A scientist casually pressing two fingers onto the anthropogenic pan of a laboratory balance, lifting natural variability into the air.

Science Blog · Methods

Method bias in climate records: proxies, albedo, and the assembled isotopic signature

The same measurements support different public conclusions depending on slope, weighting, smoothing, flask flags, and model corrections the original papers describe as required. This article keeps the frame on bias in methods, except where the fossil-fuel δ¹³C “fingerprint” is assembled after the flask and where correspondence used the word “hide” for a tree-ring decline.

Climate debate often treats a published curve as if it were a thermometer reading. The labs below treat the curve as a processed object: a local ice-core archive with a movable calibration slope; a hockey-stick shaft that flattens the Medieval Warm Period and splices over a volcanic 1940–70 albedo dip; a firn/ice δ¹³C reconstruction that the authors say agrees with Cape Grim “only after” gravitational, diffusion, and calibration corrections; a recent temperature series that tracks absorbed sunlight more tightly than a top-of-atmosphere energy-imbalance residual; an ocean-heat inventory in which deglaciation stored twenty times the industrial pile, with almost no ice-core CO₂ swing across the Medieval–Little Ice Age wiggle.

Interactive sources: Summit Uncertainty · Hockey Stick Lab · δ¹³C Ledger · Nikolov & Zeller atlas · Zettajoule ocean atlas · Stomata vs Ice · Paleoclimate nested timescales

1. Proxy thermometers are local, smeared, and calibration-dependent

The Summit lab opens with the sentence the public debate usually skips: the Alley 2000 ice-core series is a local archive, not a global thermometer. Method knobs — slope, elevation, site noise, the 1855 firn cutoff — can put the Holocene Thermal Maximum warmer than Summit today, while a naïve splice to Arctic instrumental temperatures argues the opposite.

Summit Uncertainty lab: Alley 2000 proxy chart, uncertainty cards, and method-bias panel
Figure 1. Summit Uncertainty lab. Left: Alley 2000 GISP2 δ¹⁸O anomalies versus a 1000–1850 baseline, with Arctic GISTEMP 64–90°N spliced as a sharp black spike at right. Right: which way each method bias moves “HTM vs today.” Open Summit Uncertainty

With the published knobs the lab reports an adjusted Alley HTM of −28.70 °C at Summit against an honest same-site “today” of GISP2 2001–2010 = −29.90 °C — HTM +1.20 °C warmer than that mean. The Arctic instrumental number often used as “now,” GISTEMP 64–90°N 2015–24, is +3.38 °C on a different baseline and a different place.

Cards comparing adjusted HTM, honest Summit today, and Arctic GISTEMP now
Figure 2. The lab’s comparison cards. Ice HTM −28.70 °C; honest Summit today −29.90 °C; “misleading now” +3.38 °C from Arctic GISTEMP, not Summit. Method bias can hide a Holocene that was warmer at Summit, or invent an unprecedented spike the borehole never measured. Open Summit Uncertainty
Direct temperatures of GISTEMP 64–90°N measure a different place. Using them as “now” against GISP2 “then” is how an overlay can hide a Holocene that was warmer at Summit, or invent an unprecedented spike the borehole never measured. Method bias can move the ice the other way.

δ¹⁸O is not a pure thermometer. Nearby Greenland cores still correlate only about r ≈ 0.4 because moisture source, transport, seasonality, elevation, drift, firn diffusion, and vapor exchange also move the isotope. Calibration slope was allowed to wander by about a factor of two. Boreholes smear multi-decadal wiggles. The published Alley series stops in the 1850s because snow is not ice yet. Dating of layers is the strong part; temperature is the weak part (±1.4 K on gas-isotope inversions versus ~0.05–0.1 K on instruments).

Six reasons the overlay is not a comparison, including d18O drivers
Figure 3. “Six reasons the overlay is not a comparison,” including the driver list that keeps δ¹⁸O from being read as T. Open Summit Uncertainty

2. The hockey stick flattens a volcanic century

Mann, Bradley and Hughes (1998/99) drew a flat millennium with a sharp twentieth-century blade. The Hockey Stick Lab rebuilds that shape from the published tree-ring network and then shows what the same century looks like once volcanic sulfate and a single overweighted site are allowed to speak.

