Just under ten weeks ago, 59-year-old Mike Lynch, who died Monday after the sinking of his super-yacht Bayesian off the coast of Sicily, was on trial in San Francisco on 17 charges of fraud. He was almost guaranteed to receive a 25-year sentence.
He was terrified that he would die in a U.S. prison, not because he was guilty—he had spent 13 years and roughly $35 million on legal fees arguing his innocence—but because it’s almost unheard of in the U.S. to win a case against the American Justice Department. He was nonetheless acquitted. His chances of winning the case were put at 0.5%.
Safely back in Britain, Lynch set about celebrating what he called his “second life.” Through tears, he told one interviewer how even the traffic in London seemed miraculous. “I’m just thinking this is the most beautiful thing I have ever seen,” he said.
This month, he took his family, friends, and lawyers on a celebratory cruise around the Mediterranean aboard his 184-foot sailboat. It sank in a violent squall in the small hours of Monday, killing Lynch, his 18-year-old daughter, and five other passengers.
The chance of such a yacht being knocked flat by a weather event was minuscule because it had simply never happened. Sailing expert Stewart Campbell, editor-in-chief of Boat International, said on Newsnight, “I’ve been speaking to a lot in the industry today and they are as shocked as me—enough to disbelieve that this could happen.”
As if beating infinitesimal odds twice in a few weeks wasn’t freakish enough, it has since emerged that Lynch’s co-defendant in the U.S. trial, his company’s former vice-president of finance, Stephen Chamberlain, was killed by a car on Saturday while out running in Cambridgeshire. The chance of both defendants dying within weeks of the trial are beyond absurd.
The ultimate irony, however, is that the Essex-born academic, tech innovator, prime ministerial advisor, and general member of the great and good is also one of the world’s leading authorities on probability theory. (Were Lynch’s story not true, anyone who went to Netflix to pitch a dramatized version of it would probably be shown the door for being ludicrously improbable.)
Specifically, Mike Lynch is a proponent and teacher at his alma mater Cambridge University of Bayesian mathematics, the theory of Thomas Bayes, an 18th century statistician and Presbyterian minister. Bayesian math embraces the idea that predictions of unusual outcomes—freak weather events in particular—can’t be made accurately using a conventional fixed framework of information, but have to be constantly updated with fresh data on how a situation develops. Bayesian math is also the basis of the current boom in artificial intelligence, in which Mike Lynch was a pioneer.
If Lynch sounds to many like a major-league nerd (and possibly lucky not to be serving time in prison), I can vouch for a very different Mike Lynch. I’ve been in touch with him since 2016 and found him not just alarmingly honest and straightforward, but very amusing, extraordinarily erudite in everything from technology to classical cultures, and unexpectedly unorthodox in his views on science.
Which is how I came to be in his Pall Mall office eight years ago for a sparkling one-on-one meeting-cum-tutorial—not on the subject of technology or his already mounting legal problems, but dogs.
Mike Lynch had contacted me about a column I wrote for many years in the Financial Times. He knew that I was also interested in fringe science—I once wrote a book on Uri Geller, the Israeli-British illusionist—and that we’d gone to the same Essex school, though we hadn’t crossed paths. He wanted me to help him set up a research project on dogs’ apparent sixth sense.
Lynch explained that he was convinced dogs have a hitherto unexplained sixth sense—that they really do know, for instance, without conventional sensual inputs, that you’re on your way home, even if you’re miles away.
“I’ve done this informal experiment many times,” he said. “I’ll get on a train in London for our house in Suffolk and at that exact second, they [the dogs] will get excited. My wife and I have ruled out as many possible clues as we can. She’ll note the moment they start barking and it will match up with me stepping onto the train.”
The problem was, he went on to explain, that the experimental work that had been done in this area was flawed. For his next big project, he wanted to fund some proper, rigorous, incontestable research.
“I’m not a parapsychologist or a paranormalist,” he told me. “I don’t believe in auras or energies or anything weird. I’m firmly entrenched in science, and I know this is controversial, but there’s an effect out there that needs explaining.”
