Lawrence O’Dwyer

Piedra de Sol

The workshop is a cluttered, busy space. There are clamps and drills, chisels and burins; the tools might be those of a cobbler or a stonemason. Drawers are stuffed with wires and molds. A cast-iron disc is lost in a blur of speed. There is a microscope on a bench behind me. A middle-aged man in a white coat is talking excitedly. Between index finger and thumb he’s holding a blood-red stone as he tells me that it’s the high percentage of chromium that makes it burn so brightly. The man I’m talking to is a master craftsman, a main d’or. It’s a ruby that he’s holding in his hand.
The technique that he’s explaining is the mystery setting, the serti mysterieux. It involves setting precious stones side by side in such a way that the underlying metal cannot be seen. For a bracelet of rubies, the result is a corona of light that has a lustrous, velvet shine. Patented in 1933, the mystery setting is a secret that’s passed on from master to apprentice in the workshop of Van Cleef & Arpels. The work begins with sketches and drawings, followed by models and mock-ups that establish volume. The the “rails” are prepared, thin wires made of gold or platinum that hold the stones in place.
At the beginning of the twentieth century, a wooden lapidary stick tipped with heated wax was used to hold the stones against the rotating disc. Today, the stick is made of acrylic resin, but little else has changed. Tools are handed down from master to apprentice; hammers are engraved with the names of mains d’or who are no longer alive though their tools are still in use.
Having worked all his life with precious stones, he has a fine appreciation of chemistry and mineralogy, but what he’s trying to achieve is a psychological effect, and his most important tool is a cast-iron disc. Sprinkled with water and diamond dust thickened with oil, it spins at 3,600 revolutions per minute. The noise a stone makes when laid on the grinding wheel gives him a sense of position and inclination; coordinates are aural as well as visual and tactile. Each stone has a different feel—sapphires are softer than rubies; diamonds are harder but more brittle than rubies. Each gem is carefully selected. Only stones classified at the top end of the scale of the Gemological Institute of America (GIA) are used; stones with a grade of D or E—the scale extends all the way to Z. A ranking of flawlessness is also used. He may work with stones that are designated IF (internally flawless). Rarer still but not unknown in this workshop are stones that are free from both internal and external blemishes, which are designated FL—perfectly flawless stones.
After cutting and faceting, tracks are carved into the gem that are used to slide it into the gold bezel. The secret within the secret is “the door,” a removable part through which the cutter inserts the stones; it is then closed and soldered shut. The door is the key to the architecture of the entire piece.
I am asked to praise the jewels but see the hands instead—the dirt under the fingernails. The craftsman has dedicated his life to the serti mysterieux, and it is this obsessiveness that delights the mind as much as the eye. The workshop is just a few blocks from the flagship boutique of Van Cleef & Arpels, on the Place Vendôme. We are in the center of Paris.

The French take jewelry seriously. To begin with, they have two words where English has one: bijouterie, for precious metals; joaillerie, for stones with a minimum of metalwork. But the history of the maison—it is not a company—begins in the lowlands with two Jewish families, one from Amsterdam, the other from Ghent. In 1864, Léon Salomon Arpels moves to Paris, where he becomes a jewelry merchant. He is joined in 1867 by Salomon Van Cleef, a specialist in the art of cutting stones. His son, Alfred, is apprenticed for six years to the lapidist David & Grosgogeat before he establishes himself as a maker of precious stones. In 1897, Alfred marries his cousin Estelle Arpels.
If the folktale of Van Cleef & Arpels is a love story, its language is Dutch. A seagoing nation with a kingdom as small as a ship in a bottle, not fond of loose or ragged talk, the Dutch do not say “I love you”; instead they say “Ik hou van jou,” literally “I hold to you”—a maritime expression full of anchors and stays. A curious blend—the French and their extravagant poets; the Dutch and their sober merchants. It is that mortar and pestle of high and low—the dirt under the fingernails, the fire of the ruby—that’s at the heart of the workshop. Amid the garage-like clutter, with his microscope and cast-iron disc, the main d’or reminds me of a scientist. The workshop looks very much like a lab I used to work in—a patch- clamp lab.

