Asbestos took 49 years. PFAS took 23. The next one lands in your tenure
· 13 min read
This is article 5 of 8 in the Data Centre series. Here is article 4.
Years ago I stood in a boardroom with a management team anyone who has worked in semiconductors would recognize. Sharp, numerate, allergic to nonsense, the kind of people who could smell a bad model from across the room.
A single slide went up showing a cost curve. Not a scary curve. A patient one. It rose so gently across the first ten years that you had to squint to catch it moving at all, and then in the final third of the chart it went almost vertical.
A dozen very smart people looked at that slide and registered a hockey stick, which is to say they registered nothing at all. Hockey sticks are wallpaper in that industry. It is the shape ambitious plans are supposed to take.
I was in the room and I registered nothing either.
The problem was not the chart. The problem was the eyes reading it, mine included, and I have thought about that meeting for a long time.
Everyone at that table was operating on a rule that had served them for decades: trust the readings. Check the number, compare it to last quarter’s number, and if the difference is inside the noise, the thing is stable and your attention belongs somewhere else. That rule is not laziness. It is the entire basis of process control, it is why those people were good at their jobs, and applied to a chip fab it is close to infallible.
Applied to a curve in the early part of a doubling, it is a machine for missing everything that matters. The readings really are inside the noise. They stay inside the noise right up until the quarter they are not, and by then the decision that mattered was four years behind you.
Nobody in that room was careless, and nobody was short-term in any morally deficient sense you could fix with a stronger conscience. They were running factory-standard human cognition, unmodified, against a threat our brain was never built to detect.
There is a story, old and murky in origin, usually placed in an Indian court, about a man who understood this and an emperor who did not.
A man invents chess, and a grateful emperor offers him any reward. He asks for rice. One grain on the first square, two on the second, four on the third, doubling across all sixty-four.
The emperor, hearing a modest request, grants it on the spot.
It took his treasurer a while to come back with the actual number, and it was not modest. Something on the order of eighteen quintillion grains, which at the weight of an average grain works out to roughly four hundred and sixty billion tons. More rice than the world now grows in several centuries.
The emperor was not foolish. He was thinking linearly, the way every one of us thinks by default, while the man asking for rice understood something about doubling that the emperor’s whole reign had never required him to learn.
The board is worth walking one square at a time, because the emperor’s error is not where anyone expects it.
For the first half, doubling looks like nothing at all. Square ten carries a little over five hundred grains, barely a spoonful. Square twenty carries a bit over half a million, a sack you could lift with one hand. Square thirty carries better than half a billion grains, which sounds enormous right up until you weigh it and find about thirteen tons. One truckload. Still a rounding error against a season’s harvest.
Anyone watching through square thirty-two, the halfway point, would tell you, correctly, based on everything they had observed, that this is a manageable and gradually growing obligation.
Then the second half starts, and the same doubling that produced a few silos produces a surplus larger than nations, then larger than continents, then larger than the planet’s entire annual output, in thirty-two more identical steps.
Nothing changed about the rule. The rule was one grain, doubled, every square, from square one. What changed is that we finally reached the part of the curve where our eyes, our intuition, and every linear instinct we own stop keeping up with arithmetic that never once stopped being simple.
That is the trap. Not stupidity. Linear intuition applied to an exponential fact.
Albert Bartlett spent a career at the University of Colorado trying to get this into people’s heads, and he put it as bluntly as it can be put: “The greatest shortcoming of the human race is man’s inability to understand the exponential function.”
He was not being clever. He was reporting a measurement, and he reported it before the smartphone, before the spreadsheet on every desk, before any of the machinery that was supposed to fix exactly this.
Half a century on file, and it still does not change how a room reads a chart.
An exponential curve spends most of its life looking flat. Not flat like nothing is happening. Flat like something is happening so slowly that every individual measurement looks like noise, like rounding error, like next quarter’s problem.
You check it, it is basically the same as last time, so you file it under stable and move to the next agenda item.
The curve does not warn you before it bends. The bend is the warning, and by the time we can make it out with the naked eye, we are no longer early.
Hemingway wrote the whole thing in six words, in a scene with nothing to do with business. A character in The Sun Also Rises is asked how he went bankrupt. “Two ways,” he says. “Gradually and then suddenly.”