McIntyre and McKitrick (2005) showed the blade is what you get when one strip-bark bristlecone site — Sheep Mountain, ITRDB CA534 — is weighted over 390 times the rest of the network. Fifteen bristlecone and foxtail sites accounted for 93% of that first principal component. Many of those trees are strip-bark; Graybill and Idso (1993) already argued their twentieth-century surge is CO₂ fertilization, not temperature.

Hockey Stick Lab: Northern Hemisphere reconstruction with volcanic eruption markers and 1910–1938, 1940–1970, 1976–2026 trend cards
Figure 4. Hockey Stick Lab. Hard stick from 2% Sheep Mountain weight. 1910–1938 already climbing at +1.98 °C/century; 1940–1970 trough −0.72 °C/century on a volcanic albedo dip (Bezymianny, Agung, Taal, Awu, Fernandina, Fuego); 1976–2026 blade +2.95 °C/century. Without volcanic dimming the century is one moderate climb, not a pause plus a blade. Open the Hockey Stick Lab

Sulfur dioxide from those VEI 4–5 eruptions oxidizes to stratospheric sulfate. The droplets scatter incoming sunlight — higher albedo, cooler surface. Mount Agung (Bali, 1963) is the documented drop through 1964–66. Together they dent an otherwise continuous twentieth-century rise that had already started in the 1910s, long before industrial CO₂ could carry the argument.

The stick hides that geometry by flattening the shaft and splicing the instrument on at the recovery. Tree-ring series are standardized growth indexes — ring width, not thermometers. HadCRUT5 is a mercury-and-satellite temperature field. Plotting them on one axis is the original splice. MBH98 also never published that each North American tree-ring series was centered on the 1902–1980 mean, not the full record, before principal components were taken. Short-centering throws any series that rises in the thermometer era far below zero in the earlier centuries. PCA then hunts those series. That is how a Medieval Warm Period and a Little Ice Age disappear into a handle.

Hockey Stick Lab apples-and-oranges panel: tree-ring growth indexes versus HadCRUT5 thermometer
Figure 5. “Apples and oranges.” Each tree card is independently z-scored — a tall wiggle is that site’s own variance, not degrees Celsius. HadCRUT5 Northern Hemisphere is actual instrumental temperature, 1850–2025. The overlay desk rebases both to 1981–2010 so the shapes can sit on one axis; the units still are not the same quantity. Open the Hockey Stick Lab

3. Correspondence that treated the MWP and the post-1960 decline as problems to solve

The method knobs above are on the page in the papers. The Climategate correspondence, which Kenneth Richard (@Kenneth72712993) has repeatedly posted, is the layer that shows intent: not a secret temperature decline in the thermometers, but an explicit decision to stop plotting the tree-ring series where they diverge, and a running worry that a visible Medieval Warm Period would undercut the “unprecedented” claim.

“I’ve just completed Mike’s Nature trick of adding in the real temps to each series for the last 20 years (ie from 1981 onwards) and from 1961 for Keith’s to hide the decline.”Phil Jones, 16 November 1999, to Mann, Bradley and Hughes

A later processing note, which Richard quotes alongside that sentence, is sharper still: “we have applied a completely artificial adjustment to the data after 1960”; “we set all post-1960 values to missing in the MXD data set (due to decline), and the method will infill these, estimating them from the real temperatures — another way of ‘correcting’ for the decline, though may be not defensible.”

Kenneth Richard post quoting Climategate emails on hide the decline and setting post-1960 values to missing
Figure 6. Kenneth Richard posting the Jones “hide the decline” line next to the processing note that post-1960 MXD values were set to missing and infilled from the instrumental temperatures, “though may be not defensible.” The decline being hidden is the tree-ring series after ~1960, spliced over with thermometers so the blade continues. Open the post

Richard’s follow-up is the bristlecone point the lab dial makes quantitative: the Rocky Mountain strip-bark series produced hockey sticks; trees that were not bristlecone showed cooling after the 1950s following the 1900s–1940s rise. That is why the post-1960 values had to be replaced. The WMO diagram Jones was finishing did not disclose the splice.

