A study landed this week in which researchers examined 28,000 patients on GLP-1 weight-loss drugs — a type of drug that affects appetite hormones — and found that variations in two genes explain why the drugs work brilliantly for some people and barely work for others. The Guardian, a major British newspaper known for shaping how scientific findings are understood by the public, reported this as a breakthrough. I read the headline three times. Not because I didn’t understand it. Because the spatial logic of the sentence was so perfectly inverted that I wanted to admire the architecture.
“Genetics may help explain why results from weight-loss jabs vary.” The subject of that sentence is the drug. The object is the body. The body is the thing that varies. The drug is the thing that stays fixed. Twenty-eight thousand bodies measured against one pharmaceutical standard, and the conclusion drawn is not “the intervention fits some bodies and fails others” but “we can now predict which bodies will fit the intervention.” The frame of reference never moves.
I spend my working life designing information systems. Wayfinding, signage, visual communication. In February 2024, a hospital in Rotterdam hired me to redesign their outpatient navigation. The existing system used colour-coded lines painted on the floor. Blue for cardiology. Green for oncology. Red for emergency. The lines worked perfectly for anyone who could see colour, read English, and walk at a pace that matched the spacing of the directional signs. For everyone else, the lines were decoration.
The hospital administrator who briefed me said something I wrote down in my notebook: “The system works. The problem is that some patients can’t follow it.” She wasn’t being cruel. She genuinely believed the system was correct and the patients were the variable.
That is the same sentence as “genetics may help explain why results from weight-loss jabs vary.”
Otto Neurath, an Austrian social scientist and designer working in the 1930s, built what he called an isotype system — a method of communication using simple, standardized icons instead of numbers or words. The idea was universal communication: cross every language barrier, reach every literacy level. It was brilliant and it failed. Not because the icons were bad. Because Neurath assumed a universal viewer. The icons worked for sighted, literate Europeans who already understood statistical thinking. The system was designed from inside one room and called universal.
I admire Neurath enormously. The failure is instructive.
William Stokoe, a linguist, published his analysis of sign language structure in 1960, proving it was a complete linguistic system with its own grammar, not broken English performed with hands. This work was revolutionary because for decades before Stokoe, researchers had measured sign against speech and found it lacking. The frame of reference was spoken language. Sign was the deviation. Stokoe didn’t prove sign was “as good as” speech. He proved the comparison itself was the error.
The GLP-1 study does something structurally identical to pre-Stokoe linguistics. It takes a single intervention, treats it as the standard, and then asks why bodies deviate from its expected effect. The researchers — and I have no quarrel with their data — looked at variations in two genes that control how the body processes appetite signals and regulates hunger. They found that certain people metabolize the drug differently, experience different appetite responses, different side effects. Real findings. Useful findings. But the question they asked was: how can we predict who will respond correctly? Not: why are we treating a single drug pathway as the correct response in the first place?
Cathryn Lewis, a geneticist at King’s College London, told the Guardian that understanding these variations could help “tailor treatments” — a position that matters to this argument because it represents how even well-intentioned genetic research frames the solution. Tailoring is better than ignoring. I’ll give it that. But tailoring still assumes the garment is right and the body is the problem. The most expensive bespoke suit in the world still starts from the premise that the client’s body is an irregularity to be managed.
In November 2021, I sat in a design review for a transit app in Brussels. A product manager presented a wayfinding feature that gave step-by-step directions. The test users had followed the directions correctly and still arrived at the wrong platform, because the directions referenced a landmark — a coffee kiosk — that had closed six months earlier. The product manager’s proposed fix: update the landmark. My proposed fix: stop anchoring navigation to things that disappear. Build a system that works when the environment changes, not one that requires the environment to hold still.
The GLP-1 research proposes, essentially, to update the landmark. Identify the genetic variations, screen patients, match them to the right drug or the right dose. A pharmacogenomic GPS. It will help some people. It will also deepen the assumption that the drug is the map and the body is the territory that needs to be corrected to match it.
Fat disabled people have been saying this for years, plainly: the body is not the error. Samantha Murray, a scholar at Macquarie University in Sydney, wrote in 2008 that obesity research consistently treats the fat body as a problem that the correct intervention will resolve, rather than asking whether the framework that defines fatness as pathology is itself the variable worth examining. The GLP-1 study is eighteen years later and the frame hasn’t shifted a millimetre.
I keep returning to that hospital in Rotterdam. After the redesign, the new system used tactile floor markers, high-contrast icons with no colour dependency, and directional cues spaced for varying walking speeds. The administrator reviewed it and said: “This is very thorough, but it’s a lot of infrastructure for a small number of patients.” I told her to count how many people over seventy used the outpatient clinic. She went quiet.
The “small number of patients” the system failed was, in fact, most of them. The original system worked for a narrow band of users who happened to match its assumptions. Everyone else adapted, asked for help, or got lost. The system never registered those failures because it wasn’t designed to measure them. It measured arrivals, not the people who turned around.
Twenty-eight thousand patients. The ones who “responded well” to the drug fit its mechanism. The ones who didn’t were measured as deviation. Nobody counted what it cost those 28,000 bodies to be the territory the map was drawn over.
The lines on the floor in Rotterdam are gone now. The new system works. The administrator still calls it “the accessible version,” as if the original was the real one.
This article was prompted by Genetics may help explain why results from weight-loss jabs vary, say scientists from Guardian Society.