Two Lossy Compressions in One Day
Two Lossy Compressions in One Day
Two beliefs I met in a single day, what each one keeps and what it discards, and the loss that remained in my own corrections of them
Field note, then analysis · two encounters reported plainly, then a flat account of the errors, the handling, and the residue
Version 1 · June 2026
Two people I know said something false to me on the same day, a few hours apart, and I have been turning both over since.
The first told me, with the quiet certainty of someone reporting a fact, that homeopathy works. Not as a guess, as something they had lived: they had taken it, they had gotten better, and for them that settled the matter. The second, later, asked a question rather than asserting an answer. Does cold weather make you sick, the air conditioning and the cold drinks and the chill, or is it the viruses?
I should say plainly what these people are not, on this page. I have reduced them to the two claims they happened to make and left out everything else: their histories, their names, who they are to me. That is deliberate. Those lives are not mine to publish; the claims are the only part of them this essay has any business with. So they appear here thin, on purpose, and the thinness is a boundary rather than an oversight.
I want to be honest about one more thing before the analysis, because otherwise it would sit under everything and sour it. I enjoy moments like these. Most of a working day is narrow: the same constrained system, the same local tasks, cognition pressed into a small domain. These two exchanges were the opposite, a place where the general picture of how the world works suddenly had to touch live ground, in a room, with consequences for a real person. That is what I enjoy. Not that someone was wrong in front of me; that a piece of the general model abruptly mattered.
And I am not clean at this. With the first person I reached, almost reflexively, for the blunt word. I said it was fake, a sham, which is the move most likely to make someone defend a belief rather than loosen it. In the same conversation I also did the better thing, almost by accident: I took their recovery seriously as real before I touched the explanation of it. Both are in the record of that hour. I am reliable at seeing these errors and still learning to say them in a way that lands. The seeing and the saying are different skills, and only one of them is dependable in me yet.
That is the warm half: the day, the people, what I felt about the encounters. The rest changes register on purpose, from the encounter to the account of it: what the two beliefs are, what each one quietly throws away, what I did with them, and, at the end, the loss that survived in my own corrections.
Two compressions, failing in opposite directions
Both beliefs are the same kind of object: a lossy compression of a causal model. A person observes a sequence, I took the remedy and I recovered, or the weather turned cold and I got sick, and compresses it into a short causal claim. Compression is not the error. Every usable belief is a compression; the full causal detail of even a head cold is past anyone's capacity to hold. The error is the loss. What gets discarded in the squeeze, and whether the discarded part was load-bearing.
The two beliefs fail in opposite directions, and the difference matters more than the label they share. Homeopathy compresses "I recovered" into "the remedy worked." This is false attribution: a real event, the recovery, is bound to a cause that did nothing, and the discarded part is the actual cause, whatever it was. "Cold makes you sick" compresses "cold contributed to my getting sick" into "cold caused my getting sick." This is over-weighting: a real but partial factor is promoted to the whole story, and the discarded part is the necessary cause, the virus, without which the cold does nothing.
Only the first is pseudoscience. Homeopathy is a two-century doctrine with a stated mechanism, practitioners, and the outward dress of medicine, immune to two hundred years of disconfirmation. "Cold makes you sick" is not a doctrine; it is a folk overgeneralisation with a true kernel. Filing both under "pseudoscience" would itself be a lossy compression, discarding exactly the distinction that makes them worth separating. The honest category that holds both is the larger one: two failures of fidelity, of different type.
Homeopathy: a real recovery tied to a cause that did nothing
Start with the mechanism, because the mechanism settles the case before any clinical trial is consulted. Homeopathic remedies are made by serial dilution. A common preparation is 30C: thirty hundred-fold dilutions in series, a factor of 10−60. Avogadro's number, the count of molecules in a mole, is about 6×1023. The arithmetic is not subtle. Past roughly the twelfth such dilution a dose contains, on average, less than one molecule of the original substance; by 30C the probability that a single molecule remains is, for any practical purpose, zero. The remedy is water, or a sugar pill carrying the memory of water. It is inert. This is not a finding from a study that could be re-run with a different result. It is a consequence of counting, and the counting is not in dispute.
So I do not need the clinical literature to know the remedy contains none of what it claims to act through. That is settled by counting: a mechanism cannot operate on a substance that is not there. Whether the pill therefore has no effect at all is a slightly weaker claim. Efficacy is finally about outcomes in bodies, and chemistry makes a null outcome overwhelmingly likely rather than certain. The trials close that gap. The largest review of the evidence, Australia's NHMRC, assessed the body of trials and found no health conditions for which there is reliable evidence that homeopathy is effective. I lead with the dilution anyway, on purpose: its exact wording, no reliable evidence of efficacy, has been contested at the margins by interested parties, and the arithmetic has not. The absence of the substance is settled by counting, the absence of any effect by evidence, and counting is the harder of the two to argue with.
