negative September 28, 2026

From an emerged start, an energy economy does not keep complexity rising

DESIGN §1.3 sweep 11 asked whether an existing replicator keeps getting more complicated once a treatment is switched on: 216 continuations of 18 emerged worlds, 20 000 epochs each, under the parent's own dynamics, two energy economies or host mode. Read as pre-registered, neither economy arm can be read, because each relapses more than the continuation; the rich economy's sign test (12 pairs to 1) rests on two parents. Host mode is refuted, and so is strict persistence: 52 of 54 continuations held their replicators and 2 relapsed. A held-out confirmatory reading on the survivors of 8 unseen parents did not replicate the complexity signal: refuted at economy 2048, not shown at economy 8192 (1 pair to 0). From an emerged start the economy does not keep complexity rising.

Question

Complexity rises at emergence and then stops rising on every substrate the programme has tested from a random start, and the host–parasite sweep found one priced arm that kept rising on a single measured run. A run that starts from random tapes spends most of its budget waiting for replicators, and a treatment can only work on what is left. DESIGN §1.3 sweep 11 starts from replicators instead. Once a treatment is switched on in a world that has already made them, does the existing replicator keep getting more complicated, and does it hold at all?

Method

Each parent is an economy-off control of sweep 9 whose last stored world is at least half replicators by the orientation-aware census. Its terminal world is continued for another 20 000 epochs under four treatments, seeds 1001–1003 each, paired by (parent, seed): the continuation under the parent's own dynamics, an economy with influx 2048, a rich economy with influx 8192 (both with steal 1024, a stock cap of 32 768 and loss 0.5), and host mode.

The reading is pre-registered (docs/design_record.md, 2026-09-25, "Runs that start from an emerged world", and its six clarifications). Each child held its replicators where the last-decile median replicator_share stayed high and the share never sat at the floor, and its complexity rises, plateaus or is mixed on dominant_instruction_count over the samples whose dominant tape self-replicates. Each treated arm is a one-sided sign test against the continuation over paired children, printed but not read where the arm relapses more often than the continuation; H-persistence says no continuation child relapses. A second, held-out reading (2026-09-25, "The from-emerged economy arms, confirmed on held-out parents") was written after the first ten parents had been seen, and reads the same complexity rule on the survivors of the parents that qualified afterwards: H4-survivors.

Result

From an emerged start, neither energy economy can be read as keeping complexity rising, and on the held-out survivors the complexity signal does not replicate.

• H-economy, economy 2048: 7 pairs favour the treatment and 3 the continuation, p = 0.172, not read: relapses more than the continuation
• H-economy, economy 8192: 12 pairs favour the treatment and 1 the continuation, p = 0.00171, not read: relapses more than the continuation
• H-host, host mode: 1 pairs favour the treatment and 2 the continuation, p = 0.875, refuted
• H-persistence: refuted H-persistence refuted: 2 of 54 continuation children relapsed.
• H4-survivors, held out, economy 2048: 0 pairs favour the treatment and 1 the continuation, 20 tie, p = 1, refuted
• H4-survivors, held out, economy 8192: 1 pairs favour the treatment and 0 the continuation, 18 tie, p = 0.5, not shown

The economy arms relapse more often than the continuation, so the pre-registered rule reads neither of them; the parents a held sign test rests on are listed below. The held-out reading was written to ask the question again on colonies that survived the treatment, on parents nobody had seen: there almost every surviving pair ties, and the seen rich-economy signal is not there. So the economy does not keep complexity rising from an emerged start, and the question of whether complexity keeps rising stands where the host–parasite finding left it. What the economy does do, on the same held-out children, is make the dominant replicator copy itself faster.

The colonies themselves are robust: 52 of 54 continuations kept their replicators over 20 000 more epochs. Only the strict rule that none relapses is refuted.

Every count on this page is read from the sweep's stored samples at render time, through the same readings the sweep page draws.

The pre-registered reading final

Every child of every qualifying parent has finished. Each child is read over its own samples. It held its replicators where the median replicator_share of its last decile is at least 50% and the share never sat below 10% for 3 samples running. It rises where the last-decile median dominant_instruction_count of its self-replicating samples is at least 1.2 × the first decile's, and plateaus within ±10%. It is unmeasured with fewer than 10 such samples in either decile.

Treatment	Children	Finished	Held	Relapsed	Rises	Plateau	Mixed	Unmeasured	steal_rate	distinct_tapes
continuation	54	54	52	2	3	20	31	0	—	5980
economy 2048	54	54	40	14	7	11	36	0	0.00345	8720
economy 8192	54	54	46	8	13	11	29	1	0.019	8940
host mode	54	54	54	0	2	17	35	0	—	4690

Hypotheses

• H-economy, economy 2048 against the continuation: not read: relapses more than the continuation

  54 pairs measured on both sides: 7 favour the treatment, 3 the continuation, 44 tie; one-sided sign test p = 0.172. Its children relapse more often than the continuation's, so the test (not shown) is printed but not read as H-economy.

