Litecoin.watch — one week at one node, release confirmation-week-20261014-v1
Publication release: 14 October 2026. Evidence enrollment: 6–12 October 2026 UTC.
As-of export: 2026-10-13T00:03:33Z. Frozen source index retrieved 2026-10-13T23:59:03.276769Z.

Use Python 3.10+ (standard library only):
  python verify.py.txt
  python analyze.py.txt
Run with the extracted package as current directory or pass its directory as the first argument.
Scripts perform reads and print results; no network access, package installation or data mutation.

cohort.csv is filtered ONCE from the final day 12 cumulative snapshot:
  1791244800 <= first_seen < 1791849600
Daily cumulative cohorts must never be concatenated. Polls are concatenated from each of the seven separate UTC daily archives.
The final source cohort had 118,195 retained rows; 68,640 fall in this enrollment week.
Outcome states are from actual export time, not a reconstruction of midnight or a fresh canonical audit.

Deadline comparison: require first_seen <= 1791849813 - 600 for ALL thresholds 60, 300, 600.
35 younger records (28 confirmed and 7 pending) are excluded uniformly; denominator 68,605.
Qualified confirmation requires state=confirmed, valid selected lowerhex txid, positive blockheight,
lowercase hexadecimal 64-character blockhash, first_seen<=confirmed_at<=snapshot and closed_at=confirmed_at.
Matched-by includes equality; later requires a qualified observed inclusion beyond the threshold.
Every other sufficiently aged enrolled record is unresolved, never inferred late or failed.
Full cohort counts and matched-only quantiles use all 68,640 enrolled rows, not the aged deadline subset.
Quantiles: linear interpolation at (n-1)*q; sorted integer observed waits; no miner timestamp arithmetic.

Coverage counts valid consecutive polls <= 30 seconds apart, assigned to the ending poll's UTC day.
Boundary imprecision can reach 30 seconds. Baseline exclusions are entry counts across restarts/gaps, not unique IDs.
Daily enrollment counts and daily poll coverage concern different units; daily rows are not independent trials.
64,087 matched records reference 3,253 block hashes. 55 reorg-censored records reference one original block hash;
do not translate records into independent reorganization events or attacks.

The source index, source-manifests and snapshot preserve original ZIP URLs, SHA256 values and snapshot times.
Verify the original ZIP hash against snapshot.json and each cohort/polls member against its original manifest.
This package includes the selected data and frozen published collector source for inspection.
Running the collector is neither required nor performed by the verifier. Node credentials are not included.
The inputs map in snapshot.json covers selected CSVs and immutable source metadata.
SHA256SUMS.txt covers every package file other than itself and the ZIP container. Hashes detect differences,
not third-party attestation, provenance beyond the named source or an off-site backup.

LIMITS
One pruned walletless Core node, nominal 15-second polling, txid first 8 hex integer divisible by 16.
Public mempool/block lists only; no individual private MWEB timings. Preexisting baseline entries excluded.
Very fast between-poll confirmations may be missed. Both endpoints are observations, not precise event times.
No fee rates/vsizes, merchant policies, exchange crediting, population confidence intervals or speed guarantees.
Later chain changes can change membership; version 1 remains a frozen historical report. No pilot files replaced.
Article: https://litecoin.watch/articles/one-week-at-a-litecoin-node-one-five-and-ten-minute-confirmations
Method: https://litecoin.watch/uploads/research/confirmation-week-20261014-v1/README.txt