Patch clamping is a technique that revolutionized neuroscience. With a glass pipette pressed gently against the membrane of a neuron, it lets you listen to the fine Morse code of the brain. The technique earned its inventors, Neher and Sakmann, a Nobel Prize in 1991. A long apprenticeship is needed before you can record the microscopic currents that flow through the brain. It is an art as much as a science; from the “pulling” of glass pipettes to the building of a patch-clamp rig, everything is artisanal.
I worked in a lab in the Department of Physiology at Trinity College Dublin, where shafts of sunlight struggled to pass through the great bay windows; the glass was grimed with the fumes of the city. On the shelves were tinctures and chipped cups, brushes and boxes, jars of bolts and screws. The rig was attached to an inner wall, an outer wall being more susceptible to vibration. It sat on an anti-vibration table. Every aspect of the work was sensitive to movement.
I would peer through a microscope, inspecting a slice of brain that was resting inside a little Perspex bath that was perfused with a constant flow of bubbling liquid. The slice was the size of the nail on your little finger. Lowering the pipette was like lowering a needle onto a record, only the needle was made of glass.
The rig was enclosed inside a Faraday cage—a wire-mesh box, open on one side, to allow access to the slice. Without the cage, the recording would be drowned in fifty hertz noise—the frequency of the mains supply. I used to disconnect the wires that earthed the rig just to watch fifty hertz waves rolling across the screen  of an oscilloscope. The green crests looked quite beautiful.
The stimulating electrode was made of Stablohm wire, an alloy of chromium and copper. The recording pipette I made myself. I would place a thin glass straw about ten centimeters long between the grooves of two metal clamps insider a “puller.” After I closed the lid and started the laser, a beam would focus light onto the middle of the pipette until it glowed red. The clamps pulled the glass apart, slowly teasing out the strands, before the final gunshot bang. Opening the lid: two perfect little spears of glass.
When I first arrived at the lab, there was a rig left over, but I was advised to take it apart and put it back together before starting any experiments. It wasn’t really advice, more an unspoken rule. It was the only way to learn how each part worked. After I’d built and rebuilt the rig, I had to learn how to prepare brain slices. The dissection took months to learn. The best handiwork with a scalpel I ever saw was that of Jianqun, a former surgeon, who asked me once to guess his fastest time removing an appendix. I can’t remember his record time, but the grace of his work was a joy to watch. The dissection needed to be fast— the tissue would die if it wasn’t transferred to a recovery bath in less than two minutes.

A delicate slice, a precious stone: as the main d’or showed me the ruby, he dropped it on the floor. I took a deep breath, somehow thinking that I was responsible for a mishap with serious consequences, that my presence was the cause of this lapse in concentration, but the way he got down on his hands and knees reminded me of how many times I’d dropped a slice of brain on the bench of the lab. After rummaging about, he got back on his feet with the ruby in his hand. He carried on, unperturbed. A brain slice is similarly robust; as long as it is wet and perfused with oxygen, it can handle a little tumble.
Another similarity between the two places was their unusual relationship with the digital world. Although patch clamping and stonecutting are manual skills, computers nevertheless contribute to the work. The electrodes in the patch clamp were connected in series, first to an analog-to-digital converter, then to an oscilloscope, and finally to a computer, where the voltages of neurons were transformed into peaks and troughs on a screen before being stored on the spinning disc of a hard drive.
After my tour of the workshop, I was brought to another room, where four designers were working on new prototypes. No garage-like clutter here, no lapidary sticks or melting wax. As I leaned over the shoulder of one designer to look at a rendering of a bracelet on a screen, he reminded me in a perfunctory yet polite tone that I should be aware of the secrecy of this work. Jewelry is a hermetic world. It is also competitive. We were just a few blocks from the Place Vendôme, the epicenter of the rarefied world of high jewelry. It is not by accident that Van Cleef & Arpels opened its doors here in 1906. By then two of the most exclusive hotels in Paris, Le Meurice and the Ritz, were already established on the Place Vendôme. Other neighbors included the rival houses of Boucheron and Lalique. Chaumet followed in 1907. Mauboussin arrived after the First World War. Just for color, Franz Mesmer and Frédéric Chopin lived there in the nineteenth century. Around the corner, on the rue de la Paix, you’ll find Cartier. With celebrity comes its corollary, secrecy. In that room full of computers, intellectual property was precious. In computer terminology, the operating system was closed.

Anyone who does four years of research ends up knowing at least one or two things about science. For my part, I know that the last page is always the first, and that neither would ever be written without the help of others.
Those were the opening words of my thesis. Research is a collaborative affair. Every thesis is a log of an apprenticeship. What I’d really learned was how to go about the art of science. There are a few simple rules that you assimilate, ethics that enter your soma. If you don’t know something, always say so. Read as much as you can. Share your knowledge. Even when you think you know something, be assured: you know almost nothing.
I studied the wiring of the brain, how memories are formed and stored. Along with memory I learned an architecture with many secret doors—orphan receptors, magnesium- gated channels, phosphorylated chains. After three years of research, I learned something else: I’d had enough of slaughtering animals. At the end of each day I would bring a bucket full of rat heads to a freezer in the basement. I wanted to work with humans, not rats, so I turned to a new technique after my thesis had been accepted: MRI, magnetic resonance imaging.