  Per-parent agreement

  Parent	Favour the treatment	Favour the continuation	Tie	Not both measured
  run 944	0	1	2	0
  run 950	0	1	2	0
  run 967	0	0	3	0
  run 991	0	0	3	0
  run 1007	0	0	3	0
  run 1029	2	0	1	0
  run 1087	0	0	3	0
  run 1089	3	0	0	0
  run 1103	2	0	1	0
  run 2568	0	0	3	0
  run 2577	0	0	3	0
  run 2590	0	0	3	0
  run 2654	0	0	3	0
  run 2676	0	0	3	0
  run 2700	0	0	3	0
  run 2771	0	0	3	0
  run 2802	0	1	2	0
  run 2862	0	0	3	0

• H-economy, economy 8192 against the continuation: not read: relapses more than the continuation

  53 pairs measured on both sides: 12 favour the treatment, 1 the continuation, 40 tie; one-sided sign test p = 0.00171. Its children relapse more often than the continuation's, so the test (held) is printed but not read as H-economy. Carried by parents run 1029 and run 1103: without their pairs the test falls to p ≥ 0.05.

  Per-parent agreement

  Parent	Favour the treatment	Favour the continuation	Tie	Not both measured
  run 944	1	0	1	1
  run 950	2	1	0	0
  run 967	1	0	2	0
  run 991	0	0	3	0
  run 1007	0	0	3	0
  run 1029	3	0	0	0
  run 1087	0	0	3	0
  run 1089	0	0	3	0
  run 1103	3	0	0	0
  run 2568	0	0	3	0
  run 2577	0	0	3	0
  run 2590	0	0	3	0
  run 2654	0	0	3	0
  run 2676	0	0	3	0
  run 2700	0	0	3	0
  run 2771	0	0	3	0
  run 2802	2	0	1	0
  run 2862	0	0	3	0

• H-host, host mode against the continuation: refuted

  54 pairs measured on both sides: 1 favour the treatment, 2 the continuation, 51 tie; one-sided sign test p = 0.875.

  Per-parent agreement

  Parent	Favour the treatment	Favour the continuation	Tie	Not both measured
  run 944	0	1	2	0
  run 950	0	1	2	0
  run 967	0	0	3	0
  run 991	0	0	3	0
  run 1007	0	0	3	0
  run 1029	0	0	3	0
  run 1087	0	0	3	0
  run 1089	0	0	3	0
  run 1103	0	0	3	0
  run 2568	0	0	3	0
  run 2577	0	0	3	0
  run 2590	0	0	3	0
  run 2654	0	0	3	0
  run 2676	0	0	3	0
  run 2700	0	0	3	0
  run 2771	0	0	3	0
  run 2802	1	0	2	0
  run 2862	0	0	3	0

• H-persistence, the continuation children hold: refuted

  H-persistence refuted: 2 of 54 continuation children relapsed. They were watched over colony ages 600–38,320.

  • run 3171 (parent run 1087, seed 1003) relapsed at epoch 21,520, colony age 13,350
  • run 3601 (parent run 2654, seed 1001) relapsed on its last decile alone

• continuation
• economy 2048
• economy 8192
• host mode

A held-out confirmatory reading, pre-registered 2026-09-25 final

Its rules were written after the reading above and an exploratory look at the same children, so they are read only on the children of parents from sweep 9's extension (seeds above 90) none of whose children had been seen. Past a settling window of 1,000 epochs, a child relapsed where its share sat below 10% for 3 samples running; it is extinct where its last-decile median share is below 10%. H3-latency pairs the economy child against the continuation child on the ratio of their last-decile to first-decile median copy_latency, a lower ratio favouring its side; H4-survivors reads the original complexity rule on pairs whose two children both survived. Host mode is not re-tested.

Treatment	Children	Finished	Settled relapses	Extinct	Latency measured	Survivors
continuation	24	24	0	0	24	24
economy 2048	24	24	3	0	24	21
economy 8192	24	24	5	0	24	19
host mode	24	24	0	0	24	24

Hypotheses

• H3-latency, economy 2048 against the continuation: shown

  24 pairs measured on both sides: 20 favour the treatment, 4 the continuation, 0 tie; one-sided sign test p = 0.000772.

• H4-survivors, economy 2048 against the continuation: refuted

  21 pairs measured on both sides: 0 favour the treatment, 1 the continuation, 20 tie; one-sided sign test p = 1.

• H3-latency, economy 8192 against the continuation: shown

  24 pairs measured on both sides: 21 favour the treatment, 3 the continuation, 0 tie; one-sided sign test p = 0.000139.

• H4-survivors, economy 8192 against the continuation: not shown

  19 pairs measured on both sides: 1 favour the treatment, 0 the continuation, 18 tie; one-sided sign test p = 0.5.

What is not claimed

• Not that an economy cannot raise complexity. It did not here, on this substrate, over 20 000 epochs past an emerged world. A longer budget, a different economy or a different world is untested.
• Complexity is the instruction count. The rule reads dominant_instruction_count over the samples whose dominant tape self-replicates; a replicator that grows more intricate without executing more of its bytes reads as flat.
• Children are correlated. Three children share each parent's world, so the pairs are not independent colonies, which is why every held test lists the parents it rests on.

Evidence

Transition, flagged, census and arm are used here in the senses set out in the glossary.

Sweep: From an emerged world