank is the simpler way to the same result and is the one applied. The difference between a bank of six and a bank of twelve vanishing suggests the casual forester’s remaining limit is the supply of ripe timber, not crew-days.
Exploit tests
Section titled “Exploit tests”The first test was a field of foresters differing only in how often they look in, from every step to once a day, in one shared market. A fair catch-up should leave this curve falling steadily, and a design that rewards checking in at the threshold will show a hump.
| Forester checking in | Every step | 1 h | 2 h | 3 h | 4 h | 6 h | 8 h | 12 h | 24 h |
|---|---|---|---|---|---|---|---|---|---|
| Today | 124,154 | 116,669 | 79,072 | 54,782 | 43,734 | 31,368 | 26,293 | 18,265 | 12,189 |
| Rested cliff at 8 h, 3 cuts, +60%, no bank | 124,458 | 116,597 | 79,864 | 43,881 | 31,616 | 45,465 | 31,338 | 17,468 | |
| Bank 6 + cliff at 6 h, 6 cuts, +50% | 124,613 | 119,943 | 112,664 | 67,201 | 80,376 | 63,059 | 46,000 | 25,163 | |
| Applied: bank 6 + ramp 3 to 8 h, +70% | 124,139 | 119,201 | 111,542 | 84,644 | 77,912 | 74,735 | 71,025 | 51,332 | 27,060 |
| Applied, fast eras | 121,602 | 114,110 | 104,462 | 81,598 | 70,493 | 68,545 | 65,289 | 49,690 | 31,658 |
| Today, fast eras | 120,937 | 108,846 | 75,259 | 50,683 | 39,722 | 30,443 | 25,012 | 18,719 | 12,379 |
The two cliff designs do show the hump. With a bonus that switches on at 8 hours and no bank, the forester who waits 8 hours scores 45,465, 44 percent more than the one who comes at 6 hours and more than the one who comes at 4. With a bank and a cliff at 6 hours, the 6-hour visitor beats the 4-hour one by 20 percent. A player who learns the threshold logs in on a timer, which is the pattern the bonus was supposed to reward only when it is real absence. The linear ramp removes the hump: the applied curve falls from hourly to daily, and no check-in interval beats working all the time. The one small wobble is the 3.5-hour bot (74,240) scoring 5 percent below the 4-hour bot (77,912); it checks in every 3 hours 40 because the interval rounds to eleven twenty-minute steps, so the bonus it earns is not that of 3.5 hours.
The second test is the active player who pauses. The bonus counts time since the last crew work, so a forester that waits for stands to ripen is “away” by that measure. With the ramp starting at 2 hours the active forester gained 1.8 percent (130,018 against 127,660), and starting at 1 hour it gained 4.2 percent (133,049). Starting at 3 hours the gain is zero: 127,660 with the bonus, identical to the bank alone. That is the reason for 3 hours.
The third is bursting. A bigger bank is the usual worry, because an active player could save up work. At ten-minute steps, the five active bots moved by plus 1.7 percent (logger), plus 0.4 (forester), plus 1.0 (farmer), plus 1.1 (entrepreneur) and minus 3.0 (miller) between today’s crew rules and the applied ones. At twenty-minute steps the same comparison reads plus 6.5, plus 0.3, plus 1.4, plus 2.8 and plus 4.0, but that step undercounts the old crew, as noted above. The forester does not move because it is limited by ripe timber. The miller’s 3 percent drop is most likely the market: the casuals now deliver more logs into the same yards, and that is the kind of difference a changed field produces.
Three further guards are in the rules and were not separately tested. The bonus covers timber cuts only, not harvests, Reserve stands or the sawmill. Hauling, building and planting do not restart the away clock, so opening the game for those does not forfeit or farm the bonus. And the six bonus cuts are all a rest period yields, so splitting them across visits gains nothing.
What was applied
Section titled “What was applied”A crew bank of six crew-days, each back after 20 minutes (CREW_MAX 6, CREW_REST 20 min), the rested ramp (RESTED: from 3 hours, full at 8 hours, six cuts, plus 70 percent), era needs D2 with the businesses need added and the settler need kept only in era 1, floors unchanged at 14, 28 and 42 days, and an explicit filing-shop: 1.0 in era 1 so the fade table has no gap. Hired hands were removed from the code after the test, and the option is kept as sim/variants/hands.js.