The first medical scanner I used was manufactured by Siemens. In 1857, Wilhelm Siemens invented the regenerative furnace, which led to a boom in the production of cheap glass. Until then, glass had been handblown, mostly green and brown, and quite expensive. The sudden availability of cheap glass had many consequences, but one that historians rarely mention is the boom in the art of the ship in a bottle. The first recorded date of this art is 1810; naturally enough, it comes from the seagoing nation of the Netherlands. Before the arrival of the Siemens furnace, glass would have been too valuable to be used for such a frivolous art. But a ship in a bottle is a beautiful thing. On days of shore leave, sailors made them from the leftovers of drinking sessions—matches and plug for mast and hull, a torn shirt for a sail—and then traded them for cigarettes and beer. Whenever I see a patient being reversed into the bore of a magnet, I can’t help thinking of a ship in a bottle.
Scanners have two main modes of operation. Functional MRI (fMRI) studies the activation of neurons—a kind of oceanography of the storms and currents inside the brain. I was more interested in structural MRI—the mapping of the hemispheres, simple cartography, the discovery of New Spain.
My first postdoctoral work after the patch-clamp lab was in Frankfurt. Alois Alzheimer’s grave was a short walk from the psychiatric hospital where I worked, and my first job was to analyze 1,054 scans from a study of Alzheimer’s disease. It was my introduction to “big data.”
To begin with, the scans needed tidying up. The way a photographer might crop a frame or equalize tint and tone, I corrected for motion and cropped the skull from each frame. Then the serious work began: alignment with nonlinear algebra. I had to find a template, a reference brain that would be the most representative of the entire sample. All brains would then be warped onto the globe of that reference for comparison. In short, more nonlinear algebra than a single computer could handle. I needed a supercomputer. I needed to learn how to code.

The Linux operating system is open-source, free. Almost without exception, it is the system of choice for every academic supercomputer. The reason is simple: for complex parallel computing, no other system is as powerful or as stable. Linux was conceived at the end of the twentieth century when Linus Torvalds, a student at the University of Helsinki, posted a message to the internet asking for some suggestions as to what people might like to see in a new operating system that he was hoping to build. He received useful feedback, as well as offers of help. Two months later, the first version of Linux was released, together with its source code—the program’s DNA. Without the source code, a program cannot be altered or developed by a third party. Linux was characterized from the beginning by its openness.
The kinship between Linux and the academic community is not accidental. The first computer scientists came from a small group of programmers who called themselves hackers. Today there are many kinds of hacks and hackers—good and bad—but the term originally referred to a small group of scientists working in MIT in the late 1960s, people who were passionate about writing code. It was only in the 1980s that journalists started to use the term loosely to denote those who specialized in writing computer viruses. The difference between black-hat and white-hat hackers is much like the difference between a chemist who is interested in making crystal meth and one who is interested in making insulin or quinine.
In his book The Hacker Ethic, the Finnish computer scientist Pekka Himanen describes an enthusiastic attitude to work that he pits against the idea of work as an obligation, even a moral imperative, an idea espoused by Max Weber in The Protestant Ethic and the Spirit of Capitalism. In Weber’s treatise, work is primarily a drudge, more often than not devoid of passion. By contrast, hackers are driven by a desire to create, often within teams that form spontaneously, as was the case when Torvalds and a small group of friends came together around the idea of creating a new operating system. Most of all, there is a commitment to share one’s skills. Individual breakthroughs are handed back to the community for the advancement of the project. Share early, share often is the hacker’s mantra. With many eyeballs, all bugs are shallow. This was the community that I was about to get to know.