Hide the decline graphic: proxy reconstructions versus the same series with instrumental splice after 1960
Figure 7. Graphic Richard has used at NoTricksZone: paleoclimate reconstructions with the proxy decline visible (left family) versus the same story after Mike’s Nature trick replaces the post-1960 tree rings with the instrumental blade (right). Xing et al. 2016, D’Arrigo 2006, Schneider 2015, Wilson 2016, Christiansen & Ljungqvist 2012. Open the NoTricksZone article

The Medieval Warm Period was the other obstacle. Jonathan Overpeck, writing as an IPCC coordinating lead author in January 2005, told Keith Briffa and Tim Osborn he wanted to “deal a mortal blow to the misuse of supposed warm period terms and myths in the literature.” Briffa wrote back that some of Peck’s messages could be read as trying to “nail” the MWP — “trying to say there was no such thing.” David Deming later testified that a correspondent he believed to be Overpeck had told him, years earlier, “we have to get rid of the Medieval Warm Period.” Overpeck denied the wording. The 2005 email is not in dispute.

Richard’s highest-engagement posts are the reconstructions that keep the MWP and the LIA in the picture instead of nailing them. Poland’s early-medieval July ~2.5–3 °C warmer than now; China ~2.8 °C warmer in the Roman and Medieval intervals than 1970–2000; Southern Hemisphere proxies from South America, Antarctica, Australia/New Zealand and the sub-Antarctic as warm or warmer than modern times. Those papers are why a flat shaft was so valuable.

Reconstruction showing Poland July temperatures around 20.6°C in medieval times versus 17.4°C today
Figure 8. From Richard: Poland’s July temperatures about 20.6 °C in medieval times versus 17.4 °C recently — Little Ice Age Julys not far from the modern mean. North American and European mean annual temperatures near the low end of the last 4,500 years in that compilation. Open the post
China reconstruction: Roman and Medieval Warm Periods about 2.8°C warmer than 1970–2000
Figure 9. Richard posting a China reconstruction: Roman Warm Period ~2.89 °C and Medieval Warm Period ~2.81 °C above the 1970–2000 mean. A shaft that has already erased those peaks does not need a volcanic 1940–70 dip to look unprecedented — the millennial context is gone before the blade is drawn. Open the post

None of this requires a conspiracy novel. It requires reading the methods section next to the mail. Jones’s “trick” and the “not defensible” infill are presentation choices. Overpeck’s “mortal blow” is a framing choice for an IPCC box. Together they are why a local, volcanic, calibration-dependent century could be sold as a unique blade.

4. Smoothing and weighting decide whether the present looks unique

Ice-core CO₂ and temperature are heavily smoothed archives. Plant stomata keep a growing-season pore count and still carry century-scale swings that only meet the ice at the instrumental splice. Presenting the smoothed ice as “the” CO₂ history, then declaring the industrial tick unique, is a resolution choice, not a new measurement.

Stomata versus ice-core CO2 during the industrial rise
Figure 10. Stomata vs Ice lab. Ice (points) climbs from ~286 to 370 ppm by the late 1990s and onward with Mauna Loa. Dashed stomatal means retain extra century-scale structure; Beck 2007 (dotted) spikes near 1940. Open Stomata vs Ice

On glacial timescales the nested paleoclimate desk shows eight Antarctic cycles keeping CO₂ between about 170 and 300 ppm, with the industrial rise as a vertical tick at the right edge. Raw versus 5-kyr / 200-yr smooth changes how large earlier wiggles look next to that tick. The official ICS clock also makes the point visually: everything measured by satellite sits in the top pixel of Earth history. That is context, not an argument that recent warming is unreal. It is an argument against treating a spliced, smoothed proxy as a single thermometer.

International Chronostratigraphic Chart on the paleoclimate desk
Figure 11. Instrument Record paleoclimate desk: ICS chart used as the official clock before the ice. Holocene is the sliver labelled modern time — 11,700 years. Open paleoclimate desk

5. Recent warming tracks absorbed sunlight more tightly than a CO₂ residual

The Nikolov & Zeller atlas does not claim Earth has not warmed. It claims the IPCC-style ledger — greenhouse forcing, top-of-atmosphere energy imbalance, stored heat — is the wrong way to count. Surface temperature tracks absorbed sunlight. On the 2000–2026 window the lab reports:

Nikolov and Zeller atlas showing ASR vs GSAT R2 0.84 versus EEI R2 0.45
Figure 12. Nikolov & Zeller atlas. Same months, same GSAT. Absorbed shortwave explains 84% of the variance; the energy-imbalance residual explains 45%. Open the albedo atlas

Equation 16 reconstructed from measured TSI and a CERES-style albedo path follows the GISTEMP–NOAA average through July 2026. The 2023 spike sits on a record-low planetary albedo driven mainly by reduced low cloud over oceans, not by a TSI jump and not mostly by polar ice. Polar ice is only a small share of absorbed-shortwave change.