That is the exact shape of every deferred liability that has ever landed. Nobody wakes up one Tuesday and discovers a compatibility problem with the living systems their operation depends on. It accrues in increments too small to trip any alarm we have built, for years, sometimes for a working lifetime, and then it stops accruing. Nature’s ledger becomes gradually visible, and the Third Number arrives suddenly.
Gradually is the only part of the process anyone in the room experiences. Suddenly is the part that makes the news.
There is a second mechanism underneath the first, and it is stranger.
Hal Hershfield, working with Brian Knutson’s group at Stanford, put people in an fMRI and asked them to think about themselves. Then he asked them to think about themselves in the future. The neural activity for the future self looked less like self-referential thinking and more like thinking about someone else entirely. They published it in Social Cognitive and Affective Neuroscience in 2009.
Not a version of you. A different person. Someone you do not know well and, it turns out, do not feel much urgency about protecting.
Popular retellings dull the point they are trying to make. The size of the gap, between how brightly a person’s brain lit up for their present self and how dimly for their future self, predicted how steeply that same person discounted a future reward.
The less like you your future self looked to your own brain, the more cheaply you sold them out.
That is not a metaphor for corporate short-termism. That is the machinery. And it is not a discount, a dollar next year weighted a little less than a dollar today, the way any sensible model does it. It is a substitution. Your brain actually swapped the future person that is you for a stranger it feels no loyalty toward and no longer thinks of as you.
You would think twice before handing a stranger your liabilities. Your brain does it automatically every time it trades present convenience for future cost. Which is a subtle driver of the economy and the quest for wealth, which is nothing more than grasping for a protected state of zero inconvenience in any form in one’s life.
Organizations have their own version, and theirs is structural rather than psychological, which makes it worse. A person can be shown the fMRI and change their behavior. An org chart cannot be shown anything.
Call it management mean time between failures (M-MTBF). We borrow the term from reliability engineering, where you design around the expected lifespan of a component before it breaks, and you build your maintenance schedule and your spares inventory around that number rather than around some idealized version of how long the part should last.
Apply the same discipline to leadership tenure and the picture is uncomfortable. The average run of a senior leader in a seat is measured in a small number of years. Call it three, call it five. The exact number matters less than the fact that it is short relative to almost anything biological or ecological time horizons that a business depends on. And it is certainly not a generational term.
Whatever bill comes due outside that window gets discounted toward zero in every practical calculation that leader makes. Not because they are cynical. Because the incentive architecture underneath them was built that way. The board is measuring this cycle. The compensation plan is measuring this cycle. The analysts on the earnings call are measuring this cycle.
The bill does not disappear. It gets mailed to someone else’s desk, and that person walks in with their own three-to-five-year clock running the same discount function.
Run that relay long enough and you get an organization that has never, at any point in its history, contained one person structurally positioned to own the tail, where the Third Number often resides. Every individual owner was rational. The sum of all that rationality is an institution that manufactures blindness as a byproduct of good governance, the same way a factory manufactures scrap as a byproduct of good throughput.
Nobody ordered the scrap. It is what the process produces if you do not design and keep account against it.
Here is what has actually changed, and it has nothing to do with anyone caring more than they did last year.
In 1924, a thirty-three-year-old textile worker named Nellie Kershaw died, and Dr. W. E. Cooke published the case, introducing the word asbestosis into the medical literature. That is the moment the hazard became public knowledge.
Johns-Manville received internal medical reports confirming disease in its own workers in 1933. The Sumner Simpson correspondence of 1935 documents a coordinated strategy to suppress research. A confidential 1942 report found that forty-two percent of workers in Johns-Manville factories had developed asbestosis. Wagner established the causal link to mesothelioma in 1960. OSHA regulated it in 1971.
And in 1973 the Fifth Circuit upheld the verdict in Borel v. Fibreboard Paper Products, the first successful asbestos personal injury case, awarding Clarence Borel’s family 79,436 dollars and establishing strict liability.
Borel had died before the decision came down.
From 1924 to 1973. Forty-nine years from public knowledge to the first check.
Now run the same measurement on two more, the same way, from the point the hazard became public knowledge to the point somebody wrote a check.
