Five things we learned from five rounds of experiments on Barry Becker’s economy simulator about fixed-supply money, ordinary central-bank money, and what each means for the average citizen.
“Bitcoin”: a fixed money supply“Fiat”: central-bank money that grows
Based on Barry Becker’s agent-based simulator and five rounds of 20-run experiments · October 2026
A tiny economy you can rerun at will
Barry Becker is a retired software engineer and the author of the novel Into the Singularity. He built an agent-based economy simulator: a few hundred simulated households, dozens of firms, a handful of banks, a government and (under fiat) a central bank. They work, shop, borrow, save and go bust month by month for twenty simulated years. Nobody tells the economy what to do. Booms, slumps, inflation and deflation emerge from thousands of small decisions made every month.
That makes it a laboratory for a question that is usually argued in slogans: would ordinary people be better off with bitcoin-like money that can’t be printed, or with ordinary government money? Over five rounds of experiments, each run 20 times with different random luck, the answer kept changing. Why it changed turned out to be the most interesting part.
“The simulation cannot show that one monetary system is better in the real world. It shows which assumptions a conclusion depends on.”
Barry Becker
How to read the app screenshots
Each line is the median of 20 simulated histories. Hovering a month shows a readout like 0.82 → 0.67, which means baseline → scenario. Shares such as 0.15 mean 15%.
Two kinds of numbers
Across rounds come from Barry’s 20-run test suites, as written up in his round-by-round reports.
Live app numbers are read straight off the screenshots. The live site has evolved a little since the suite runs, so the two can differ slightly.
“Barry’s preset” is his richest setup: prices and output respond to demand, and there are bank loans, mortgages and thinly capitalised banks.
01
Finding one · Assumptions drive conclusions
The 52% that wasn’t
The very first round produced a headline any bitcoin fan would frame: under fixed-supply money the typical household was 52% better off. It was too good to be true. In that version, fiat’s money supply never actually grew while prices kept rising, so fiat households were slowly starved of cash. A separate “phantom inventory” leak also let bitcoin households consume goods that didn’t exist. Once Barry fixed both bugs, the landslide shrank to about +3%.
+52%
Bitcoin’s edge in round 1 (typical household’s real consumption)
Across rounds
+3%
The same edge after the two bugs were fixed (+2.8%, round 2)
Across rounds
0.82vs0.67
Typical consumption with working fiat vs fiat whose money growth is frozen, as in the bug
Live app
Across roundsHow to read: each bar is bitcoin minus fiat in one round (grey = round 1, blue = round 5). Left: living-standards gap in % (up = bitcoin better). Right: unemployment gap in points (up = bitcoin worse). The grey round-1 bars at +52% and +46% tower over every later round.Live appRe-creating the bug: solid = normal fiat, dashed = the same economy with fiat money growth set to 0. Follow the dark-brown median real consumption pair: it ends at 0.82 (solid) vs 0.67 (dashed), and the dashed line goes nowhere for 20 years.Live appSame comparison, measured as well-being: the Compare panel confirms the only change (Fiat money growth 1 → 0). Normal fiat (solid) climbs above zero while frozen fiat (dashed) sinks.
What about today’s numbers? On today’s default settings bitcoin still comes out about 11% ahead, but Barry’s round-5 report traces that gap to how the government rebates surplus taxes. With the rebate switched off it falls to +2.8%. So it isn’t a bitcoin win. It’s another assumption at work (see Finding 2).
So what
The most exciting result a model produces is often the most fragile. Check the plumbing before you celebrate.
02
Finding two · Deflation on trial
The deflation spiral was really the treasury stuffing taxes under a mattress
In rounds 2 and 3, Barry let prices respond to demand instead of following a preset trend, and bitcoin suddenly looked dangerous: living standards 12% lower and unemployment 4–5 points higher, a textbook deflation spiral. The real culprit was the government. Under bitcoin rules it kept collecting taxes and sat on the surplus, quietly pulling money out of circulation. Once the treasury gave back any cash it didn’t need, the spiral vanished. Prices still drifted gently down, but nobody was poorer for it.