A friend of a friend gave me the email address of a physicist who specialized in MRI at a research institute in France. I wrote to him. Despite the fact that he knew very little about me, he invited me to his lab. I gave a vague explanation to my supervisor and booked a flight. I had no idea what was going to happen.
The man I met on that cold December morning in Normandy seemed quick and sparrow-like. He was informally dressed, in his mid-forties, with short black hair and glasses; his words were fast and clipped. He seemed interested in the “project”—my code word for a big problem that I had no clue how to solve.
Only later did I learn how fortuitous my timing had been. Franck had been managing the supercomputer at his institute for many years. He was also running his own experiments—almost on the side, though it was supposed to be the other way around. A smoothly running supercomputer (which doesn’t exist) depends on a team of dedicated computer scientists working round the clock to keep a very complex machine purring over. Hundreds of emails needed answering every week—regarding software bugs, hardware upgrades, and the allocation of computing time, which gave his job a quasi-political tone, as research groups are always in competition when it comes to using a valuable resource like a supercomputer. The week before I showed up, he had handed over all that oily, dirty work to a colleague. I caught him in the transition phase: he was still enjoying the luxury of not having a hundred bugs to fix before the end of the week.
Curious, then, that his first meeting was with a stranger. As he settled into his desk, I noticed that his screen was filled with lines of green code—a command-line interface. No GUIs—graphical user interfaces. If you work on a laptop, you know what GUIs are: the windows and tabs that we click on every day when we use Facebook or Zoom or Instagram. In computing there is a clear distinction between the front end and the back end. Veteran programmers are notorious for building dated-looking home pages. What counts is the code; it is almost a point of principle for those who spend their lives working at the garage end of things to spend as little time as possible with something as frivolous as a GUI or a home page.
It didn’t take Franck long to realize that he was looking at a very “heterogeneous” data set—another code word, meaning a mess. As he called up images on the screen, lines of code started to flow. He asked me what I wanted from the “pipeline.” Maybe a T1-weighted analysis? What about tractography? Yes, tractography was what I was after—a technique I was attracted to primarily for its beauty. Tractographic maps trace the vast arboretum of white-matter pathways in the brain; the images this technique creates are the equal of any serti mysterieux and a good deal more complex: color-coded labyrinths of tangled trees and roots, a botany of forking paths that no one will ever fully understand. The timeworn analogy of the brain as a computer is lazy and incorrect. As an organic structure, its complexity is far closer to that of a jungle than a computer.
As the hours went by, some semblance of understanding began to form, but it was more feeling than understanding. I was beginning to see what the back end looked like. Ultimately, I wanted the “pipeline” to crop and turn, to warp and calculate every kind of analysis we could think of. Each scan consisted of three-dimensional pixels called voxels. Each voxel represented about a cubic millimeter of brain. Once we’d made a tractographic map for a single subject, each voxel in the map had to be projected into the coordinate space of the reference brain. These transformations with nonlinear algebra needed to be performed on each and every voxel. The total number of voxels in the data set was roughly equivalent to 1 followed by enough zeroes to fill a hundred pages. Neuroscience is a form of jealous counting.
On my second morning, I got to the lab at about nine thirty. We had worked for ten hours the day before. By the third day, even if I couldn’t keep up, I was familiar with the speed of Franck’s thought. In the manner of a chess player, he was always projecting three or four moves ahead, scanning the horizon for openings. The code itself was engraved with his own peculiar touch: he called the pipeline “Bushmills”; for debugging, he often named his variables “Moo.” But what I remember most from that first week is the way he would pause after he’d reached a confluence or a barrier in the code. I could see him working out the solution, but before he veered down any new line of attack he would always say, “I propose.” He made it sound as though I might have an alternative point of view; as though I might be able to point to something that he’d overlooked. But it was indicative of his way of thinking—a reflex and a style at once. Never presumptuous, that pause before the lines of code would start to flow again. Always “I propose.” For Franck, problems were adversaries to be outwitted. But there was an etiquette to his attack. On my last day I gave him a present of a black belt.

The academic model is far from perfect, and my circumstances in Frankfurt should make it clear that I am not naive about science. I lived on Weberstrasse, and the drudgery of the Protestant work ethic that Himanen had railed against was alive and well in Germany. I was based in a psychiatric hospital where patients danced and screamed in the garden beyond my office window. I might have even been working on the scans of some of those patients who were dancing in the garden. This hospital was the most hierarchical environment I have ever known. The director was a well-known autocrat and bully. He parked his Mercedes-Benz by the front door, where he insisted on having a space marked with his initials. A psychiatrist once mentioned in passing that he knew where the score would fall if he could ever get the director to complete a questionnaire measuring clinical psychopathy. The atmosphere was one of fear; the director’s strategy was to pit researchers against one another. Dozens of doctors left the department. The board of the hospital made an unprecedented effort to dismiss him.
I was immune from these politics, mostly because I’d found a way of working absolutely alone. It was made clear to me that if I didn’t produce a paper, I would be out of a job, but I was the only one automating MRI in the lab: they needed my code, or rather Franck’s code. It was an old-fashioned place; my colleagues were expected to squeeze a few drops of research out of the nanometer gaps in their clinical schedules. They had no time to learn new techniques. I was not a clinician, so I spent all hours of the day and night logged on to the supercomputer in France. Once or twice a week, Franck would also log on and we’d work together on the command line. I felt an acute sense of having to be on my toes for those joint sessions. I was slow, but how far I’d progressed can be noted by the simple fact that I had not coded before. Franck saw that I was learning. He must have also seen that my learning was a direct result of his teaching. That kind of feedback loop can go a long way.

