I’ve been fascinated with expertise since childhood. And it started over the chessboard. My dad would beat me—swiftly, crushingly, and above all, effortlessly—time and time again. He understood lines and positions in a way that I just couldn’t, and, as it seemed to me, would never be able to. My first question at the end of most games was: “where did I go wrong?”
Almost more unnerving than my dad’s ability was the fact that there were people out there who could, just as easily, beat him. “In college in Russia, I played a classmate of mine, who was a master,” he told me once. “I would think all night about my move, and then the next day in class, he’d move right away. Still, he beat me easily.”
Thus my interest in expertise was born. It seemed that some just had some sort of divine gift, which beckoned them onto a higher plane of understanding. For me to attempt to reach those heights would be futile. I could only watch in awe from below.
As I grew older, my skills improved. My games with my dad grew stricter and cleaner, until, one day, I beat him. In time, whether I won or lost, I was always able to give him a fair fight. I came to appreciate chess as an incredibly rich and rewarding game.
But my view of expertise—now that I had a taste of it—had lost a bit of its sparkle. We’ve all heard that 10,000 hours of practice can make an expert (1). I myself was approaching expertise, and it wasn’t by virtue of some sort of visitation. Expertise was, it seemed, simply a matter of paying dues.
Is 10,000 hours all there is to it? I sought to understand expertise further—even at the risk of growing to admire it even less.
Practice time certainly does matter. In his seminal paper The Role of Deliberate Practice in the Acquisition of Expert Performance, Ericsson studies violinists at the Music Academy of West Berlin. Students from a single department were classified as “best violinists” and “good violinists” by their professors.
There’s a clear correspondence between ability and hours practiced. By 18, the best violinists had practiced 7,410 hours on average, compared to the good violinists’ 5,310 (1). By age 20, the best students had largely reached the revered 10,000 hours. Maybe practice really is all there is to perfection.
The 10,000 hours figure shows up elsewhere. Chase and Simon estimate that chess masters have spent between 10,000 and 50,000 hours studying chess positions (2). What’s being achieved during these many hours? Chase and Simon go on to claim that chess masters have amassed stores of roughly 50,000 patterns. Experts are studying long hours to build massive vocabularies.
This finding, though, leads to what Ericsson and Staszewski (3) call “the paradox of expertise”: how do experts so quickly and effectively process the massive amounts of information they store? My dad’s college classmate moved instantly. If he did in fact store ten times the information, one would expect him to take ten times longer to find a move. Experts, though, aren’t just more effective than novices. They’re also quicker.
Ericsson and Staszewski propose that experts don’t just amass memory; they build skilled memory, where knowledge is stored in complex hierarchical structures. They study a runner named DD, who, after several years of practice, attained a digit span of 106 digits, the highest ever recorded. And, in examining the strategies he used to memorize this amount of digits, they found that his storage schemes were not linear, but rather strictly hierarchical.
A 100-digit string was broken into supergroups, each of which contained several small strings. Each small string was coded as a running time, age, or date. Each supergroup, then, might be coded as a certain runner, of a certain age, who achieved a given time on a given date. In this way, strings of random numbers were given meaning. Only by drawing meaning from disorder was DD able to recall a hundred digits at a time.
Of course, DD’s meanings were meaningful in a superficial sense. Each time a new 100-digit string was assigned, DD produced new runners, ages and dates. But in the case of chess, the same patterns are seen over and over, and eventually they develop nontrivial meaning. And once patterns develop meaning, they’re organized into the hierarchical structure according to their meaning and significance.
The difficulty of the task may lie in knowing what’s meaningful in the first place, and in knowing which hierarchical arrangements are instructive and which ones aren’t. But, once a simple scheme is built, the expert-in-training may use this scheme to develop a better understanding of the terrain, which he can then use to build more-sophisticated schemes. The process can then repeat itself. Through this procedure, the chess player builds much larger, deeper, and more complicated structures than the mnemonist. As DD learned to memorize 100 meaningless numbers, so the chess player memorizes 50,000 meaningful patterns.
This hierarchical tree allows the chess player to easily and quickly make decisions. In a game, an expert might see a position, which he would then associate with a given node in his tree. This node might correspond to the move white moves pawn to f5, in the context of black’s Sicilian Dragon opening, for example. An expert could quickly accept or disqualify that move, because he has easy access to the information below that node.