−12%→0
Bitcoin’s living-standards gap, rounds 2–3 vs round 5 (+0.3%)
Across rounds
+4–5→0
Extra unemployment points under bitcoin, before vs after the rebate
Across rounds
0.94vs0.96
Typical consumption, fiat vs bitcoin, even though bitcoin prices fall ~2% a year
Live app
Before: one change at a time
Across roundsHow to read: round 3, bitcoin minus fiat. “S1” and “S3” are the two demand-driven setups, and the “+ surplus rebated” rows change only the treasury rule. Left: about −10% (S1) and −9% (S3) consumption shrinks to roughly 0% and −1.5%. Right: the unemployment penalty falls from 4.1 and 5.3 points to under 2.
After: today’s app with demand-driven prices
Live appBitcoin really does deflate: solid = fiat, dashed = bitcoin. The red inflation pair splits early and stays apart, with fiat just above zero and bitcoin just below. The summary line under the chart puts the final month at about +1.7%/yr vs −2.1%/yr.
Live app…and nobody is worse off: same run. The dark-brown median real consumption lines nearly sit on top of each other and end at 0.94 (fiat, solid) vs 0.96 (bitcoin, dashed).
Honest footnote: the treasury rule still decides the sign. When Barry’s round-5 tests switched the rebate off again, the bitcoin penalty came back (−10% to −12%). The lesson isn’t “deflation is fine”. It’s that the spiral came from money leaving circulation, not from falling prices themselves.
So what
Mild deflation on its own was roughly harmless here. What hurt was a government hoarding cash.
03
Finding three · Where the jobs go
Sticky wages are the real villain
Under Barry’s preset, bitcoin’s falling prices cost a lot of jobs. Deflation isn’t the problem by itself, though. The trouble is deflation plus paychecks that won’t come down. When prices fall but wages stay put, every worker gets more expensive in real terms and firms stop hiring. Let wages adjust and the jobs gap disappears. Make them stickier and it explodes.
15%vs9.6%
Unemployment, bitcoin vs fiat, with normal sticky wages
Live app
≈6.3%
Unemployment under both systems once wages can fall freely
Live app
0·+6.6·+21.6
Bitcoin’s extra unemployment points as wages go from flexible to the default to very sticky
Across rounds
Live appSticky wages (the default): solid = fiat, dashed = bitcoin. The orange unemployment pair splits almost at once. Bitcoin jumps above 12% within months and stays around 15%, ending at 15.1% vs fiat’s 9.6%.Live appSame switch from fiat to bitcoin, but with wages free to fall (nominal wage rigidity = 0): the two orange unemployment lines now overlap and end at 6.4% vs 6.3%.
Careful: flexible wages remove the jobs cost, not every cost. In the same flexible-wage runs, the typical household’s consumption still comes out lower under bitcoin. So the claim is narrow: the unemployment penalty depends on sticky wages.
So what
“Is deflation bad?” mostly turns into another question: how quickly can paychecks adjust?
04
Finding four · Barry’s point
It’s the printing press, not the coin
Is there something special about bitcoin, or is it just money that stops growing? Take ordinary central-bank money, freeze its growth at zero, and you get the same thing: prices slide about 7% a year, almost exactly as under bitcoin, and unemployment is as bad or worse. In this model, “bitcoin problems” are really fixed-money-supply problems. As Barry puts it, the real question isn’t coin versus coin. It’s whether politicians can resist printing.
−7.3%vs−7.2%
Yearly price change: frozen fiat vs bitcoin (Barry’s preset)
Across rounds
19.7%vs15.0%
Average unemployment: frozen fiat vs bitcoin. Normal fiat: 8.4%
Across rounds
~24%
Frozen fiat’s unemployment peak, around 2034, vs bitcoin’s ~15%
Live app
Live appHow to read: solid = bitcoin, dashed = fiat with money growth set to 0 (two changes: regime bitcoin → fiat, money growth 1 → 0). Frozen fiat’s unemployment (dashed orange) tracks bitcoin for the first couple of years, then climbs to roughly 24% around 2034 while bitcoin holds near 15%. The two only converge near the end.