Competition is integral to all work, and the open-source model of Linux uses competition to develop its operating system in an innovative way. Different versions are made by competing groups, and then Torvalds and a few principal developers decide which versions will be incorporated in updated releases of the software. If there is disagreement with Torvalds, any individual or group within the community is free to develop the project in whatever direction they choose by releasing their own version of Linux.
That sense of constructive competition was entirely absent in Frankfurt. Despite repeated threats from the director, conveniently conveyed by a middleman, I did exactly as I pleased. I felt no loyalty to anyone in that lab, but I did have a loyalty to Franck. I wanted to make use of the work he’d done. In the hacker model, work always begins with a problem that the hacker finds personally interesting. I enjoyed that sense of combat that I’d learned from Franck. I could feel how satisfying it is to write good code. Working with an inert mass of zeroes and ones, I tried to get light to pass through my code.

In the week that I sat alongside Franck, there was one ritual that took him away from the pipeline. He called it his “technology watch”—the half hour or so that he set aside every day for reading the online forum of the MRI software we were using. It was a program developed at the University of Oxford, also open-source. Beginners, professors, postdocs—everyone could be found there. The questions were often as useful as the answers—they pointed to new analyses that needed tweaks and hacks that might take the code in unexpected directions.
As I got used to asking questions on that forum, I was struck by the informality of the community. The only protocol was that you should read the documents provided by the group in Oxford; if you couldn’t find your answer there, then you should look for your question in the threads of previous discussions. If you still couldn’t find your answer, you were guaranteed to get help from the community. I was surprised when one of the original creators of the software answered a question that I posted. No “Herr Professor,” as was the case in Frankfurt. It was Steve, plain and simple. Another feature of the back end: titles were irrelevant.

I wish I’d known about Prince Boris when I worked in Frankfurt. I only learned about him before I visited Van Cleef & Arpels. Aside from his assertion that he came from a noble family of White Russians, not much is known about Boris’s life before he pitched up in Andorra in 1934 with the idea of becoming the monarch of that tiny principality. We do know that he arrived in England after fleeing the Russian Revolution without even a passport to his name. His flair for languages landed him a job with the British Foreign Office. After that, he claims to have spent time at the Dutch royal household, whence the further supposition that Queen Wilhelmina had granted him the title “Count of Orange” even though the title “Van Oranje” is a privilege of the royal family that cannot be given to a commoner. It is true that he became a Dutch citizen; the Dutch consulate at Dijon issued him a passport in 1923. His given name in that document is “Monsieur le Baron Skossyreff.” A year later, he appears on a list of “prominent foreign revolutionaries” compiled by the Dutch Central Intelligence Service. More succinctly, they call him  “an international swindler.”
On his brightest adventure, Boris was accompanied by an American millionairess, Florence Marmon, the ex-wife of Howard C. Marmon, owner of the Marmon Motor Car Company of Indianapolis. Using a generous allowance from her ex-husband, she provided the financial backing for Boris’s “reasoned presentation of his claim to the throne of Andorra.” The quotation is from a confidential dispatch preserved in the archive of the Foreign and Commonwealth Office of the UK. Some days later, the Council of the Valleys of Andorra, essentially the Andorran government, expelled Boris along with Marmon, who was acting as his secretary. When he decamped to the Hotel Mundial, in La Seu d’Urgell, three miles down the road, he was, according to contemporary accounts, sporting a monocle and generally affecting the airs of a prince.
While in “exile” he printed ten thousand copies of the constitution of his prospective kingdom, which more than irritated the bishop of Urgell—who was, and still is, the de facto co-prince of Andorra, together with the president of France. Remarkably, on July 8, the Council of the Valleys met at the manor house of the Casa de la Vall, the parliament of the principality, to discuss the matter further. With its wooden benches and rustic dais, the chamber of the Casa de la Vall is more like the banquet hall of a medieval fiefdom than a seat of government. At that meeting, the attorney general affirmed that the “dynamic outsider” was committed to making Andorra one of the most important business centers in the world. The vote of the chamber was 23 to 1 in favor of installing Boris as the new monarch. We should note that the mayor of Encamp refused to support the motion. The council met once more, on July 10. Another vote was taken—23 to 1. The mayor of Encamp would not be swayed, though it hardly mattered, as the new monarchy had been declared on the previous day and Boris was already busy outlining to journalists his plans for the kingdom: “protection for the needy, education for all and sport, a lot of sport.”
Sadly, the new sporting paradise was not to be. On July 12, he issued a proclamation declaring war on the bishop of Urgell, and the following week the Spanish Civil Guard sent three constables and a sergeant to apprehend him. His “subjects” watched as he was handcuffed and packed off, first to Barcelona, then to Madrid, where he was interviewed at the central police station by a number of journalists, to whom he displayed a fine knowledge of the Andorran dynasty from medieval times to the last incumbent of the throne, the Duke of Guise. When asked if he himself descended from the duke, Boris clarified, after some hesitation, that his claim was based not on historic rights but rather on “principles of chivalry.”

