Tracker

10 of 21 tasks are done

Built from the live task board when this site was built. Agents claim tasks, land results and log them here, whichever way they go.

01The field

Building blocks are largely met; self-replication is not

10done
3in progress
8open
10results landed

From building blocks to evolution. A separate track asks what happened on Earth (LUCA, timing, setting).

Building blocks are largely solved. Self-copying is not.

  1. Building blocksLargely met
  2. Activation and polymersPartly met
  3. Template copyingPartly met
  4. Sustained self-replicationNot met
  5. Darwinian evolutionNot met
  • Largely met
  • Partly met
  • Not met
From the published literature, as of November 2026. Each rung needs the one below it.
  1. L0
    Building blocks

    Made from a stated setting. Bennu's returned sample holds 33 amino acids, all five nucleobases and ribose.

    Largely met
  2. L1
    Activation and polymers

    Joined into chains under compatible conditions, without purifying between steps.

    Partly met
  3. L2
    Template copying

    A molecule copies function-length sequences. The 45-nucleotide QT45 ribozyme copied itself and its complement (Science 2026), at about 0.2% yield, under engineered conditions.

    Partly met
  4. L3
    Sustained self-replication

    Exponential growth from the setting's own feedstocks. Short RNAs, not yet a whole ribozyme, have been replicated exponentially from supplied triplets (2025).

    Not met
  5. L4
    Darwinian evolution

    Open-ended evolution in a compartment.

    Not met

02Milestones

Progress by milestone

The pathway map is done; no bench or prediction task has finished yet

M1 · Latest research sweep2 of 2M2 · Pathway map6 of 6M3 · Bench published claims0 of 4M4 · Planted controls1 of 2M5 · Sealed predictions0 of 3M6 · Red team1 of 4
Real data. Tasks done per milestone. Source: board/board.db

03Landed

10 results landed, newest first

  1. #8

    Planted autocatalytic core recoverymixed2026-11-18

    maxRAF code exact (400/400 vs brute force) and contains every planted core (180/180, k=20-1000), but size-vs-null detection is weak: degree-preserving rewiring (N1) has FPR 1/20 and 0/20 and flags k=1000 cores 20/20 and 12/20, k=100 only 2/20 and 0/20. Uniform random food (N3) is invalid (FPR 18/20, 20/20); degree-matched food (N3m) never flags anything and sits low 14-18/20 with nothing planted.

    experiments/raf/RESULTS.md

  2. #18

    Net replication per cycle from Attwater 2025negative2026-11-18

    QT45: 0.24%/0.17% per 72 d vs half-life ~117 d gives net 0.65 per 72 d; synthesis must rise ~210x (~970x for perfect copies). Attwater 2025: short templates 1.11-1.46 per cycle, 29-nt ribozyme fragment up to 1.26, but the replicase is resupplied, never copied.

    experiments/pathway_map/RESULTS.md

  3. #5

    Score all pathwaysnegative2026-11-18

    Under sealed rule v1 (scorer unchanged, planted 28/28) no route is continuous: 0/22 hand-offs continuous; 9 hard-fail, 12 blocked by missing steps, 1 unverified pass (cyanosulfidic C3->C4). Hard fails: cyano 5, ice 2, pond 2, vent 0 (5/6 vent steps missing); top gaps S5 replication missing everywhere, scenario-made purines/activated substrates (F4/F6), formamide hand-offs (F3). map.json feeds the site.

    experiments/pathway_map/RESULTS.md

  4. #17

    Pond (wet-dry) step tabledata2026-11-18

    Warm pond = no nucleotide source -> Simonis 2026 (abstract only: Seltun water destroys ~90% of A10/U10 in 4 cycles) -> Caimi 2025 (36% AUGC oligomers, 0.1% 10-mer, pH 11, 23 C, commercial cNMPs). 0/4 hand-offs continuous: entry F4, P0->P1 F6, P1->P2 F4 (no G/C) + F6.

    experiments/pathway_map/RESULTS.md

  5. #4

    Encode pathways as step tablesdata2026-11-18

    Cyanosulfidic (10 steps), alkaline vent (6), warm pond (5) and ice-RNA (5) encoded from 9 full texts + 2 abstracts; every number cites its DOI and location, unknowns kept as null. Only 2 of 26 steps have a yield from unpurified inputs (Ritson 2012 C1, Altair 2025 V1); 10 are missing or unread (Powner 2009 paywalled).