As a first step, he wanted me to write a column inviting people to describe their experiences with dogs’ sixth sense, and then help him launch a dedicated Web site for them to post their accounts. “That way we build a database of supposedly psychic dogs to work with.”
He was fascinating as he expounded his theories further for nearly 90 minutes. He was an inspiring teacher and patient when I lagged behind, seemed a bit thick, or got the wrong end of the stick.
“Dogs are incredibly good at reading people’s emotions,” he said. “They can pick up very subtle cues, which, of course, we do as humans. We just don’t realize we do it. The key thing is that dogs are unusual among creatures in that they have mirror neurons, which are very important. These are the neurons in the brain that put you in someone else’s shoes. If I hit my hand with a hammer, you’ll wince. It’s the basis of empathy.”
“Now interestingly, this makes it possible for children around two and a half to learn to lie. Deceit is fascinating because to be deceitful, you have to realize that not everyone has the same information you have. So it’s a major step and a highly intelligent function. So dogs run a model in their mind of what it’s like to be you, or another dog. This is how they become pack animals.”
This, he admitted, was all traditional science. The big question he wanted to know the answer to was whether there was a realm beyond that. “You have to have a good understanding above all of probability—the probability that this or that effect is explicable using known science.
“I’m totally a believer in the scientific method but one has to bear in mind that the mammalian brain is the most complex item in the known universe … so the idea that we would understand everything about it sounds arrogant,” he insisted.
He said this way he could show—or not—that dogs are “Bayesian calculators.” Bayesian logic is kind of magical thinking in a way because it’s so much more subtle and complex than classical logic. “What dogs do in their Bayesian way is use intuition—see data in a different way. But this is not magic, it’s science and it’s going to lead a few years from now to astonishing developments in AI.”
This leap to AI eight years ago from the dry theories of a long-dead clergyman wasn’t surprising. Lynch had the type of brain that was always ahead of its time.
Regarded by many in both academia and business as a tech genius, it’s why in 2011 his technology company, Autonomy, sold for $11.7 billion to the fading computer giant HP. It was HP who seemed to have suffered buyer’s remorse later and started blaming Lynch and his team—incorrectly—for fiddling the books.
For a hard-headed academic his view of intuition made sense in terms of data processing and how we predict things happening. He was adamant that “hunches, gut feeling, intuition” were all sophisticated measures by which decisions are taken, “not boring spreadsheets. This is why big companies make such bad decisions. The danger of analysis drives me nuts but I’ve had to learn this.”
He recounted a story from the past when a “posh dreamer” came to him with a great idea to download music over the Internet. Lynch said he spent a long time explaining to him why his vision wouldn’t work. That it would take four days to download a single over the Internet and the disc to store it on would cost the same as the average house.
In retrospect, he said: “What I should have done is said all of that and then gone away and plotted the cost of disc drives and realized he could be onto something. But the over-analysis stopped it happening and I missed out.”
Mike Lynch’s dog project never happened. My interview with him never ran thanks to the question mark over him with the, as then, unsettled court case.
“That’s really sad,” he e-mailed me after I told him. “But thanks for trying.” He seemed truly upset.
We stayed in touch throughout the trial, and were still discussing the Ultimate Dogs Experiment while he was in San Francisco in the run-up to the trial and the court case itself. As the judge proved, I wasn’t the only person to find Lynch convincing and exuding both character and integrity.
I discussed the whole, heartrending story yesterday with a medical consultant friend from school who had been following his travails. We agreed that this latest tragic twist in his incredible story feels as improbable as it does just desperately sad.
Bayesian math is used extensively in forecasting probable weather, so what are the odds of a boat named Bayesian being capsized by a tornado/waterspout in the Mediterranean?
“Smart arse question, sorry,” my friend wrote back. “But with all the money in the world, why does an expert on predicting natural events sleep with friends and family on a boat when the weather is so awful?”
Unfortunately, we will probably never know.
Based in London and New York, AIR MAIL’s tech columnist, Jonathan Margolis, spent more than two decades as a technology writer at the Financial Times. He is also the author of A Brief History of Tomorrow, a book on the history of futurology