“Well, I could tell you that I’m Prince Boris, and if I was going to buy that watch over there, you’d have to say, ‘Thank you very much, Prince Boris.’”
I was talking to the manager of Van Cleef & Arpels’ boutique at the Place Vendôme. Leafing through an old ledger, my eyes had rested on the name of a princess who had bought a necklace in the 1940s. Her name contained the z’s and y’s beloved of the Polish language. The watch I was pointing to had a price tag of 126,000 euros. From Boris’s pre-Andorran days, I’d read a newspaper article about an attempted swindle of a gold watch.
Earlier, I’d asked the young man about the Alhambra, a line of jewelry based on the motif of the four-leaf clover and inspired by a recurring pattern in the sultan’s palace in Granada. It is one of Van Cleef & Arpels’ most recognizable motifs. Although my guide had an impressive knowledge of the titles of his clientele, his knowledge of the Alhambra was less formidable. The work on display was the same as what I’d seen in the workshop—near perfect, technically flawless—but if the workshops breathed an air of obdurate craftsmanship that elevated the work, the showroom had the opposite effect. The Queen of Persia, the Princess of Monaco, the Prince of Wales: the price tags were discreet, but titles were everywhere.
A monarch’s desire for precious stones is natural enough—all rare and beautiful objects are signs of power—but it goes a little deeper than that; probably it goes back to water. Glittering stones remind us of light shimmering on the surface of flowing water. Our innate reverence for gems as simulacra of flowing water has been coded into our DNA.
We can exist for only a few days without water. Stagnant water is more likely to be contaminated with bacteria. Early nomadic humans were more likely to survive where they were able to find fresh water, especially clear and flowing water.
After I put the old ledger back into the glass cabinet, I was brought to an inner chamber. An automatic door opened and closed behind me. I was in a space like an egg, with a velvet lining; the young man informed me that only a select few of the maison’s clientele were welcomed into that egg. A panel on the wall opened to reveal a floor-to-ceiling mirror in which a princess might admire a diamond necklace. But it was a minaudière that I was shown next, a rectangular box made of gold and lacquer. Conceived by Charles Arpels in 1933, the first minaudière was designed to replace a lady’s vanity case. Although its function is largely redundant, it still retains an air of exclusivity, possibly because it is now a purely ornamental object. The interior and exterior of every minaudière provides ample opportunities for the designer to show off his skill. The store manager showed me a hidden watch that could be looked at discreetly when raised from the surface of the box.
“It would be rude for a princess to be seen looking at the time,” he added. Almost on cue, his mobile phone went off. Turning aside with the tick of a smile that reminded me of a grasshopper, he took the call.

Authenticity is difficult to define, but we know when we are in its presence. True works of art are almost always discreet and unobtrusive. What is purely ostentatious is no more than fashion, and fashion is based on novelty—the mother of death, as Leopardi noted. In a letter to a friend, John Keats described how we dislike poetry that “has a palpable design on us,” poetry that tries to catch our attention by grabbing us by the scruff of the neck. How the flowers would lose their beauty, he said, if they called out, “Admire me, I am a violet, dote on me, I am a primrose.”
The most perfect gem I know is not a stone; it is a song of thirst and water. I’m thinking of “Piedra de Sol,” a poem by Octavio Paz that begins with a river and flows for 584 lines before turning back to its source. The first six lines are the last six lines. There is a line for every day of Venus’s journey around the sun. The eleven-syllable—hendecasyllabic—lines are as finely carved as the facets of a ruby in the mystery setting. As a work of art, Paz created a song that shimmers like a sunstone: it is light-in-a-word. As the main d’or works with rubies and Franck with code, Octavio Paz carves words. Looking through a microscope at Van Cleef & Arpels at a ruby that had failed quality control, I could see a crack at the tip of a facet. We feel the same errors in poetry when a line is broken by ugly enjambment, the turning point where a phrase is folded over two lines. The finest work should flow in such a way that the river feels inevitable. We say that it must be so—it could be no other way. Or at least, the work gives us the illusion that we know in advance what comes next because there is no alternative; what is flowing is something that we have always felt, even if we have never experienced the current before. Few works of art reach this plane, but those that do surpass their creators.