Modeled versus observed GSAT using only albedo and TSI
Figure 13. Atlas time series: black curve is albedo + TSI only — no ΔF_CO₂, no optical depth, no EEI. If a greenhouse term were required, the solar-only model would miss the trend. It does not. Open the albedo atlas

Whether the cloud decline itself is anthropogenic, natural, or mixed is a separate question. The energy-budget arithmetic on this window does not require a large leftover CO₂ term once ASR is allowed to speak. That is method bias in the choice of diagnostic: forcing and EEI as the ledger, versus absorbed sunlight as the ledger.

6. Ocean heat is sunlight. Forcing is a lid we have not isolated.

The Zettajoule atlas puts the industrial ocean-heat pile next to the last deglaciation. Gebbie’s review of noble-gas mean-ocean-temperature estimates puts deglacial uptake at 12,000–20,000 ZJ — best estimate about 14,000 ZJ — against roughly 500 ZJ for the industrial ocean through ~2017. That is a factor of at least twenty. Kenneth Richard (@Kenneth72712993) has been posting that same inventory from the literature: deglaciation 20–30× today’s stored heat, MCA about 1,500 ZJ, modern about 500 ZJ — only a third of the way back to the ocean of a thousand years ago. Melting ~130 m of sea-level-equivalent ice took another ~15,800 ZJ of latent heat. Public charts that start in 1955 hide the rest of the inventory.

Zettajoule atlas: each square is 500 ZJ, deglaciation 28 squares, MCA 3, modern 1
Figure 14. Zettajoule atlas. Each square is 500 ZJ — the entire modern-era ocean-heat pile. Deglaciation ~14,000 ZJ (28 squares). MCA versus Little Ice Age 1,500 ZJ (3 squares). Modern era 1 square. Inventory, not rate: deglaciation ran ~10,000 years at ~0.1 W/m²; the industrial ocean is faster, and still a rounding error as a pile of joules. Open the Zettajoule atlas

The Medieval Climate Anomaly ocean held about 1,000 ZJ more than the year-2000 ocean. Preindustrial ice-core CO₂ sits quietly at ~277–285 ppm across that whole MCA–LIA swing of ~1,500 ZJ. The deep ocean still carries Little Ice Age water. Whatever mechanism filled and drained those zettajoules, it was not a CO₂ knob we can point to in the ice.

Common Era ocean heat: MCA about 1000 ZJ above year 2000, ice-core CO2 nearly flat
Figure 15. Common Era inversion (Gebbie & Huybers): MCA stored ~1,000 ZJ more than year 2000; the ~500 ZJ of modern uptake is one-third of the way back. Law Dome CO₂ is almost flat across the same centuries. Cycles are in the papers — glacial, millennial AMOC, Holocene drift into the LIA — they are not a 500 ZJ alibi, and they are not a CO₂ story. Open the Zettajoule atlas

That inventory sits on a physics point the energy-budget language usually skips. Seawater is heated by shortwave sunlight, which is absorbed over metres in the mixed layer. Thermal infrared from the atmosphere — the “downwelling longwave” in a forcing diagram — is absorbed in about ten micrometres of skin. It never reaches the volume that Argo and XBTs call ocean heat content. Greenhouse forcing and feedbacks are therefore not a heat engine for the ocean. In theory they thicken a lid on skin cooling (less net longwave leak, a little more time for the mixed layer to keep yesterday’s sunlight). That is a slower leak, not a new source.

Ocean heat enters as shortwave sunlight, not as greenhouse forcingVisible sunlight is absorbed over metres in the mixed layer. Thermal infrared from the atmosphere is absorbed in a ten-micrometre skin and cannot heat the ocean volume. Forcing, if it does anything, can only slow cooling at that skin.SunShortwavevolume heatAtmosphere · H₂O, CO₂, cloudsLongwave IR (~10 μm) cannot enter the water columnSKIN ~10 μm — every downwelling IR photon stops hereMixed layer · metres to tens of metresThis is where sunlight becomes ocean heatThermocline / deep oceanNo sunlight left · infrared never arrived · heat only by mixing from aboveskin coolingForcing and feedbacks are not a heat source for seawaterIn theory a greenhouse lid slows skin cooling. That lid has not been isolated in the ocean-heat inventory.
Figure 16. All ocean heat that is actually in the water arrived as sunlight. Longwave from the air dies in the skin. Forcing, if it does anything, can only slow that skin’s cooling. We do not have a verified isolation of that lid in the deglacial 14,000 ZJ, the MCA 1,000 ZJ surplus, or the industrial 500 ZJ.

