In this commit, we swap the live-run panel's lineage data from the
exp-018 gepa arm to code_deg1, the most recently completed run, so
the telemetry matches the runs the page now discusses.
In this commit, we add findings sections 15 and 16: the ceiling arm
(meta_harness at ten times the budget converges to a lower shelf,
closing both halves of the exp-018 question) and the lying-channel
breed (the first evolved attribution-confidence machinery, the
flattest degradation profile measured, and the attempt-cap subsidy
that reframes the verdict). Timeline gains exp-022, the exp-023
spec, and exp-024; the challenger count moves to eight everywhere it
appears; the live-run panel now states plainly that no optimizer is
live and that the interval-router branch is unbenchmarked in this
simulator.
In this commit, we add §14 with both halves of the distillation
result — the soft_unknown recovery tables and Fig. 6, the four-way
descent trace that is the null's whole proof — plus the timeline
entries, and we resolve every place the site promised the patch as
future work with a pointer to the measured outcome. The writeup's
mainnet recovery cell gains the footnote the site pass flagged: 17%
is of the success loss, since lnd's objective rose under degradation
on that tier.
In this commit, we add §13 — three engines, one starting line: the
engines table, why the claude arms produced nothing, the adjudication
statement with its limits, omni1 as challenger failure number six,
and the two idea-ledger mechanisms — plus the timeline entry, the
ceiling sections' resolution notes, and the freshened counts (twenty
experiments, six failed challengers). The stale live-run panel now
points at the adjudication's gepa arm, and its run telemetry ships
with this version.
In this commit, we fold the shift-isolated mainnet arm into the
exp-019 record. Shift does not help lnd on mainnet: both CIs straddle
zero, the sign flips between levels, and the decomposition shows the
small positive at 0.1 is the give-up spiral's attempt term with
success falling on eight of eight files. The anomaly is hard-tier
only, and the route-geometry story fails by its own premise — it
predicted the vanishing because mainnet routes would be longer, and
they are three times shorter (hub source, 1.9 mean first-attempt hops
against 5.4). An outcome can match a prediction while refuting its
premise; this one did, and the check ran before anything was
published rather than after. The anomaly now ships with two facts and
no mechanism, and the queue drops the isolation arm as done.
In this commit, we add §12 — the degraded-attribution ladder, its two
tables, the five findings, and the anomaly labelled as an anomaly —
plus the two exp-019 timeline entries, and we sweep the 8.6x claim
across the whole site: every place that quoted the attempt ratio as a
live headline now carries the retirement (the ratio was a
perfect-channel figure; under realistic degradation the edge converts
to success), while historical narration keeps its numbers and gains a
pointer instead of a rewrite. The masthead moves to nineteen
experiments, title-defended champions, and attribution-tested status.
In this commit, we close the loop the v58 publish opened: every
"title in adjudication" caption across the findings page, drift page,
index and chart series notes now carries the exp-020 resolution, the
§11 note box gains a same-day postscript with the split-test result,
and the timeline gains the exp-020 verdict entry including the
sealed-corpus provenance findings.
In this commit, we sync the run lineage data the v58 bundle shipped
with (code_hybrid1, 28 candidates), so the committed site matches
what Litbucket serves.
In this commit, we add §11 to the findings page — the thirteen-world
sweep, the full scores table, the seed-as-control argument, atomic1's
regime specialization, and the eroding generalist title — plus two
timeline entries, and we bring the stale framing current everywhere
it appears: masthead counts and date, the §00 circularity note now
pointing at its half-closure, and the mx_c3 "generalist / best
overall" captions on index, drift, and the chart series notes now
carrying the in-adjudication qualifier. Design untouched; everything
additive or a one-line text correction.
The published explanation of why served weights hurt lnd was wrong, and
it reached the dashboard. It said lnd files a failure as a pair penalty
carrying no amount, suppressing a corridor for every payment size.
probability_apriori.go:363 returns the unpenalized prior whenever
amt < FailAmt, so the estimator gates on amount correctly. A Fable
advisor found it by reading the estimator rather than the summary.