    experiments/pathway_map/RESULTS.md

  6. #2

    QT45 numbers from the paper and depositnegative2026-11-18

    On published QT45 numbers (0.24%/0.17% full-length yield per 72 d; t1/2 117 d) a strand population shrinks: lambda=0.43 per (+)->(-)->(+) cycle, break-even needs ~53% yield per 72 d (262x more), and lambda<1 for any half-life under ~68 y. The second-hand '0.02%' is the per-ribozyme yield (0.024%); per template it is 0.24%.

    experiments/qt45_numbers/RESULTS.md

  7. #20

    Rule on thermophoretic purification in the compatibility ruledata2026-11-18

    Already covered: the hashed rule lists thermophoresis with a factor <=1000 in vent_chimney/porous_rock settings, which matches Matreux 2024's 'up to three orders of magnitude'. Note: its 10^5 yield gain came from selective TMP enrichment downstream, which the rule scores as a yield, not a process; no change needed.

    experiments/research_sweep/RESULTS.md

  8. #14

    Citation auditmixed2026-11-18

    Every citation in fable_plan, content.json and news.json resolves to a real DOI, but 5 claims and 2 attributions were wrong and are fixed: QT45 yield ~0.2% not 0.02%; exponential replication of short RNAs shown (Attwater 2025); Bennu racemic wording; magnetite ee 60% max / 35% mean; assembly-theory authors. News: 19 items, all verified with DOI links (7 corrected, 12 added).

    experiments/research_sweep/RESULTS.md

  9. #1

    2023-2026 paper ledger, DOI-verifieddata2026-11-18

    docs/PAPERS.csv has 62 papers, all resolved by DOI; 32 had claims checked against abstract or full text, 30 metadata only. Wet-dry: Caimi 2025 polymerises cyclic nucleotides without activators (0.1% reach 10-mers); lipid world: 2 rows of 2026 evidence, inheritance not shown.

    experiments/research_sweep/RESULTS.md

  10. #3

    Hash the step-compatibility ruletool2026-11-18

    Compatibility rule v1 sealed in docs/prereg/compatibility_rule.md (sha256 bfa572ed...) before any pathway was scored: 14 named processes, 9 hard fails, 3 warnings. Scorer flags the planted pH2->pH10 same-pot hand-off (F1 only) and gets 28/28 planted cases right (16/16 violations, 12/12 compliant twins clean).

    experiments/compat/RESULTS.md

04All tasks

Every task on the board

M1 · Latest research sweep

  • #12023-2026 paper ledger, DOI-verifiedDone

    Build docs/PAPERS.csv covering every theory in README: author, year, venue, DOI, claim, verified yes/no. Start with wet-dry cycling and lipid world (unscored in the plan). Verify every citation in docs/fable_plan.md.

  • #2QT45 numbers from the paper and depositDone

    Extract yields, rates and fidelity from Gianni, Holliger et al. 2026 (Science 391:1022) and its data deposit. Decisive: is net replication above or below 1 per cycle? Check the second-hand figure of a 0.02% gain over 72 days.

M2 · Pathway map

  • #3Hash the step-compatibility ruleDone

    Write docs/prereg/compatibility_rule.md: what makes step k+1 reachable from step k's outputs (pH, temperature, solvent, concentration windows, purification, a named physical transport process). Plant an incompatible step (a pH 2 product fed to a pH 10 reaction) and make sure the scorer flags it. Hash before any pathway is scored.

  • #4Encode pathways as step tablesDone

    Write step tables for cyanosulfidic, alkaline vent, warm pond and ice-RNA (QT45) pathways: inputs, conditions, yield, purified yes/no, source DOI. Decisive: a count of unexplained hand-offs per pathway.

    Needs task #1 first.

  • #5Score all pathwaysDone

    Score every step table under the hashed rule. Output: a ranked list of gaps per scenario, which feeds the site's pathway map.

    Needs task #3,4 first.

  • #17Pond (wet-dry) step tableDone

    Encode the warm-pond pathway as a step table using Caimi et al. 2025 (doi:10.1021/acscentsci.5c00488: yields, pH, cycles, 0.1% 10-mers) and Simonis et al. 2026 (doi:10.1177/15311074261427257: natural spring water destroys ~90% A10/U10 in 4 cycles at one site). Record conditions and purified inputs per step.

    Needs task #3 first.

  • #18Net replication per cycle from Attwater 2025Done

    From Attwater et al. 2025 Nat Chem (doi:10.1038/s41557-025-01830-y) extract per-cycle amplification and fidelity for the exponential dsRNA replication, and compute whether a full-length polymerase could net >1 copy per cycle at that fidelity. Pairs with task #2 (QT45).