Richard has been hammering that physics from the papers, not from a slogan. Absorbed solar radiation, unlike CO₂, directly heats the ocean and is what tracks global heat content. CO₂ longwave can radiatively touch only the top 0.01 mm of a ~1 mm thermal skin — about 1% of the skin, none of the mixed layer, none of the deep Pacific with its centuries-long lag. A 4 W/m² forcing on that film is a calculated 0.008 K skin-temperature change. That is not a heat engine for 0–2,000 m.

Kenneth Richard post: CO2 can only radiatively affect the top 0.01 mm of the 1 mm thermal skin layer
Figure 17. Richard on the deep-Pacific lag: CO₂ longwave does not reach water that takes centuries to mix. It stops in the top 0.01 mm of the millimetre-scale skin. The volume Argo reports as ocean heat was never in that film. Open the post
“Longwave radiation penetrates less than a millimeter through water, thus its heating and cooling effects are restricted to the very surface of the ocean: the skin layer.”Wong and Minnett 2018, as posted by Richard — who adds that any thermal-skin longwave effect is dominated by cloud forcing, with CO₂’s longwave impact too small to detect. Open the post

The next step in the literature is the one the energy-budget diagrams skip: it is not possible for the extra energy in the thermal skin to be conducted into the bulk ocean beneath the viscous skin layer. Richard posts that line because it is the difference between a lid on cooling and a source of zettajoules. Nikolov and Zeller 2024, in a post of his that circulated widely, put the observational corollary: a global longwave radiative forcing of the sort the models attribute to rising CO₂ “does not exist in reality” — absorbed solar, not the greenhouse residual, accounts for the post-2000 trend.

Kenneth Richard quoting literature that extra thermal-skin energy cannot be conducted into the bulk ocean
Figure 18. Richard quoting the thermal-skin literature: CO₂’s radiative reach is the skin, and the extra energy there is not conducted into the ocean beneath the viscous layer. That is the published reason forcing is a lid, not a heat source — and why it has not been isolated as the cause of 12,000–20,000 ZJ of deglacial uptake. Open the post

The honest residual is large. Deglacial-average uptake (~99 mW/m²) is in the same neighborhood as geothermal heat through the seafloor. Modern imbalance estimates (0.5–0.7 W/m²) are faster, and still reconstructed from sparse floats, XBTs, and inversions — “every ZJ here comes from a model of a measurement,” as the atlas puts it. We do not yet have a closed, instrument-verified explanation of why the ocean held more heat a thousand years ago, at 280 ppm, than in 2000. Treating the industrial 500 ZJ as a greenhouse receipt is a choice of ledger, not a completed energy audit.

7. The isotopic “fingerprint” was assembled, not observed

This is the section where the labs go beyond ordinary calibration bias. The δ¹³C Ledger’s headline is the claim:

Delta-13C Ledger hero: isotopic signature assembled not observed, intercept -13.3 versus fossil -28
Figure 19. δ¹³C Ledger hero panel. Observed Keeling intercept −13.3‰ at four stations versus the fossil-fuel claim −28‰. Industrial Δδ¹³C ~1.6‰; largest ice revision 0.2‰. Open the δ¹³C Ledger
Francey et al. (1999) and the CSIRO GASLAB flask programme are cited as the fossil-fuel fingerprint in air: falling δ¹³C. Rebuild their record, strip the documented corrections, and the mixing intercept stays near −13‰ — not the −28‰ of coal, oil and gas. The fossil number is a model isoflux, not a measurement.

