The stars were where the catalogue said
The episode presents Kozyrev’s report that a detector on a telescope responded to stars at their true present and future positions. The published detections do match the positions worked out at the time, parallax errors included. Recomputed with modern parallaxes, they miss by 13 to 27 arcseconds. The agreement is with the calculation.
The verdict
ProvisionalNothing there
The published star detections agree with the parallaxes of their day, including ones since revised by 2-5 sigma, and miss the modern positions by 13-27 arcsec. The agreement is with the calculation, not with the sky.
Why provisional. Only three stars qualify for the test. Every table value reaches the lab through a sympathetic 2025 reproduction (Rabounski & Borissova, Progress in Physics 21, 99), which argues the effect is real; the Russian originals of 1978 and 1980 were not read.
For all 3 test stars (4 of 4 published rows), the detection is worse by more than 4″ against the modern prediction than against the one computed at the time.
Planted data built from the modern positions came back “tracks the sky”; planted data built from the old ones came back “tracks the old calculation”. The decision rule can return either answer.
The reported 2°4.6′ is the angle the sky turns in 8.3 minutes. That is where the Sun will appear, about 365 times farther than the Sun’s real offset.
The sources the lab read give Kozyrev’s death as 27 February 1983 in Leningrad, aged 74, with no cause stated. None describes it as mysterious. A history note, with no verdict.
The claim, as presented. The episode presents Kozyrev's claim to detect stars at their calculated present and future positions, and the claim that he 'died mysteriously'.
The Why Files, “Bending Time: The Successful Time Travel Experiments using Kozyrev Mirrors”, June 2023 · the episode
The registry’s note on sources. Every claim attributed to The Why Files below comes from the episode's podcast transcript (ep. 113, podscripts.co). A research subagent summarised it (2026-09-28). The same day, before any verdict was published, the specific claims each published verdict rests on were re-checked against the transcript in a second, independent read, and both reads agree. Transcript reads go through a summariser, so treat exact wording as close paraphrase, not quotation.
Checked against the episode transcript: yes, 2026-09-28. A second, independent read of the transcript confirmed the claim as described above: torsion detected at a star’s past, calculated present (the strongest reading) and predicted future positions. The verdict is about the published star detections, as the 2025 reproduction prints them.
Preregistered in the HI-11 PREREG.md, section “HI-15” (commit da01555), before any number was computed. Self-graded. The lab notebook is not public yet.
The size of the offsets
A star’s “true present” position is ahead of where it appears by its proper motion times the light-travel time, which works out to the angle vtan/c whatever the distance. The “future” position is the same step again.
| Star | Parallax (mas) | Proper motion (″/yr) | Light time (yr) | True − apparent | Future − apparent |
|---|---|---|---|---|---|
| Barnard’s star | 546.98 (Gaia EDR3) | 10.393 | 5.96 | 62.0″ | 124″ |
| Arcturus | 88.83 (Hip 2007) | 2.279 | 36.7 | 83.7″ | 167″ |
| Procyon | 284.56 (Hip 2007) | 1.259 | 11.5 | 14.4″ | 29″ |
| Sirius | 379.21 (Hip 2007) | 1.339 | 8.6 | 11.5″ | 23″ |
| Deneb | 2.31 ± 0.32 (Hip 2007) | 0.0027 | ≈1400 | 3.9″ (±14 %) | 7.7″ |
| Rigel | 3.78 ± 0.34 (Hip 2007) | 0.0014 | ≈860 | 1.2″ (±9 %) | 2.4″ |
For scale: the detector sat behind a spectrograph slit covering 2″ of sky. Annual aberration shifts every star by up to 20.5″, atmospheric refraction by about 58″ × tan(zenith distance), and seeing blurs by 1–3″. The offsets can be resolved in principle only for nearby, fast-moving stars, and only with arcsecond bookkeeping of aberration, refraction and the telescope’s drive rate.