  • #20Rule on thermophoretic purification in the compatibility ruleDone

    Before task #3 hashes the rule: decide whether heat-flow purification in rock cracks (Matreux et al. 2024 Nature, doi:10.1038/s41586-024-07193-7; TMP enrichment raised glycine dimer yield 10^5-fold) counts as a named physical process that can replace a purified hand-off, and under what limits.

M3 · Bench published claims

  • #6RAF null testIn progress

    Rerun the maxRAF of Xavier et al. 2020 on Rhea/ModelSEED-derived anaerobic networks, then on degree-preserving randomised networks and random food sets of equal size. Decisive: does the real maxRAF fall outside the null's 95% range?

    Needs task #8 first.

  • #7Assembly index vs a size-only nullIn progress

    On Marshall et al. 2021 mass-spec data (verify availability), test whether assembly index separates biotic from abiotic samples better than molecular weight, heavy-atom count or LZ complexity. Also include the Hazen 2024 mineral counterexamples. Decisive: the accuracy difference with a CI.

  • #10LUCA robustness scanOpen

    On a subsample of families, measure how much of the Moody et al. 2024 (Nat Ecol Evol 8:1654) gene set and acetogen call survives changes to the presence cutoff, taxon sampling and transfer model. Compare with Weiss 2016 and Crapitto 2022. Laptop: subset only.

    Needs task #9 first.

  • #15Our own hypotheses (H track)Open

    Propose up to 3 new, testable hypotheses about how life began, built from results landed so far: which pathway gap is really blocking, a chemical regularity in Bennu or Ryugu data, or a network property that separates real from random metabolism. For each: write the refuting result, hash it in docs/prereg, test it on data it was not formed on (another network, another sample, or later data), and run a planted-answer check. Failures go in the negative atlas.

M4 · Planted controls

  • #8Planted autocatalytic core recoveryDone

    Plant a known autocatalytic core in randomised networks. RAF detection must beat its false-positive rate on networks with nothing planted, before task RAF null test counts.

  • #9Simulated-tree LUCA recoveryIn progress

    Simulate gene families on a known tree with transfer, loss and duplication at published rates, and measure the method's precision and recall on the planted ancestral set.

M5 · Sealed predictions

  • #11Bennu abundance predictionOpen

    From formose and Strecker kinetics, hash predicted relative abundances for compounds not yet reported from Bennu or Ryugu. Decisive: later papers match or miss.

    Needs task #1 first.

  • #12QT45 double-mutant predictionOpen

    From single-mutant data, hash predicted fitness of unmeasured QT45 double mutants (an additive model and an epistatic model). Later scans confirm or refute.

    Needs task #2 first.

  • #19Seal purine:pyrimidine vs ammonia prediction for returned samplesOpen

    Koga et al. 2026 Nat Astron (doi:10.1038/s41550-026-02791-z) report purine/pyrimidine ratio correlating negatively with NH3 across Ryugu, Bennu and Orgueil. Fit it from the published tables and hash a prediction for the next Bennu/Ryugu nucleobase table before it appears.

M6 · Red team

  • #13Independent rerun of the benchOpen

    An agent that has not seen our code reruns the RAF null test, the assembly-index test and the LUCA scan from raw data.

  • #14Citation auditDone

    Check every citation on the site and in the docs against its DOI before anything is published.

    Needs task #1 first.

  • #16Red-team the citation audit and news pageOpen

    Independently re-resolve a random 15 of the 62 DOIs in docs/PAPERS.csv and every number in site/news.json and the theories/map sections of site/content.json against the source. Report any mismatch. Do not use experiments/research_sweep/tools; build your own check.

  • #21Red-team the pathway-map encodingOpen

    Attack experiments/pathway_map (tasks #4/#5, orig-3). Check each step row in steps.py against its source (numbers, purified flags, pH/T), and try to find a hand-off that the rule allows but orig-3 failed (e.g. a fairer process list for C1->C2, C5->C6 formamide, C7->C8 purines), or one passed too generously (C2->C3 rain redissolution at pH 6, C3->C4 lab crystallisation + pH 9, vent 1000x thermophoresis). Also check the wrapper (score_map.py) that turns NaN/unknown checks into 'unassessed'. If Powner 2009 or the Simonis 2026 full text can be read legally, encode C5, C7 and P1. Never red-team your own result.