A Keeling plot of δ¹³C against 1/[CO₂] has intercept equal to the net isotopic signature of the added carbon. Seasonally adjusted flasks at Utqiaġvik, La Jolla, Mauna Loa and South Pole recover intercepts about −12.9 to −13.3‰ with R² near 0.99. If the rise were a fossil dump, that intercept would sit near −28‰ or drift toward it as the fuel mix changed. In the ledger’s epoch table it does not:

Keeling plot with observed intercept near -13.4 per mil and fossil line at -28
Figure 20. Keeling geometry as used in the ledger: observed mixing intercept near −13.4‰, fossil claim drawn at −28‰. The intercept does not migrate to the fossil line as emissions climb. Open the δ¹³C Ledger

Francey et al. (1999) themselves wrote that firn-air δ¹³C agrees with Cape Grim only after correction for gravitational separation, diffusion/disequilibrium, and a latitudinal gradient, and that “complex calibration strategies are required.” Those are not hidden emails. They are the methods section. Later Rubino revisions moved a ~0.2‰ South Pole firn mismatch on paper and applied ICEBASE selection plus a 50-year spline. Cape Grim baseline hours reject a large fraction of flasks; rejected air is systematically more depleted.

Table of documented isotopic processing corrections
Figure 21. Processing chain magnitudes compiled from Francey 1999, Allison/GASLAB, and Rubino 2013/2019 as laid out in the ledger. Yellow rows are the steps that most change the public fingerprint story. Gravity and diffusion sculpt tenths of a per mil. The leap from −13‰ to −28‰ is the isoflux term in an inversion. Open the δ¹³C Ledger

Call the physical firn corrections ordinary ice-core practice if you want — every high-resolution firn record uses analogues. The bias claim is narrower and sharper: the public talking point that “the isotopes prove fossil fuel” presents a hybrid reconstruction plus an inversion identity as if it were a raw flask intercept. The flask intercept is −13‰. The −28‰ number is assembled afterwards. That is why this article treats the isotopic signature as the case that crosses from ordinary method bias into fabrication of a fingerprint.

8. Alternative reconstructions meet dossiers, not just reviews

Scientists who keep these knobs visible report a second kind of bias: identity-based rejection and reputational packaging.

Judith Curry, former chair of Earth and Atmospheric Sciences at Georgia Tech and an AMS/AGU Fellow, has described a fortress mentality after she began emphasizing uncertainty and engaging outsider blogs. Search results filled with “climate denier” frames. She left her academic chair. DeSmog’s Climate Disinformation Database entry on her is the standard package: photograph, credentials, consulting clients, selected quotes, resignation framed as citing the “craziness” of climate science.

DeSmog Climate Disinformation Database profile page for Judith Curry
Figure 22. DeSmog database page for Judith Curry. Breadcrumb: Databases → Climate Disinformation Database → Judith Curry. The page leads with a portrait and then a compiled biography — a pre-built personal file that travels with the name. Open the DeSmog dossier

Ned Nikolov and Karl Zeller stated that manuscripts challenging the conventional greenhouse framing were rejected, sometimes without review, after editors Googled their names. They published under reversed-name pseudonyms (Den Volokin, Lark ReLlez) to get a content-first reading; one paper was later withdrawn over the names rather than the calculations. Their later invited Geomatics paper is the atlas shown above. DeSmog maintains a Karl Zeller page that leads with the pseudonym episode.

Willie Soon is the other recurring dossier: solar papers plus energy-related funding packaged as the story, with journal ethics reviews and congressional interest following the funding narrative more loudly than the energy-balance critiques.

DeSmog presents this as accountability journalism about fossil-fuel influence. The scientists above present it as a blacklist that makes dissenters unhirable and that hands activists a ready-made personal file. Political organizations that publish such dossiers know they will be used for targeting. What is not in dispute is the mechanism: a searchable personal dossier attached to a methods disagreement.

Harassment of climate researchers runs in more than one direction. Mainstream scientists have also received abuse and threats. The point here is specific: when the reply to a slope slider, a Keeling intercept, or an albedo residual is a compiled biography rather than a competing reconstruction, the bias has left the data.

What the figures carry

The labs are public. The underlying papers — Alley 2000, Mann et al. 1998/99, McIntyre & McKitrick 2005, Vinther 2009, Kobashi, Francey 1999, Allison/GASLAB, Rubino 2013/2019, CERES EBAF, Nikolov & Zeller 2024, Gebbie 2021, Gebbie & Huybers 2019 — are public. The Jones, Briffa/Osborn and Overpeck correspondence is public. Readers can move the sliders and watch the headline move. That is ordinary scientific hygiene. Treating the sliders as heresy is the bias that does not show up in the CSV.