Its replacement hypothesis was wrong too. Node-level contagion is real
-- getNodeProbability folds pair results into a node prior used for all
that node's untried channels, which the "761 edges, 761 pairs" check
never ruled out -- but disabling it with apriori.weight=1.0 leaves the
loss at -0.038 against -0.046.
The third guess, staleness, looked decisive and was a confound.
Failures from a one-payment server cost lnd +0.000 on 0 of 10 files,
but that set holds 232 observations against the stale set's 2,808. A
size-matched random subsample of the STALE set gives -0.003. At equal
volume stale and fresh are indistinguishable.
What survives is volume. Each imported failure blocks one directed edge
at or above its amount, and server and consumer draw amounts from the
same distribution, so the bounds land where the consumer is about to
send. At 232 observations nothing happens; at 2,808 over a 761-edge
graph lnd finds its amount blocked almost everywhere and can only route
around, onto longer and worse paths.
The upstream thesis survives in a sharper form. lnd's estimator does not
ignore amounts. Nothing DOWNSTREAM of it can act on an amount bound,
because findPath takes the amount as a fixed argument -- so knowing that
at least X fails on an edge can only subtract routes, never resize the
payment. That is exp-002b's finding from the opposite direction, and the
two now converge on one patch instead of two observations.
Two further claims are labelled rather than deleted. exp-011's paradigm
ceiling is confounded with the optimizer, since every run in this
program used engine="gepa" and three lineages converging on one band is
evidence about that engine's attractor as much as about the problem; the
adjudicating run is specified in the writeup. And the proposer law from
exp-010b was fitted to two opposite-sign runs, so it is downgraded to a
hypothesis.
Also stages params_lnd_no_contagion.json, the weight=1.0 configuration
that isolates node-level aggregation, so the contagion control is
reproducible.
Adds a section on served weights, the result that answers what a
weight-serving API should carry: the same observations help the
champions and hurt lnd, and splitting the stream shows the whole of
lnd's loss is the failure evidence, because a bound carries an amount
and a penalty does not.
Also fills four holes in the timeline. exp-013 through exp-016 had no
entries at all, so the site still described exp-013 as live and the
champions as leading four tiers when they lead six. The 8.6x headline
now carries its own caveat rather than hiding it in a writeup: it is an
upper bound until degraded attribution runs.
In this commit, we write up exp-012 and exp-002b on the findings page, and
put a corrections callout above everything else on it.
exp-012 gets §08. Every number we've published is a cold-start number, so
the section walks the four arms that finally tested the other regime: the
warmup curves (lnd's disadvantage grows 4.7x to 11.9x across a mainnet
batch, the champions are already cheap on payment one, atomic1 is the only
router that measurably learns), the depletion mistake that the snapshot and
restore control fixed, the three failure modes under a maximally stale
cache (lnd thrashes, the champions abandon, atomic1 shrugs at a 0.012
probability floor), the staleness null that indicts our own churn engine,
and the vantage transfer result. atomic1's +0.428 at p=.002 is the first
significant win over a champion here, so it gets a stat tile, and the note
says plainly that the negative is about probe-warming rather than
weight-serving, since no arm can buy knowledge without spending liquidity.
exp-002b gets §09. Seven bimodal scales including the environment-matched
one, and none of them beats lnd's own apriori default. How it fails is the
part worth reading: a better prior raises success while more than doubling
attempts, because findPath takes the amount as a fixed argument and the
estimator can't change what lnd retries.
The corrections callout is §00, unnumbered as a section but linked first in
the TOC, and it retracts three claims we made about ourselves: the bimodal
prior was in the harness prompt all along, the evolved prior constants fit
sim_liquidity.go's generator (so the mainnet tier is real topology and
policies with our liquidity), and lnd's decay never fires on the static
tiers. The one flatly contradicted sentence in §04 now carries a retraction
marker pointing at it. A research site that corrects itself where readers
can see it is more credible, not less.
Downstream sections renumber to §10 and §11, the timeline picks up two
entries, and the masthead and index status lines move to exp-012 and
exp-002b closed with exp-013 live.