Working on code and working on poetry are not as dissimilar as you might think. For a start, both almost always begin with too many lines. The first pass is full of bugs. Algorithms and poetry throw error messages in similar ways—stumbling rhyme, a lost quotation mark, a stranded dollar sign—blemishes that take a long time to find and fix. Another rule of code that applies to poetry: If five lines can do the job of fifty, the shorter code always wins. This isn’t only a question of computational resources, or the possibility of sending a program into a tailspin, although both are important considerations; it is primarily a question of style. Poets might call it something else, but they would certainly recognize what computer scientists call syntactic sugar: well-written, elegant code. In 1913, Ezra Pound published his Imagist poem “In a Station of the Metro” in the magazine Poetry. The first draft was thirty lines long. The final poem is two lines long. William Carlos Williams was rebuked for calling his poems machines, but I think he was correct when he said that prose may carry a load of ill-defined matter like a ship, while poetry is the machine that drives it, pruned to a perfect economy. “As in all machines, its movement is intrinsic, undulant, a physical more than a literary character.”
But before I get too carried away, I should add that poets are very earthly creatures. Williams called his rival T. S. Eliot “that bastard” and went so far as to say that the most influential poem of the early twentieth century, The Waste Land, set him back twenty years. Make no mistake, competition is also an integral part  of poetry. The high and the low, the mortar and pestle—we seek to lose ourselves in work, to be consumed and absorbed by it, but the mulch and juice of our lives  is never far away,  even if the goal is to live for just an instant beyond the fray of  the world.

After months of writing code, I had a working “pipeline” that was built over endless hours of sharing scripts back and forth with Franck. There is nothing  more satisfying than submitting a “job” to a supercomputer, pressing the return key and, wonder of wonders, no error message is thrown. By the magic of parallel computing, I see that my single “job” has exploded into 1,054 sub-jobs. The code is working, and all engines are whirring. For comparing a patient group with control, MRI analysis often uses something called permutation testing. It’s a bit like playing a fruit machine five thousand times and logging the results of each spin. There is no greater pleasure than programming a machine to automate that kind of drudgery. Millions of calculations would still be running when I left the office on a Friday night. They would run through the weekend. They would be waiting for me when I logged on to the supercomputer on Monday morning. Some calculations whirred for more than a week before they came to a halt. Something complex and pleasing had spun out from the simple base pairs of our collaboration. Just over a year after my journey to France, our first paper was published in the Journal of Alzheimer’s Disease. I sent Franck a bottle of Bushmills.

The atelier of the jeweler, the lab of the patch clamp, the command line of the hacker, the white page of the poet—at the back end, the work is much the same. In all four cases, Gaudi’s dictum holds: love, then technique. It is only at the front end that differences emerge.
When I read the word poem inside a poem, almost always it is a sign that something has gone badly wrong. In my mind, an error message is thrown. When we call attention to the work at hand in such an obvious way, we are usually in the realm of the secondhand car salesman, the first cousin of Prince Boris. If there is poverty in a language in which there is only one word where two would be preferable, the opposite is also true. “Piedra de Sol” has been described as a surrealist masterpiece. I call it, plain and simple, beautiful. In the boutique on the Place Vendôme, I heard the word poetry so often that it reminded me that I once counted the number of times I heard the words “mi corazón” blaring from the stereo of a chicken bus that was taking me over the Andes. The scenery was sublime, the pop music less so.
The other tension between the back end and the front end is economic. Poetry has almost zero economic impact on the world, and that is, partly, its strength. As the Polish poet Zbigniew Herbert said, the problem of commercially exploiting genuine tears is a real headache for technologists. “Piedra de Sol” is not an expensive or an exclusive work of art. It is not a $126,000 poem. Likewise, a main d’or does not earn enough money to be able to buy the bracelet he has created. Yet he possesses it in a way that is more profound than the Polish princess who wears it on her wrist.