Underneath the expert’s f5 node, of course, lies much more detailed information: the fact that f5 unduly gives up control of e5, which is held by black’s black-square bishop in the Dragon defense, for example. Or the fact that white, by playing f5, allows black the opportunity to open up the f-file, if he so pleases, which extends the range of his own rook, which is favorable for black. And under each of these nodes lies more detailed, more fundamental information. Why is it bad for white if black opens up his rook file? Because the rook covers more squares, and can target pieces on white’s back rank. Under these nodes lies information even more fundamental. Why is it good for a rook to cover more squares? Because the piece has better control over the board. Perhaps we’ve descended to the level of the axiom, here. Of course, the expert doesn’t concern himself with any of this information. It’s modularized under the much-higher pawn to f5 is bad for white if black played the Sicilian Dragon node. The expert works with largely with the f5 node alone, only quickly dipping into what’s underneath it if he needs to. And herein lies the power of working with a well-built skilled memory tree: the expert can apply his short-term memory towards larger, more abstract tasks, in effect automating the smaller, more rote tasks.
And nodes exist even higher than the f5 node. Above this, for example, might be the much-larger idea that the Sicilian Dragon is a battle for the black squares and for the long, a1-h8 diagonal. In fact, it’s probably this node–the highest node we can come up with–that the expert truly works with. This way, instead of having thousands of things on his mind, he has perhaps just one. He reduces the entire, complicated game of chess to a single feeling.
This phenomenon explains why Capablanca, when asked how many moves he thinks ahead, replied:
I see only one move ahead, but it is always the correct one.
It also explains why the violinist can pour his heart out into a solo piece, or how the athlete can act largely on instinct. The smaller tasks–like putting one’s fingers in the right place on the fingerboard, or grasping the football along the laces–are automated, living in nodes far below that one with which the violinist or athlete is working. Instead, an expert works with much larger ideas. “Mozart intended for this piece to feel dance-like, and I want to convey that to the audience,” the violinist might say. Or, “I’ll be able to hit my receiver just past the 70-yard line, assuming my guards protect me in the pocket.” Experts and novices likely process information equally-fast, but experts are working with much-larger ideas, at the tops of their masterfully-built skilled memory trees.
Skilled memory allows the expert to overcome the paradox of expertise.
Skilled memory, then, is at the heart of what it means to be an expert. Experts aren’t just committing to memory lists of 50,000 units. They’re building massive hierarchical structures, where one travels not just forward and back, but, over, down, around, past, through, and, most importantly, up. The expert’s skilled memory tree isn’t a list; it’s an entire world, writhing with texture and structure, and the expert sits at the top of it. Experts aren’t list-learners; they’re world-builders. And the worlds they build are fascinating indeed.
In a way, the idea of world-building puts the mystique back into expertise. Sure: only through long, strenuous practice can an expert amass his knowledge. But this doesn’t mean that his practice is rote and mindless. Instead, he’s entering unknown terrain, searching for meaning, and using the meaning he finds to build a complex world that only he and fellow experts can access. I’ve got a bit of a chess world built up inside my own brain, and that’s something special. Meanwhile, I may be able to see the hours the grandmasters spend, but I certainly can’t see the worlds they build. Just because expertise is built with time doesn’t mean it’s not fascinating.
We can hope that, by practicing skills in world-building, we can more easily achieve expertise. In my graduate medical studies, my classmates and I are tasked with learning incredibly massive volumes of information. Surprisingly, though, I’ve heard very little talk about how best to assimilate, understand, and eventually commit to memory that information. And I have encountered quite striking instances of my classmates memorizing information using a list-learning approach, where that information could be much better and easier understood with a world-building approach. I hope to focus more on world-building in the context of my studies, and to share my ideas with my classmates when possible, so that we might all more easily attain the expertise required to practice medicine.
And, apart from expertise, we can use world-building to make the task at hand more enjoyable. I, for one, would much rather head to the library to build a world than to study a list.
- The Role of Deliberate Practice in the Acquisition of Expert Performance by Anderssen, et. al.
- Skill in Chess by Chase and Simon
- Skilled Memory and Expertise: Mechanisms of Exceptional Performance by Anderssen and Staszewski