In this commit, we write up exp-010b on the findings page as a new
section 07, "The honest arena: atomic commitment, and the fifth
challenge", with the process and timeline sections renumbered to 08 and
09 behind it.
The section tells the arena story first: shards that hold liquidity
until the whole payment settles, siblings contending for what is held,
and background traffic drifting on every attempt boundary, so a long
reactive ladder finally pays for the churn it sits through. Flag-off
byte-identity means none of the earlier results on the page moved.
The headline is the baseline, which reordered the field before evolution
ran at all: lnd falls from second place to last, spending 105 attempts
per payment where it spent 23 with instant settlement, and exp-010's
persistent-plan router pulls statistically even with mx_c3 on both
atomic tiers without ever having seen an atomic shard. Two tables carry
the numbers, the seven-router baseline and the six-tier paired sweep
with sign-test deltas against the champion.
Then the verdicts. mx_c3 survives its fifth direct challenge on an arena
built expressly against its evidence ladder, but the shape of the
frontier changed: the codex arm's hybrid of cross-payment memory and
reservation-ledger planning is the first challenger in the program with
no collapse tier, and it routes mainnet payments in 1.6 attempts, the
lowest figure we have measured. The Opus arm lost outright, its
drift-bred bound relaxation burning 57 attempts per payment, which flips
the exp-010 proposer A/B and adds the clause that proposer strength
interacts with environment variance.
We also close the forward pointers. The exp-010 sidenote now says how
its designed follow-up turned out, the drift page's "one drift
intensity" caveat gets the attempt-boundary answer, and the index byline
and live-run panel move to exp-010b closed with exp-012 next.
In this commit, we write up the close of exp-010 on the dashboard, as a
new section 06 in findings.html. The corridors corpus asked whether joint
route-set planning shows up once unequal splitting is mandatory, and all
three proposer lineages (codex/gpt-5.6-sol, Opus 5 at default effort,
Opus 5 at medium) answered yes, at increasing depth: one-step lookahead
with reservation, up-front corridor-sized shard sets, then persistent
parallel flow plans with concurrency-first dispatch at 1,931 lines.
The section carries the full five-tier sweep with paired deltas against
mx_c3 and sign-test p values, so the two headlines can be read off the
table directly. The Opus-default arm posted the program's first
statistical tie with a champion on any tier (+0.005 on split validation,
success 0.958 vs 0.917) and then collapsed off-corpus at 0.303 on the
hard test, which is what a corridor-tuned fail budget buys you. The
medium arm matched codex throughput, took the family's best val score
and its worst held-out, a val overfit the sealed sweep caught. Champions
of record stay hb1 + mx_c3, and the pre-registered resolution caveat
plus the exp-010b successor are logged in a sidenote so nobody reads the
nulls as evidence that joint planning can't win.
The process and timeline sections shift to 07 and 08, the timeline's
exp-010 entry moves from "next" to the verdict, and the stale forward
pointers on the overview and drift pages now point at the result. No
design changes: same sections, tables, notes and sidenotes the rest of
the site already uses.
In this commit, we fold the exp-008 outcome into the site. The drift
page's outcome ledger, written before the run, is resolved honestly:
the realized result is the first row's mechanism in the second row's
shape with the third row's verdict, and each row now carries its
resolution inline. A new verdict section presents the evolved clock --
the 35 minute confidence half-life, the 20 minute bound expiry, and
the interpolation back to the prior -- alongside two new figures: a
decay curve contrasting drift1's confidence fade with lnd's one hour
penalty weight, and the six-router drift-test comparison on the same
axis scale as the baseline figure so the two read together. The
four-tier results table and the drift1-versus-gen2 cut carry the
argument, and the closing section separates what the experiment
settles from what it cannot.
The index and findings pages get the matching surgical flips: status
lines move from in-flight to verdict-in, the evidence claim grows to
four of four tiers, and the hand-off now points at exp-010 and the
code_split1 run. Published as Litbucket version 42.