Then the person's real experience, which I did not dispute and will not. They got better. That is not in question, and treating their recovery as imaginary would be both false and the surest way to lose them. The only question is what to attach the recovery to. I told them, flatly, that it was placebo. The reasoning is by elimination: if the remedy is inert, and they improved anyway, and the improvement had anything to do with taking the remedy, then the one channel left open is the expectation around it, placebo. As a rebuttal to their specific claim, the remedy did this, that is correct and appropriately narrow.
But here is the first place my own compression leaks, and I would rather mark it than hide it. "It's a placebo" still grants too much. It quietly attributes the recovery to placebo, when the larger and duller truth is that most "it worked for me" is neither the remedy nor placebo. It is regression to the mean and the natural course of a self-limiting illness. People reach for a remedy when symptoms are at their worst; symptoms then drift back toward baseline on their own, and whatever was taken at the trough collects the credit. The honest correction is not "the remedy didn't work, but your mind did." It is "you were going to recover, and the remedy, and quite possibly the placebo, were both along for the ride." My corrected shorthand, "it's a placebo," is higher fidelity than "it works" because it strips the false chemistry, and it is still lossy in the same direction, tilting credit toward an effect that was probably minor. I will come back to this, because it is the whole point.
One study is worth naming, because it draws the line cleanly between feeling better and being better. Wechsler and colleagues gave asthma patients a real bronchodilator, a placebo inhaler, sham acupuncture, or nothing. On what the patients reported, the placebo inhaler and the sham acupuncture performed as well as the real drug, all three far better than no treatment. On the objective measure, lung function, only the real drug did anything. The placebo edited the dashboard and left the engine untouched. That is the shape of nearly every homeopathic success: a real change in how the symptom is experienced, sitting on top of a body that is healing, or not, for its own reasons.
Cold: a real contributor mistaken for the cause
The cold case is the mirror image, and the correction runs the other way: not amputation but re-weighting.
The load-bearing fact is that the virus is necessary. No virus, no cold; the air conditioning and the cold drink cannot make you ill on their own, because there is nothing there to make you ill. The popular shorthand puts the cause in the wrong place, in the temperature, and a clean version of the correction simply restores the necessary cause to its position.
But the clean correction over-corrects, and this is where the standard sceptical reply, "it's only viruses, the cold is a myth," is itself too strong. Cold is a real contributing factor. A 2022 study by Huang and colleagues gives a direct mechanism: the cells at the front of the nose release swarms of tiny vesicles that act as antiviral decoys, and cooling the nose, which they induced by exposing subjects to cold air for fifteen minutes and dropping nasal temperature a few degrees, cut that defence by roughly forty percent. The same virus clears a colder nose less well. So the precise shape is that the virus is necessary and cold genuinely raises the odds that exposure becomes infection. "Cold contributes" is true; "cold causes" is not. The person I spoke to had already half-inferred this, which is why the correction was light. I was adjusting a weight, not removing a belief.
That difference is the substance of this case, and it is not only a difference in content. Amputating a false belief asks the person to give up something whole: the remedy did nothing, full stop. Re-weighting a partly-true one asks only that they hold what they already had at a different magnitude, keeping the intuition that cold matters and demoting it from cause to contributor. The first meets resistance because something is taken away; the second often does not, because nothing is removed, only resized. That the person had already half-arrived at the correction is the signature of a belief that was mostly right and only mis-weighted, and such a belief does much of the work of its own repair, if you let it keep the kernel. The homeopathy case was harder for the opposite reason: there was no kernel to keep, so the whole attribution had to go.
The symptom half, which I gave them too, holds the same lesson at smaller scale. Once a throat is sore, cold genuinely helps; it numbs the inflamed tissue, and ice cream is not folly. Warm fluids and steam give real but modest and largely subjective relief when there is congestion. Neither touches the viral course; both change how the illness feels. I will not oversell these, the evidence for steam in particular being thinner than the folk confidence in it, but the structure is the same as the asthma study: comfort is real and is not the same as cure.
And even this is compressed: the full causal picture of why colds cluster in winter folds in indoor crowding, humidity, behaviour, and more, and my corrected version discards most of it.
What the handling actually did
What the handling did reduces to five separations, made across the two conversations; they are the whole of the skill, such as it is. Experience from explanation: they did get better, but the remedy is not why. Cause from contributor: the virus makes you ill, the cold only tilts the odds. Symptom relief from disease mechanism: cold soothes the throat, it does not shorten the cold. Private comfort from transmissible risk: what a belief does for the believer is not what it does once it travels. And the epistemic claim from the compassionate one: what is true, kept apart from how hard I am willing to press it. None of these is clever. All of them are just refusing to let a real thing and a false thing ride in the same sentence.
The worry, and the thing that outweighs it
There was a worry under all of this, and it was real. If the belief was doing some placebo work, then correcting it might cost the person something, might, in the limit, leave them slightly worse. I want to take that worry seriously and then size it correctly, because feeling the weight of altering someone's model is exactly what keeps this from being point-scoring, and mis-sizing it is its own failure.
Sized correctly, it is small. The benefit I could remove is the placebo slice, and by the earlier argument that slice is already minor next to regression to the mean and natural healing, a small part of an already small part. And what I did was not delete the benefit; I affirmed the recovery and corrected only the story attached to it, which is closer to leaving the comfort intact than to taking it away. The worry is sound as a sensitivity and overgrown as a probability of harm.