Barry’s insight · the missing experiment
Real politicians rarely hold money growth at zero. The temptation runs the other way, toward printing more, especially in a crisis or before an election. The simulator doesn’t model that temptation yet. A central bank that sometimes gives in would be the natural next experiment, because that is where fiat’s real-world risk lives.
So what
The coin isn’t the issue. The money rule is. Fiat’s advantage here depends entirely on a printing press used with discipline.
05
Finding five · Who pays
Old debts turn gentle deflation into a crisis, and mortgage holders pay
Most of the damage in Barry’s preset traces back to mortgages signed in the old fiat world: 62% of homeowners start the simulation owing fixed sums of money. When prices and wages start falling, those debts get heavier every year. Borrowers default, banks fail, money is destroyed, and the deflation feeds on itself. Run the very same bitcoin economy without the inherited mortgages and the damage roughly halves.
Same bitcoin economy, with vs without old mortgages
71%vs28%
Households renting by 2046
Live app
15%vs10%
Unemployment at the end of the run
Live app
31.5vs16
Bank failures over 20 years. Inflation: −5.5%/yr vs −2.6%/yr
Live app
Live appHow to read: solid = bitcoin with the inherited mortgages (62% of owners), dashed = same economy without them. The blue Renting line climbs to 71% with old debts. Without them it falls to 28%, and two-thirds own outright.Live appJobs: the orange unemployment pair separates after about two years and ends at 15% with old debts (solid) vs 10% without (dashed).Live appBanks: running count of bank failures, which ends at 31.5 with old debts (solid) vs 16 without (dashed). The two run level for about a decade, then the solid line pulls away.
Who pays for a switch?
Barry’s round-5 tests tracked households through a one-step switch from fiat to bitcoin. The bill landed on mortgagors (42% of households), especially the middle-skill ones, whose consumption fell 40% (Q3) and 29% (Q4). Renters, people who owned outright and the most skilled came out ahead.
−37%
Mortgagors’ consumption, plus 15 more foreclosures per 100
Across rounds
+15%
Renters’ consumption
Across rounds
+69%
Outright owners’ net worth
Across rounds
Across roundsHow to read: Barry’s preset, change in real consumption ten years in, switch to bitcoin vs stay on fiat, by household group. Orange is the default switch; the other colours are variants. Debtors, mortgagors and middle-skill (Q3) households sit near −37% to −40%, while renters and low-skill (Q1) households gain.Live appThe switch itself: here solid = fiat for a year, then a one-step switch to bitcoin (grey band); dashed = stay on fiat. Renting (blue) rises to 44% after a switch but falls to 19% if fiat stays, and mortgages (purple) shrink to 31% instead of growing to 54%.
Context: in Barry’s gentler default setup the same switch has no jobs effect, and mortgagors lose about 8% instead of 37%. Inherited debt plus the preset’s harsher deflation is what turns a switch into a crisis.
So what
A switch’s costs aren’t shared evenly. They fall on ordinary borrowers, so any real transition would have to deal with existing debts first.
The big picture
It isn’t about the coin
Five rounds in, the simulator never delivered a simple verdict. Whether bitcoin-like money helps or hurts the average person kept depending on four dials that have nothing to do with the coin itself:
1Wage flexibility
If pay can fall along with prices, deflation costs few jobs. If it can’t, unemployment soars (Finding 3).
2Inherited debt
Debts signed in an inflationary world get heavier as prices fall, and borrowers, banks and jobs pay for it (Finding 5).
3How the government runs its budget
A treasury that hoards taxes drains money from circulation. One that gives back the surplus makes the “spiral” disappear (Finding 2).
4For fiat: political discipline
Fiat with zero money growth acts like bitcoin. Fiat with too much growth is the risk the model doesn’t test yet (Finding 4).
A teaching tool, not a forecast. The simulator is a small, simplified economy that still misses real-world features. In Barry’s preset, bank failures and foreclosures run far above anything in history. Money changes hands too slowly. There’s no innovation, no world with two currencies circulating side by side, and no government deficit path. Treat every number here as evidence about assumptions, never as proof about the real world. That is exactly the point.
Try it yourself
Pin a run as the baseline, change one slider, and watch the dashed line drift away from the solid one. You can check any finding here in a few minutes.
