What could have been known, release by release
The Signal Before the Signal
The design forecast said Advanced LIGO should see binary black holes. That is not the same claim as saying the first three-month run was more likely than not to see one. This instrument freezes the official public record at each release and keeps those questions apart.
At the last public ER7 sensitivity release, 9 July 2015
calculating
for at least one binary black hole detection in the planned first run
The frozen rule versus what followed
The rule lost.
It classified the run as unlikely. A confident event occurred. Its probability was not disproved by one event, but its Brier score against that outcome is calculating.
A score against the historical actors is refused. Their publications did not lock one comparable BBH probability.
First-run probability
calculating
Expected count, lambda
calculating
Poisson model, three calendar months
Range proxy
calculating
The probability is rebuilt from quantities available by the selected release. The 2010 collaboration paper supplied a realistic merger density and design horizons. The 2013 scenario supplied three months, a 40 to 80 Mpc BNS range, and 80 percent duty per detector. Later releases replaced the range proxy. Nothing uses GW150914's event time. The event enters only when the detection paper became public on 11 February 2016.
A future design year
calculating
From the published realistic design rate of 20 per year. This was a broad statement about the mature instrument.
This short first run
calculating
From the unrounded density and horizon calculation, then the released ER7 range, three months, and coincident duty.
The mass the record did not contain
The 2010 rate paper fixed every black hole at 10 solar masses because it said the mass distribution was not known well enough for a detailed model. The public detection paper later reported component masses near 36 and 29 solar masses. Before that release, the second case below stays locked.
A declared chirp-mass sensitivity exercise
calculating
This depth layer uses the leading inspiral scaling distance proportional to chirp mass^(5/6). It is a sensitivity calculation, not an exact high-mass search model. Merger and ringdown, the detector noise curve, cosmological redshift, search thresholds, spins, and the population distribution all matter. It cannot recover the unchosen historical branch.
What the frozen rule can and cannot say
The rule said "below one half," and the outcome was one confident detection. That makes the binary classification wrong. It does not make a low probability logically impossible. The official publications described wide ranges and strong anticipation, but did not provide one locked probability for a binary black hole detection in that specific run. There is no honest same-scale historical score to compute.
The counterfactual stops at scoring the rule against the outcome that actually followed. It does not claim what the collaboration would have done differently, or what would have happened if any operational choice had changed.
The check
The first anchor is Table IV and Table V of the 2010 rate paper. Its realistic BBH density is 0.005 Mpc^-3 Myr^-1. Its horizons are 161 and 2187 Mpc. The published rates are 0.007 and 20 per year. The browser recomputes them from volume:
N = R x (4 pi / 3) x (D_horizon / 2.26)^3
Uncertainties: the pre-detection merger-rate range spans orders of magnitude; released spectra were preliminary and had large calibration uncertainties; a BNS range is only a proxy for a BBH search volume; the Poisson model assumes a stationary population and independent events; the 2.26 orientation factor and Euclidean volume omit cosmological corrections; the archive is a bounded official-record manifest, not all of astronomy.
Free choices: the headline uses the smaller of the two last released ER7 detector ranges; before two ranges exist, it uses the sole released detector range, then the published 54 Mpc worked scenario, then the arithmetic midpoint of 40 to 80 Mpc; it uses the realistic density rather than the pessimistic or optimistic bounds; it squares 80 percent for coincident duty; it treats a probability below one half as "unlikely." The selector makes the detector-range choice movable. None of its options crosses one half.
Timestamp precision: arXiv releases carry UTC seconds. The DCC list supplies calendar days only, so the firewall treats those days inclusively and does not invent a time.
Open the complete 15-release replay
Every selected release appears. Rows before the run plan refuse to print a first-run probability. Rows after the public detection remain visible but are marked as outcome-known. The chart draws every defined point.
| Released | Record | First-run state | Bucket at selected cutoff |
|---|
Open source and licence notes
- Abadie et al. 2010: Tables IV and V, the 2.26 factor, horizons, fixed 10 solar-mass black holes, and the stated lack of a detailed mass distribution.
- Aasi et al. 2013: the three-month 2015 scenario, 40 to 80 Mpc BNS range, and duty-cycle assumptions.
- Singer et al. 2014: the worked 54 Mpc two-detector 2015 scenario.
- L1 ER7 spectrum: observed 6 June, released 8 July 2015, 62 Mpc and preliminary.
- H1 ER7 spectrum: observed 9 June, released 9 July 2015, 65 Mpc and preliminary.
- GW150914 detection paper: public release time, outcome, and 36 plus 29 solar-mass components.
- First BBH merger-rate inference: post-detection rate estimates and dependence on the unknown mass distribution.
No source paper or DCC data file is redistributed here. The shipped manifest contains bibliographic facts and short numerical parameters. Older arXiv records use the arXiv non-exclusive distribution licence; arXiv:1304.0670 is CC BY 4.0; the GW150914 paper is CC BY 3.0; the DCC cards state no explicit reuse licence. Full notes are in research/when-gravitational-waves-became-likely/README.md.