Back end, front end; open, closed. We have an obsession with dichotomies that really don’t exist. Everything sits on a spectrum. It is true that there are differences between left and right brain functions, but they are more subtle than the usual clichés would suggest. We create a belief system about ourselves and the world around us by merging information from the senses with our memories. This belief system is constructed to a large degree by the left hemisphere. If there is anomalous information that doesn’t fit into this schema, the right hemisphere may smooth over the discrepancy in order to preserve the image we have of ourselves. All of this is inevitable and to a certain extent useful. A degree of optimism is necessary for all tasks beyond the trivial, especially for work that may consume us for many years. The architect of any dream needs a breezy indifference to the enormity of the task ahead. We must focus on the day-to-day details while keeping the final goal just a little bit blurred. Too near a focus and the problem overwhelms us; too soft a focus and discipline dissolves.
If our optimism—or our delusions—have their uses, there is also a limit, and we cross the threshold of that limit when we fabricate information to such an extent that we end up sounding like Prince Boris. But even Boris deserves our sympathy—a man who lost everything in the Russian Revolution, a man who arrived in England like thousands of other émigrés, looking desperately for a way to survive. The newspaper accounts of his life are always black-and-white: the swindler-thief in a land of gullible peasants, the operatic prince seducing the craven millionairess.
We are all fractured creatures striving to reconcile our contradictory natures. The profoundest experience that life can offer us, said Octavio Paz, is the possibility to discover reality as a oneness in which opposites agree. Gems also have a divided soul; it is difficult to reconcile a glittering ruby with the mine from which it came.
Mineralogy is a dirty business; it needs arsenic as well as little railway tracks that go deep down into Hades. Soot and dust fill the lungs of miners. Outside the town of Potosí, Bolivia, are the mines of Cerro Rico. Half a millennium after they were first opened, they were still in operation when I visited them at the beginning of the twenty-first century. It is impossible to know how many people died in   those mines. A reasonable estimate might be five million souls, although there’s nothing reasonable about the history of mineralogy or mining. Descending into a shaft with head bent low, I came to a shrine for El Tío, the devil-god, the lord of the underworld. The miners still offer him cigarettes, dynamite, vodka, beer. He has a large erection. It is his union with Pachamama, the earth goddess, that created the bulging veins of Cerro Rico. El Tío has an insatiable appetite for souls. The miners try to appease him every day.
In the boutique at the Place Vendôme, a lady in a blue dress walked past me on a carpet that was as deep as snow. She bore a tray with little distillations of black: espressos for two sheikhs I’d seen at a desk looking at diamonds. Perhaps those two cups were an offering, a libation poured for El Tío.

I should add that I, too, ended up at the shiny end of science. After Frankfurt, I moved to Nijmegen, in the Netherlands, where I became the coordinator of what was at the time the largest clinical study of autism in the world. Front-end, jazz-hands kind of work. I thought I could continue to collaborate with Franck just as I had before, but the circumstances were very different. It wasn’t just me and Franck anymore. It was me and the Dutch Research Council, the ethical committee, Hoffmann-La Roche in Basel, EEG recordings in London, the translation of clinical questionnaires into five languages, PhD candidates to interview from Monday to Wednesday. I had time left over to code with Frank for a half hour in Thursday mornings. Each time I called him, I was unprepared. I felt terrible. I was relieved when I decided to call a halt to that scrappy coding after three short weeks, though I carried on with my jazz hands for another three years before I’d had enough.

On a bright morning in May, I left my apartment on Van Spaenstraat, in Nijmegen, headed for the Pyrenees—by bicycle. On my way to the mountains I called on Franck, who had moved to a lab in Lyon. He welcomed me into his home. In a peculiar way, that journey reminded me of my work in Frankfurt. I was free again to do whatever I wanted. Just as I was consumed by Bushmills, I was obsessed now with a single goal: I wanted to complete a hundred-mile trail run over the peaks of Andorra. I didn’t know it at the time, but that project would take me four years. There are quicker ways to get to the Pyrenees but there was no rush. Once arrived, I would have weeks by myself in the mountains to prepare for that trail race. I’m sure Franck understood that obsessiveness. Whatever I was going to do, I definitely couldn’t contort myself into a managerial post. The shiny surface, the Jesus bug walking on water; all that fla-fla, as Franck would say.
It was good to be at the back end again. The first six hundred miles had taken its toll. Franck was in the garden tinkering with the gears of my bike. He had the right tools for the job; he said that his grandfather had been a bicycle mechanic, that he’d grown up with chains and gears and oil. While he worked on the bike, I put my dirty clothes into the washing machine in the basement. I had a day of rest
before I would hit the road.