Set against it is the thing that actually decides the case, and it only appears once the horizon extends past this one person and this one mild illness. A belief is not a private object. It gets recommended. The next person it reaches may not have a self-limiting cold; they may have something that needs a real drug, and a confident "this worked for me" may be what stands between them and getting it, a danger the NHMRC names directly when it warns against using homeopathy for conditions that are serious or could become so. The harm of my correction is bounded and small. The harm of leaving the belief intact is unbounded in the tail. On the short horizon and the typical case, correcting a placebo-bearing belief looks faintly cruel; on the full distribution it is clearly right.
One objection remains, and it is fair: beliefs do more than track truth. They steady mood, hold identity together, let people coordinate, and a false belief can serve those ends well. I am not dismissing that. But two things bound it here. The comfort a false model gives is borrowed rather than free, because a model that misfits reality pays for the misfit where prediction meets the world, sooner or later, and the bill is often handed to someone else. And the function at stake in these particular cases is not identity or coordination but the one fidelity governs, getting the cause right, because the cost of getting it wrong is the next person reaching for sugar instead of medicine. Where a belief's other work is doing no harm, I leave it alone. Here it was not.
I put it to myself bluntly at the time, and I will keep the blunt form because the smoothed version loses the structure. The worry about the person losing the small comfort of feeling better is heavily outweighed, probabilistically, by the good of intervening, and I have sufficient tooling to make the call. The tooling is real for the diagnosis. It is, as above, still imperfect for the delivery.
The carve-out I had to correct in myself
I told that first person something else, and it is the part I have had to inspect most, because it was my error, not theirs. I put it on their heart not to recommend homeopathy to anyone, except, I said, as a last resort, once someone has exhausted real medical options; and there, who am I to say otherwise.
That carve-out is generous in exactly the way the rest of the day was about catching. It feels like humility. On inspection it is a lossy compression of my own. The danger was never the inert pill as such; it was substitution and transmission, and the last-resort case escapes neither. "After medicine has failed" selects for the most serious, refractory conditions, precisely where the cost of a sugar pill is highest rather than lowest, in money, in false hope, in palliative care foregone. And "a reasonable last resort" is not a stable thing to pass between people. It sheds its conditions in transmission and arrives at the next person as "it works," the same lossy squeeze I had just spent the day correcting. I had released a new belief, caveats attached, into the same machine that strips caveats.
There is also a quieter slip in "who am I to say." I am, in fact, someone to say; that was the entire premise of intervening at all. The deference is selective. I did not defer on the mechanism, I corrected it. Deferring only at the last-resort boundary is not humility, it is the conceding reflex finding the one spot it is still allowed to operate. The honest version separates the two claims it had fused. The epistemic claim, that the pill is inert everywhere, the refractory case included, stays fixed and true. The compassionate claim, that I will not press a suffering person who is out of options and wants the ritual, is a decision about my own conduct, and a defensible one. What I had done was let the kindness quietly rewrite the fact, granting that maybe the pill works there when all I meant was that I would not be the one to take it away. Those come apart. Keeping them apart is the same firewall as everything else: the fact does not move, and what flexes is what I do around it.
The same loss, in my own corrections
So here is the shape of the day, and it is not the shape I would have written before living it. I did not spend it finding errors in other people and dispatching them. I spent it watching beliefs lose their caveats as they compressed, intervening where the loss was load-bearing, and then finding the same loss in my own corrections. "It's a placebo" tilted credit toward placebo. My picture of winter colds discarded most of the causes. My last-resort carve-out shed its conditions the moment I imagined it travelling. Every correction I made was itself a compression, truer than what it replaced and still lossy, and there is no final version underneath them where the loss goes to zero. Fidelity is not a place you arrive. It is a direction you keep moving in, and the most you can do is leave less out than the thing you are correcting did.
That is the whole of it. Two people said two false things on one day. One belief needed its cause amputated, the other needed its weight corrected. I did both imperfectly, and the useful part was not being right about them. It was turning the same instrument on what I had just said and finding it, too, leaving things out. A field note, not a verdict. The errors were ordinary. So were mine.
Sources
verified · National Health and Medical Research Council (Australia). Information Paper: Evidence on the effectiveness of homeopathy for treating health conditions. NHMRC, 2015. nhmrc.gov.au/about-us/publications/homeopathy
verified · Wechsler ME, Kelley JM, Boyd IOE, Dutile S, Marigowda G, Kirsch I, Israel E, Kaptchuk TJ. Active albuterol or placebo, sham acupuncture, or no intervention in asthma. New England Journal of Medicine 2011;365(2):119–126. DOI 10.1056/NEJMoa1103319
verified · Huang D, Taha MS, Nocera AL, Workman AD, Amiji MM, Bleier BS. Cold exposure impairs extracellular vesicle swarm–mediated nasal antiviral immunity. Journal of Allergy and Clinical Immunology 2022. PubMed 36494212
Comments
Post a Comment