This is a speculative report translated for non-specialists. The narrator is an investigator who arrived long after humans were gone. Everything described as measured relies on real Bitcoin mechanics: block intervals, difficulty/target, timestamp rules, and data available from block headers and the coinbase transaction.
We arrived on a silent planet. The last clocks still ticking were embedded in a ledger whose authors were gone.
REPORT START
Team: Survey Unit 3
Artifact: Global ledger (“Bitcoin”)
Technique: Lightweight chain analysis (headers + coinbase), mapped to solar time
Method
We analyzed the digital artifact known as Bitcoin using what we identified as block headers (timestamp, target / “bits”, version) and each block’s coinbase transaction (height, output value, and tag text).
From our previous initial review we’ve constructed the following data points:
- Fees were treated as: coinbase output − programmed subsidy (fees actually claimed by the miner).
- Timestamps were calibrated to the planet’s solar day and year and bounded by Bitcoin’s median-time-past (MTP) rule.
- Evidence of tip contention (stale blocks) was inferred from timing irregularities and MTP edge effects; where any stale-block archives survived on isolated nodes, they corroborated those periods.
- Difficulty retargets occurred every 2016 blocks with the actual_timespan clamped to 0.25×–4× of the two-week target, implying a per-epoch difficulty change bounded to at most 4× in either direction.
Findings
Cessation of payments
We recorded ΔH (blocks before present) to be ≈ 86,000. Coinbase outputs were equal to the programmed subsidy, implying fees ≈ 0. Over that same interval, average block spacing settled near ~60–70 minutes with a long-segment mean of ~65 minutes.
Interpretation: Human-directed payments had ceased. Mechanical issuance continued.
Dating: 86,000 blocks × ~65 minutes ≈ ~10.6 years before our arrival.
Power-source timing signatures
Post-collapse block arrivals were not memoryless. Diurnal and seasonal cadence encoded the unattended power mix:
- Daytime clusters with nighttime gaps repeated across low-latitude longitudes → unattended solar with degrading storage.
- Irregular multi-hour bursts punctuated by multi-day voids at mid-latitudes → wind that faulted during storms and wasn’t reset.
- Persistent overnight presence at a few longitudes → small hydro or geothermal operating islanded.
We aligned repeated intraday timestamp clusters to local solar noon to estimate longitude bands of surviving sites. The strength of seasonal variation in block arrivals gave coarse latitude bands. Precise site coordinates were not recoverable.
Difficulty terraces (the fade, timed)
Immediately after the hashrate shock, average block time jumped from ~10 minutes to hours. Because difficulty only retargets every 2016 blocks and each epoch’s change is bounded, the chain formed terraces, plateaus of near-constant average interval separated by discrete down-steps.
Representative sequence observed in the global ledger:
- Terrace A: ~16–17 h/block for 2016 blocks → elapsed ~3.8 years.
- Terrace B: ~4.1 h/block for 2016 blocks → ~0.95 years.
- Terrace C: ~62–65 min/block for 2016 blocks → ~87–91 days.
- Terrace D: ~15–16 min/block for ~22 days, after which renewed hardware failures re-slowed the chain.
Where residual hashrate was ≈1% of pre-event, Terrace A alone spanned ~3.8 years at ~16.7 h/block. At ≈0.1%, the same 2016-block epoch would have stretched to ~38 years at ~167 h/block, still within the protocol’s adjustment bound. One region’s cadence matched the ~16–17 h/block case.
How to read a terrace (worked calc):
Epoch length = 2016 blocks. If the observed interval on a plateau is 16.7 hours, elapsed time for that epoch ≈ 2016 × 16.7 h ≈ 3.84 years.
Network decay captured in the record
Once accurate clocks vanished, miner timestamps drifted in coherent regional patterns. Bitcoin’s MTP rule limited abuse of timestamps (each new block had to be later than the median of the prior 11) but did not eliminate drift signatures.
Interval variance and clustered MTP-bounded timestamp advances revealed intermittent partitions and tip contention; when any link resumed (e.g., satellite, microwave), competing branches reconciled and only the winning branch remained canonical.
Without preserved stale-block archives, measured contention is a lower bound.
Maker marks that outlived their makers
Coinbase tag strings (pool labels) and stable nonce/version fingerprints persisted for years after fee activity ended. Defaults were never changed once operators were gone, leaving software/hardware families identifiable in the record. (Coinbase tags are visible via the coinbase transaction; headers alone do not carry them.)
Dating key events (worked examples)
- “Payments ended.” Window where coinbase output = subsidy began at ΔH ≈ 86,000. Using the observed ~65 min/block: ~10.6 years before present.
- First post-shock retarget completed. The initial 2016-block reduction finished ~3.8 years after the hashrate collapse (plateau at ~16.7 h/block).
- Final detectable hydro cadence. The last night-heavy, near-constant signature ceased ~1.9 years before present; the prior seven spring seasons showed increasing multi-day outages consistent with intake clogging and flood damage.
All conversions use observed segment averages, not the nominal 10-minute target.
Duration estimate (how long machines ran)
- Minimum confirmed: >10 years after economic activity ceased (from fee collapse to last hydro-like cadence).
- Plausible upper bound (regional): Multi-decadal operation at extremely low hashrate, where a single 2016-block epoch spans decades due to the adjustment bound.
The only requirements were: (a) at least one surviving power source and (b) an intermittent path for some blocks to reach the global network.
Summary report
Ultimately, the ledger shows when payments stopped, how energy tapered, how networks frayed, and how long unattended machines kept writing time, enough to reconstruct the end of activity from headers and coinbase alone.
END OF REPORT
What readers should take from this
- Bitcoin behaves like an instrument. Difficulty rules and timestamp constraints transduce physical reality, power availability, operator absence, and network partitions into a durable time series.
- Physical failure, not price, ended the write. Dust, clogged screens, tripped breakers, drifting clocks, broken links.
- These forensics apply today. Block spacing, fee pressure (via coinbase delta), timestamp drift, and retarget dynamics are actionable diagnostics for present-day outages and partitions.
Limits
- Longitude bands were estimable; precise sites were not. Latitude was inferred only coarsely from seasonality strength.
- Fully isolated “shadow mining” may have produced blocks that never reached the global ledger.
- Without preserved stale-block archives, contention estimates are lower bounds; some races leave no canonical trace.
- Once synchronized time sources failed, MTP primarily preserved relative ordering, not accurate civil time; long-range calendar dates carry additional uncertainty even when intraday/seasonal structure is clear.
- In very low-hashrate regimes dominated by a single surviving operator, timestamps could be marched within MTP limits, partially masking diurnal signatures; cross-checks with nonce patterns and coinbase tags mitigate but do not eliminate this.
- Most OP_RETURN payloads were not decodable at scale and were not interpreted.