In this commit, we pick up the dashboard refresh for the code_drift1
run: the exported lineage data now reflects 167/400 evals, best score
0.9209, with the budget still open.
In this commit, we fold the two newest findings into the site without
touching the approved design. The settled result, exp-011's paradigm
ceiling, becomes a new findings section: three independent lineages
converge within 0.014 objective of one another on the same
interval-belief design, shown as a dot plot against the lnd and seed
baselines, with the two unrewarded inventions and the honest limits
(three samples, shared reflection-model culture) called out alongside.
The in-flight experiment gets its own page instead: drift.html, a
third nav tab covering the virtual clock and background traffic
mechanics, the pre-evolution baseline in which the champions' hard
bounds survive drift while lnd's decay still trails, and a ledger of
what each possible outcome of the live code_drift1 run would mean.
Keeping a verdictless experiment out of the findings page preserves
that page's register, and the new tab gives future in-flight chapters
a home.
The masthead, timeline, and live-run telemetry now reflect eleven
experiments with code_drift1 running, and refresh_dashboard.sh
bundles the new page so scheduled publishes carry it. Published as
Litbucket version 36.
In this commit, we write up exp-011: the code_gen2 run seeded from the
small original router with the champions' discovered insights supplied
only as prose in the reflection prompt. The run completed its full 400
eval budget cleanly and accepted 31 candidates, roughly four times
code_mix1's acceptance rate, confirming that small seeds iterate much
faster than champion-sized ones.
The headline result: insight transfer works, but hits the same
ceiling. The best gen2 candidate reaches champion-class performance on
all three held-out tiers (combined 0.638 vs mx_c3's 0.652 and hb1's
0.640, with identical 2.3 attempts/payment on mainnet) yet passes
neither champion. Three independent lineages now converge on the same
interval-belief paradigm and the same performance band, which reads as
a local optimum for the current environments. gen2 also evolved two
novel mechanisms the sim never rewards -- in-flight local liquidity
reservation and weakest-edge failure attribution -- suggesting the
next lever is changing the environment (exp-008 background traffic,
exp-010 splitting pressure) rather than spending more eval budget.
We archive the best candidate source next to the writeup, log the
paradigm-ceiling learning in IDEAS.md, update the notebook and
CLAUDE.md open-work section, and check in the final code_gen2 lineage
data for the command center.
In this commit, we replace the dashboard's console theme with an
editorial report design: a single light paper theme, serif headings
and body with mono reserved for quantitative type, and one oxide
accent used only where it carries meaning. The palette was validated
for CVD contrast against the actual background rather than eyeballed.
We also add findings.html, a standalone write-up of the research so
far: the mainnet validation headlined by the 8.6x attempt reduction,
the paradigm-over-parameters negative result, the three tiers of
held-out evidence, and an anatomy ledger of what the evolved routers
dropped, rediscovered, and invented relative to lnd's stack --
including the honest caveat that the champions' rejection of time
based logic is partly a simulator artifact.
Four hand-drawn SVG diagrams explain the system: the GEPA loop, a
schematic contrasting clock-decayed penalties with evidence-bounded
liquidity intervals, the rediscovered bimodal prior plotted from the
champions' own constants, and the champions comparison chart.
The live-run wiring is hardened for code-mode runs: the candidate
view now detects Go-source candidates, shows line counts and source
diffs, and degrades gracefully on missing exporter fields. Finally,
refresh_dashboard.sh now bundles findings.html, so scheduled
refreshes no longer drop the page from the published site.
In this commit, we pick up the dashboard refresh for the new code_gen2
run: the exported lineage data and a live-run note that the small-seed,
insight-enriched follow-up is in flight.
In this commit, we add the static command-center site: an overview phase
tracker, live run charts with candidate lineage, a corpus explorer, and
an explainer covering the simulator, mission control's probability
models, and the evolution results. The site is self-contained (no build
step), serves via any static file server, and publishes to Litbucket
through the refresh_dashboard.sh script, which exports a GEPA run
directory into the dashboard's data schema via export_run.py.
