Uncited Press › The Three-Body Problem Open the interactive journal →
three-body · Military Science & Strategy

Where the Staircase Probe Went: Planning a Passive Detection Search over the Space of Off-Nominal Trajectories

Dr. Arkady Vesnin-Wu1
1 Fleet International Academy, St. Petersburg, Department of Autonomous Systems
Received 10 Sep 2026 · Revised 8 Oct 2026 · Accepted 25 Oct 2026

Abstract

The probe of the Staircase Program left its planned trajectory during the acceleration after a failure in the rigging of its sail, and tracking was lost. This paper reconstructs the space of off-nominal trajectories from 38 archived tracking fixes, in a reduction of the paper's own, and plans a passive detection search for six survey vehicles that observe from stations 30 AU from the Sun. Each vehicle has 40% of full charge and a window of 2,160 h. The reconstructed family has a median asymptotic speed of 14 km/s, a median range of 747 AU after 253 years, and a sky patch of 573 deg2 (95%) centred 19° from the nominal axis. The best plan uses 18 drift-scan strips, 480 deep near-infrared tiles and 42 wide-field tiles, for 5,269 vehicle-hours, and gives a recovery probability of 0.309. Energy limits the plan and time does not. The detection law was checked against 214 recovery trials of catalogued fragments (Pearson χ2 4.83 on 4 df), and the track fit against 8 held-out fixes. The probability ranges from 0.230 to 0.394 when the sail area is halved or doubled. The plausible alternative is a fragmented sail, which lowers it to 0.285 after weighting and to 0.024 if the fragmentation is certain. A single sail is the more probable case, since the final fixes show one return.

1. Introduction

The posterior-coverage approach to search planning was developed for survey vehicles that must find a faint body in a large volume with a fixed reserve of energy. It is applied here to the probe of the Staircase Program (Crisis Era, Years 5 to 7, or CE 5–7), a sail-driven probe that was pushed by nuclear pulse detonations laid along its path. The probe carried the Staircase subject, whose continued existence is unconfirmed. During the acceleration a failure in the rigging of the sail, a cable, took the probe off its planned trajectory, and tracking was lost.

Written in Deterrence Era, Year 59 (DE 59), this paper rests on sources that all predate that year. Nothing about the fate of the probe after the last fix is assumed or used. The tracking sheets were released through the Planetary Defense Council (PDC) records office (United Nations Planetary Defense Council, Records Office, 58 DE) and are held in the declassified series (United Nations Planetary Defense Council, Staircase Tracking Desk, 5–7 CE). Speeds, directions, masses and areas given below are outputs of the paper's own model, and the reduction of the sheets is not an official record.

Only the autonomous vehicles of the Oort survey have the endurance for such a search (Oort Debris Field Survey Office, 56 DE), and their limit is energy. The vehicles are assumed to be staged at observing stations 30 AU from the Sun before the window opens, and the transit lies outside the window and the charge budget. For example, a deep near-infrared tile of 6.15 h costs 0.368% of full charge with slewing, so a vehicle that may spend 40% can take at most 108 of them and a team of six can take 648.

Three quantities are sought, namely the instruments to fly, the observing hours to spend, and the probability of recovery that results under a fixed time and energy budget. Sophons read anything written, and a sophon may set the logged outcome of any single measurement. The plan is therefore assumed to be read in full, and no step of it depends on concealment. Interference is partly suspended in the Deterrence Era on terms set by Trisolaris (United Nations Planetary Defense Council, Secretariat, 2 DE), but the tracking sheets were made when it was not.

2. System Description

Thirty-eight usable fixes from five stations survive in the reduced record, and they span the first 6.4 h of the powered phase. Range, Doppler shift and two angles are carried by each fix, and the sheets end where the return was lost. Fixes made before the rigging failure are few, so the 38 are treated as one arc with a change of direction at the failure. The robust step of Section 3 follows an earlier method for sparse tracking (Varga-Nwosu, 55 DE).

Six survey vehicles make up the search team, and each is allowed 40% of full charge for observing and a window of 2,160 h, or 90 days. Transit to the observing stations lies outside both budgets and is not studied here. The team budget is therefore 12,960 vehicle-hours and 240 units of charge, where one unit is one percent of one vehicle's full charge. The ordering of moves between tiles follows the online replanning method of the author's earlier survey paper (Vesnin-Wu, 56 DE) and is not studied again.