Regression with eras pinned off (the new mechanics against today’s, same seven-bot field, twenty-minute step; both runs repeated and identical):
| Bot | Today | Applied | Change |
|---|---|---|---|
| Logger | 18,307 | 19,500 | +6.5% |
| Forester | 127,339 | 127,660 | +0.3% |
| Casual | 27,182 | 48,485 | +78% |
| Farmer | 36,344 | 36,861 | +1.4% |
| Entrepreneur | 58,555 | 60,196 | +2.8% |
| Miller | 50,523 | 52,537 | +4.0% |
| Casual forester | 27,399 | 73,220 | +167% |
The two casual bots are where the catch-up is meant to act. The rest move by the step artifact and by market effects (at ten-minute steps, within 3 percent either way).
Scorecard against the three targets
Section titled “Scorecard against the three targets”| Target | Result | Verdict |
|---|---|---|
| Leader over casual at most 3.0 | 2.63 with eras off, 2.63 at real pacing (no era reached in five days), 2.10 with fast eras on. Today: 4.68, 4.68, 4.01. Against the casual forester: 1.74, 1.74 and 1.76. | Met |
| Every lane at least 50 percent of the leader | Fast eras: miller 50.1, entrepreneur 52.8, casual forester 56.8, casual logger 47.5, farmer 31.0, logger 17.1. Eras off: 41, 47, 57, 38, 29, 15. | Not met |
| No era change moves a bot’s net worth by more than 1.5 times | Era-attributable effect between 0.90 and 1.03 for every bot and every change. The forester’s raw change over the first era is 1.50 times, of which 1.03 is the era. | Met |
The era-attributable effect compares a bot’s growth over the six to nine hours after a change with its growth over the same hours in the eras-off run, so ordinary growth does not count against the era. The largest effect is the casual forester at the power-saw era, 0.90, which is the white pine demand falling to half.
The lane target fails for three reasons that the work here does not touch. The logger is not a lane: it clears its claim in the first day and stops, and no setting of the eras or the crew changes a bot that does nothing after day 1. The farmer sits at 31 percent of the leader because farming returns a few hundred board feet per crew-day against about a thousand for a ripe stand, and the farmer is not crew-limited, so the bank does not help it. The casual logger is short by 2.5 points (it needs 57,050 and has 54,222, which is 5 percent more).
Two exploratory runs, not applied, say what would close more. Making the white pine demand fall harder (0.9, 0.55 and 0.35 in eras 1 to 3 instead of 1.05, 0.7 and 0.5) takes the forester to 100,274 in the fast run, and then the casual logger is at 53 percent, the miller 57, the entrepreneur 60 and the casual forester 58 of the leader. Adding better farm yields and prices on top (yield 1.3 and 1.8 in eras 2 and 3, prices 1.1, 1.1 and 1.0) lifts the farmer from 35,482 to 43,549, which is 43 percent: closer, still short. Only the logger would remain far off, and that is a bot design question. These are single runs on one field and the levers were not otherwise explored.
What stays open
Section titled “What stays open”- How the server counts “active players” for the lumber need, and what happens in a town of two people. The need is per active player so a town of five and one of five hundred meet the same numbers, but a window of 7 days is a guess.
- Whether a town with nobody running a business should wait for its floor. As built it does.
- Whether the bonus should key off real absence (no session) instead of time since crew work. The simulator cannot tell the two apart, because bots have no sessions, and a player who opens the game to haul would reasonably expect the clock to run.
- The ceiling on catch-up. A casual forester now earns 57 percent of the always-on forester, and a once-a-day forester 22 percent. Whether that is the right gap is a taste question the simulator cannot answer.
- Forester dominance. Fast eras reduce the forester by about 10 percent and the harder white pine decline by about 12 more, and the forester still leads the miller and the entrepreneur by about a factor of two. The unknown is whether a human would find the forester strategy as quickly as the bot does.
- The era levers themselves are untouched from the first draft. This work tuned when the eras arrive and who they help, and it did not tune what they do.