PFAS. 3M began mass-manufacturing PFOA in 1947. Internal toxicity findings accumulate from 1961. Public surfacing arrives between 1998 and 2000, when Bilott takes the Tennant case and the compounds turn up in blood banks nationwide. Attribution lands in 2012 with the C8 Science Panel. Financial pricing (aka the Third Number quantified) arrives on 22 June 2023, when 3M settles public water system claims at 10.3 billion dollars present value.
Twenty-three years.
Microplastics. Richard Thompson and colleagues published “Lost at sea: Where is all the plastic?” in Science on 7 May 2004, coining the term. Between 2018 and 2023, a review of the Lexis and Westlaw dockets by the American Bar Association’s products liability section found exactly one microplastics case filed. Between January 2024 and December 2025 it found at least eighteen in federal court and five more in state courts.
About twenty years, and the docket went from one case in six years to twenty-three in two.
Forty-nine. Twenty-three. Twenty.
A capital decision with a twenty-five-year asset life, approved by somebody with maybe twenty years of career remaining, has a window of about twenty years before the bill lands on a successor.
Asbestos at forty-nine crossed two careers. The people who signed for the production were dead or retired before the first verdict came down.
PFAS at twenty-three missed the window by three years, which is about the length of one promotion cycle.
Microplastics at twenty is inside it.
The relay has been breaking for a century, one interval at a time, and it has now broken. The person who signs for the material is increasingly the person who will be asked about it.
And three things are closing the gap, none of which is a change of heart. Detection: biomonitoring at parts per trillion that once needed specialized laboratories and years is becoming routine, and this is why biosensing is in the critical path for nearly all new technologies. Attribution: the C8 panel needed eight years to reach probable linkage, and today epidemiology combined with computational toxicology doesn’t need eight years. Pricing: insurers repriced PFAS exposure across general liability, business owner and umbrella forms in 2024 and 2025, because carriers do not wait for consensus. They wait for exposure.
The magnitudes in all three of those arcs are argued over; however, the dates are not.
You cannot out-willpower a perceptual system. None of us can decide, one morning, to feel an exponential curve the way we feel a car swerving into our lane. It takes retraining, deliberately, against its own grain, for as long as you hold the job.
Two questions do most of the work.
Before you file any curve under stable, ask what it looks like doubled twice more.
Before you approve any plan that clears on someone else’s watch, ask what it would look like with your own name still on the door the day it comes due. You need to spend time learning to make generational decisions by always consulting nature’s ledger and finding the Third Number.
Because the account always clears. Everything here narrows to one discipline: making certain it clears while your name is still on the door, and not a chair or two down the line, when all your successor can do is sign the check you left them to fill out and wonder why nobody saw it coming. And those checks will keep getting bigger if we don’t pay attention to the Third Numbers accruing gradually but presenting to the balance sheet suddenly.
Three arcs is three arcs, and I would rather have thirty. The BiosVerse-Compatibility Index (BCI) scores whether a design fits. The Generational Cost of Ownership (GCOO) model prices what the fit or the misfit costs across generations. Both exist and both run today. What is not settled is how they are governed: who holds the standard, what counts as evidence, and who has the standing to fail a project. That is the work of Biomimetics26, on November 9 and 10 at the Minnesota Landscape Arboretum, and it is open. Bring a dated arc of your own, or bring the one above that you think we measured wrong. Both make the instrument better, and the second one makes it better faster.
Next: the most exhaustively stress-tested design archive in existence has been sitting in the public domain, unpriced, the entire time. What changed is that we can finally read it.
illuminem Voices is a democratic space presenting the opinions of leading Sustainability Thought Leaders, their views do not necessarily represent those of illuminem.
The world needs sustainability knowledge. At illuminem, no interest group or shareholder can influence our work. Thank you for supporting our mission to make high-quality and independent sustainability information free for all. Every contribution helps. Thank you for donating today.
Michael Wright
Pollution · Corporate Governance
illuminem briefings
Pollution · Climate Change
Michael Wright
Petrochemicals · Pollution
CNN
Pollution · Climate Change
BBC
Pollution · Nature
BBC
Pollution · Climate Change