Three passive instruments are admitted, and Table 1 lists them. The wide-field camera (OWC) is shallow and fast. The deep near-infrared imager (DIR) stacks long exposures at 6.15 h per tile and reaches a flux limit 27 times lower, and the drift-scan array (DSA) sweeps a strip that spans the whole sky patch in one pass. Flux limits are nominal figures from the instrument papers (Okonjo-Brandt, 56 DE, and Szabo-Quill, 58 DE) and are inputs of this model. The limits assume photon-counting apertures of about 200 m2 and are set by the galaxy background, from which a moving target is separated only by its motion across exposures. An active radar was considered and excluded for the reason given in Section 6. The signature is reflected sunlight. A deployed sail is taken as 1.0 × 106 m2 and a furled one as 4.0 × 104 m2, with an albedo of 0.8 from the film study of the teaching archive (Teixeira-Mohan, 29 DE). Light is scattered into π sr, and the vehicles observe from stations 30 AU from the Sun, which is 4% of the median range of Section 3. The station-to-probe distance is taken at its worst case, the heliocentric range plus 30 AU, so that the received flux falls as r-2(r + 30)-2 for a probe at r AU, and the detection range is the heliocentric range at which that flux equals the instrument limit. The sky patch is defined as seen from the Sun, and each pointing is corrected for the parallax of its station, which reaches 2.3° at the median range. No record fixes the sail area. Both areas are assumptions of the paper.

3. Analysis / Model

An off-nominal trajectory is described by a direction on the sky and an asymptotic speed. The speed is log-normal with a median of 14 km/s and a log standard deviation of 0.45, so that the 68% interval is 8.9 to 22.0 km/s, and radiation push on a deployed sail is folded into it. The speed is an output of the paper's reduction of the tracking sheets and not a recorded figure. Elapsed time is taken as 253 years, counted from the close of the program in Crisis Era, Year 7, which the paper takes as the launch year, through the rest of a Crisis Era of about 201 years to DE 59, with an uncertainty of a few years. The median range is then 747 AU, with 476 to 1,172 AU at 68% and 309 to 1,805 AU at 95%. The direction on the sky, as seen from the Sun, is Gaussian, with a standard deviation of 7.8° along the track and 3.9° across it, and its centre lies 19° from the nominal axis, or 2.4 standard deviations of the long axis. The 95% region is spread over 573 deg2, and only 0.0035 of the mass lies within 2° of the nominal axis, so the nominal corridor is not where the probe is likely to be.

The median of pairwise slopes is used in the fit, and it tolerates 29.3% of corrupted fixes, or 11 of 38, where ordinary least squares tolerates none. For example, one end fix displaced by five standard deviations moves the least-squares slope by 1.37 slope standard deviations, or 10.7° on the sky. The robust fit is chosen because any single logged value may have been altered.

Sail state enters as a mixture with weights of 0.65 deployed and 0.35 furled, taken from the strength of the return in the last fixes. A body at flux ratio q to the instrument limit is detected with probability 0.905 Φ(ln q / 0.925), where Φ is the standard normal distribution function. Both constants are fitted in Section 4. Averaged over speed and sail state, the chance that a probe inside a covered cell is detected is 0.465 for DIR, 0.190 for DSA and 0.082 for OWC.

Plans are built by fixing the number of DSA strips, placing them across the densest part of the patch, and filling DIR and OWC tiles by marginal gain until the charge or hours of each vehicle run out, with strips shared equally among the six vehicles, on a sky grid with cells 0.25° wide. Every strip count from 0 to 60 was scored. The best plan is reached with 18 strips, 480 DIR tiles and 42 OWC tiles, so that each vehicle flies 3 strips, 80 DIR tiles and 7 OWC tiles, and its recovery probability is 0.309, as Table 2 shows.

Energy is exhausted at 240.0 units, while 7,691 vehicle-hours of the window remain unused. Time does not bind.

Single-instrument plans on the same budget are used as benchmarks. A value of 0.266 is reached by DIR alone, with 648 tiles, which is 99.9% of the ceiling that the Gaussian patch allows for that area, so the mixed plan is 1.16 times better. DSA alone, with 42 strips, reaches 0.190. OWC alone covers the whole patch with 153 tiles in 245 h and reaches 0.082, or 3.8 times below the plan. The optimum is flat, and plans with 16 to 23 strips lie within 1% of the best.

4. Validation Against Field Data

Recovery trials from the survey seasons were used to check the detection law. In each trial a vehicle looked for a catalogued fragment of known cross section at a known position. Two vehicles logged each outcome independently. Of 219 trials, 214 had agreeing logs and 5 were dropped, and the 214 were sorted into six bins of flux ratio q, as Table 3 shows.

A ceiling of 0.905 and a softness of 0.925 in ln q were obtained by maximum likelihood. The Pearson statistic is 4.83 on 4 degrees of freedom, a p-value of 0.31, and the worst bin is q = 1.0, where 21 detections were seen against 17.2 expected, a deviation of 1.2 standard errors. The fragments tumble and a sail would not, and the fit covers q from 0.2 to 12 only, so a sail seen at higher q lies outside the tested range.

The track fit was tested by holding out the last 8 of the 38 fixes, and the model was fitted to the first 30 and used to predict the rest. One combined normalised residual per fix, formed from its four measured quantities, gives an RMS of 1.02 and a sum of squares of 8.28 on 8 degrees of freedom. Neither test reaches the extrapolation to 253 years, a factor of 3.5 × 105 beyond the 6.4 h of arc.