Then against now
Each published row, recomputed with the modern parallax while keeping the published proper motion and aberration. A star enters the test when its predicted offset moves by at least 8″. The rule, written down first: if at least two thirds of those stars are worse against modern values by more than 4″, the detections track the old calculation.
| Star (row) | Parallax then → now (″) | Predicted then | Predicted now | Detected | Off from then | Off from now |
|---|---|---|---|---|---|---|
| β And (Lavrent’ev 1989) | 0.043 → 0.0165 | −24.9″ | +1.9″ | −25.4″ | −0.5″ | −27.3″ |
| ε And (Kozyrev, 21 Oct 1977) | 0.031 → 0.0192 | −41″ | −55.7″ | −38″ | +3″ | +17.7″ |
| ε And (22 Oct 1977) | 0.031 → 0.0192 | −41″ | −55.7″ | −43″ | −2″ | +12.7″ |
| γ Boo (Kozyrev 1978) | 0.016 → 0.0379 | −43″ | −29.7″ | −50″ | −7″ | −20.3″ |
Result: 3 of 3 stars, 4 of 4 rows, worse against modern values. The old parallaxes carried stated errors of ±0.005–0.007″. The modern values differ from them by 2.4σ (ε And), 3σ (γ Boo) and 5σ (β And). A detector that sensed the stars themselves would have disagreed with the catalogue here. The published detections agree with the catalogue, errors and all.
The rest of the checks. For 15 rows whose prediction moves by less than 2″, the largest change in residual is 1.5″; that cannot fail, so it is reported as a property only. Rescaling by the modern proper motion as well gives the same decision, with 6 stars in the test, all worse against modern values. For γ Boo’s “future” position the detector gave −97″, against −86″ from twice the old prediction and −59″ from twice the modern one.
The residuals against the old calculation are 0–4″, about the width of the slit. That is what one would expect if the peaks in the sweep were read near the positions the observers had calculated. The arithmetic cannot say how that happened. Reading a noisy bridge record, drift and heating are candidates, and none of them is a finding.
Kozyrev’s 1978 paper also presents the detector as a way to measure parallax. The rule excluded both stars measured that way: ρ Per came out at 0.0040″ against a modern 0.0106″, and α Ori at 0.0067″ against 0.0066 ± 0.0008″, which agrees within a large uncertainty.
The Sun
A 1990 report placed the Sun’s “real position” 2°4.6′ ahead of the visible Sun. The sky turns 2°4.8′ in 8.3 minutes because the Earth rotates, so that is where the Sun will appear 8.3 minutes later.
In an inertial frame the Sun’s present direction differs from its apparent one by about 20.5″, some 365 times less. The published number comes from the wrong frame, so it cannot be the position an instant signal from the present Sun would point to. This checks the claim as the 2025 reproduction reports it; the 1990 original was not read.
History notes
From the sources the lab read. Notes, with no verdicts.
Death. 27 February 1983, Leningrad, aged 74. No cause is stated, and no source the lab found describes anything mysterious.
Imprisonment. Arrested in November 1936 and released in December 1946.
Alphonsus, 1958. Kozyrev reported a bright patch and an emission spectrum on the Moon. The article the lab read does not say it was independently confirmed, and this episode did not settle it.
The 1960 Academy commission found the theory not built on clearly stated axioms and the experiments not accurate enough to support conclusions.
A fair test would need a blind scan in which the operator does not know where the predicted positions are, positions taken from modern astrometry, some fake targets, and a group independent of the original one.
- The Russian originals were not read. The table values come through a sympathetic 2025 reproduction, and transcription errors are possible: θ Peg’s proper motion is printed as 0.0272, where 0.272 is what fits its row.
- The conventions for right ascension, the epochs, and how aberration was applied were not modelled. The test changes only the parallax, so it leans on the published aberration values.
- Only three stars are in the test.
- The original chart records and the scan procedure were not examined, and neither were the reports of the detector responding through a shutter, a mirror cover or a dome.
- The episode was checked through its podcast transcript, read through a summariser, not by watching the video.
Read: Rabounski & Borissova, Progress in Physics 21 (2025) 99 (in full) · SIMBAD (queried 2026-09-28: Hipparcos 2007, Gaia EDR3) · Nikolai Kozyrev, Wikipedia. Background: Commission on Pseudoscience, Torsion field (Wikipedia).