5. Failure Modes

Fragmentation is the first failure mode, since a parted cable could have broken the sail into pieces. One return is shown by the last 12 fixes, and a split into 50 equal pieces worsens their fit by 2.6 in chi-squared. With prior odds of 1 to 3 against a split, the posterior weight of a split is 0.083. If the split is certain, the detection range of each piece falls by a factor of 0.367 and the plan reaches 0.024, a lower bound because the pieces would spread, and the weighted figure is 0.285.

Sail area and signature dominate the sensitivity. A value of 0.230 is given by half the area and 0.394 by twice the area, while a flux limit three times worse for every instrument gives 0.188 and one three times better gives 0.445. In the worst tested case the sail is furled and the plan reaches 0.068, or 4.5 times below the headline, whereas a known deployed sail gives 0.439.

Elapsed time and energy matter less. An elapsed time of 240 years gives 0.334 and one of 265 years gives 0.287, and a charge allowance 25% lower gives 0.274 while one 25% higher gives 0.337. A reserve of 5 units on each of two vehicles for confirming a candidate takes the plan to 0.304. A candidate must be confirmed by two instruments on two vehicles before it is accepted, since one logged detection can be altered. Two-vehicle confirmation guards only against isolated alteration, and every probability in this paper is conditional on the partial suspension of interference holding for the duration of the search.

Corrupted fixes are limited by the robust fit to 11 of 38, and a larger or cross-station alteration would shift the patch by an amount that the record cannot reveal.

The form of the model is the last source of failure. It is assumed that no force other than gravity and radiation pressure acted after the final fix. The record cannot test that. Range and direction are treated as independent, although a faster probe on a deployed sail would follow a slightly different path. Active illumination by radar would raise the detection range but is excluded, since it adds emission that the framework named below treats as a cost.

6. Conclusion

This paper reconstructed the space of off-nominal trajectories of the Staircase probe from 38 reduced tracking fixes and planned a passive search for six survey vehicles that observe from stations 30 AU from the Sun. In the best plan 18 drift-scan strips, 480 deep near-infrared tiles and 42 wide-field tiles are flown over 5,269 vehicle-hours, and the recovery probability is 0.309. The probability ranges from 0.230 to 0.394 when the sail area is halved or doubled.

The plausible alternative is a fragmented sail, which would cut the probability to 0.024. A single sail is the more probable case, because one return is shown by the last 12 fixes and the weight left for a split is 0.083. The weighted figure of 0.285 remains conditional on a signature that the paper supplies. Energy limits the plan, so the first gain in recovery could come from more charge and not from more hours.

We write within the Luo-Ji / Ye framework, and under it the search treats emission from the Solar System as a cost, so only passive instruments are admitted.

The plan says nothing about the state of the Staircase subject. A track of 6.4 h is carried over 253 years, the detection law was tested on tumbling fragments only, and the sky patch is assumed Gaussian with one peak. Future work will add a late impulse to the family, and the independence of range and direction is the simplification to be removed first.

Staircase Programoff-nominal trajectorysearch planningsurvey vehiclespassive detectionposterior coverageenergy-limited survey

References

  1. United Nations Planetary Defense Council, Staircase Tracking Desk (5–7 CE). Staircase Program tracking sheets, powered phase, range and Doppler fixes. PDC Declassified Holdings, New York, Series SP, sheets 1–14.
  2. United Nations Planetary Defense Council, Records Office (58 DE). Release note for the Staircase Program tracking sheets. Bulletin of the UN Planetary Defense Council, declassified series, Bulletin 37, annex 1.
  3. United Nations Planetary Defense Council, Secretariat (2 DE). Record of the Trisolaran declaration of partial suspension of sophon activity, as received and logged under the Dark Forest Deterrence. Bulletin of the UN Planetary Defense Council, declassified series, Bulletin 6, annex 3.
  4. Oort Debris Field Survey Office (56 DE). Survey catalogue release, seasons one to six, fragment state vectors and breach records. Oort Debris Field Survey Bulletin, 10, supplement 2.
  5. Vesnin-Wu, A. (56 DE). Online replanning of multi-vehicle survey teams under energy and time budgets in the Oort debris field. Oort Debris Field Survey Bulletin, 10, 41–66.
  6. Okonjo-Brandt, P. (56 DE). Deep near-infrared imaging of faint cold bodies from survey vehicles, flux limits after long stacking. Fleet International Proceedings, 77, 12–38.
  7. Szabo-Quill, E. (58 DE). Drift-scan arrays on survey vehicles, strip sensitivity and slew cost. Oort Debris Field Survey Bulletin, 11, 9–31.
  8. Teixeira-Mohan, N. (29 DE). Reflectance and scattering of metallised sail film at large phase angles. Fleet International Academies, Combined Teaching Archive, Series F-6, file 9.
  9. Varga-Nwosu, D. (55 DE). Robust trajectory fitting from sparse tracking fixes by the median of pairwise slopes. Fleet International Proceedings, 76, 150–171.
Read this article inside the full journal experience — browse by faculty, search across universes, and filter by discipline.
Open in Uncited Press →