In one of the more obscure corners of the internet, there is a raging debate going on in the chess world (‘A bird-seed seller beat a chess master online. And then it got ugly’) Well worth a read – here’s the nub of the issue: on chess.com, an obscure player from Indonesia rose rapidly in the ratings and horror of horrors (!), defeated an International Master in a match-up. The controversy: the Indonesian player was accused to have used an AI chess software to make the moves. What is interesting is why the accusation in the first place: expertise is expected to be earned the hard way – hours toiling away at the game, mastering moves, learning strategies. You don’t get there without the world noticing you along the way. AI and the rise of machines is not what caught my attention –this whole discussion on expertise in the game is interesting. What would have been filed as a flash-in-the-pan in almost any other sport, but then chess is clearly another game altogether.
Chess is often talked about as a game where the best players are romanticized as prodigies (don’t miss ‘The Queen’s Gambit’ on Netflix, if you haven’t watched already) with vastly superior analytical skills that they use to play out a game far ahead of their opponents. Research in the recent past is contrary to this notion – turns out that many of the chess master’s most important decisions about which moves to make happen in the immediate act of perceiving the board. What this implies is that the grandmasters are really retrieving information from long-term memory – or if you think about it, expertise appears to be a combination of vast amounts of knowledge, pattern-based retrieval, and planning/response mechanisms acquired over long periods of experience in the specific domain. And here’s an interesting sidebar: while the grandmasters could remember an exact layout of the chessboard by just glancing at the board for around 5 seconds, these very grandmasters were unable to remember, much less recreate the chessboard when they created nonsensical layouts. Which suggests that the expertise is not just domain specific – but also context sensitive.
This has been researched/written about over and over again – we have all heard of the 10,000 hours rule. And this is probably what ‘purposeful practice’ does in the chosen domain:
- Creates a deep repository of patterns that are available on tap
- Develops a pattern-recognition and retrieval capability that is specific to the given situation and works in two stages:
- Draw on the best match pattern from the repository. Research has shown that the best do the pattern matching not by getting lost in the details but looking at the big picture – in other words, taking a systems view, which naturally hinges on the ability to spot adjacencies.
- Assign a response mechanism that would be the best match for the current situation. And this too is often not the most optimal response, but the first option that is good enough – in other words, taking an incremental, learn-as-you-go approach
- Develops the ability to respond within the context of the situation. While speed is a key factor here, the experts have the uncanny ability to read the situation and tailor the response that is contextual.
This is what we call ‘expertise’: it is domain specific, context sensitive and takes focused effort over sustained periods of time to develop this kind of mastery.
The psychologist Gary Klein came up with a model called the ‘Recognition-primed decision’ (RPD) on how people make rapid, effective decisions when faced with complex situations. He studied diverse groups who are required to make quick decisions, often with incomplete information and poorly defined goals – think ER doctors, firefighters et al. In this model, the decision maker, when faced with a situation, activates the pattern-recognition process (see above) to pick a possible course of action, compare it to the constraints specific to the situation, and selects the first course of action that is not rejected.
What does this mean in the context of an organization? One clear trend that I see in the best organizations is problem solving with AI that doesn’t happen by accident – it is a carefully constructed pyramid with: Data and Technology as the base, teams of Data Scientists/Analysts and a small, elite team of expert problem solvers. It is increasingly clear that expertise is becoming essential – as the rapidly changing landscape keeps throwing problems with an ever-increasing velocity, it is critical to have people with the relevant expertise and with the ability to use that in what would look like RPD decision making.
And why is it important to build this kind of a capability? That goes into how the nature of problem solving itself has changed over the years. There was a time when you defined a problem in very clear terms and then set out to build a solution – the classical world of Software Development, the process of engineering and building to scale. Typically, these problems occur with a higher frequency and it is often possible to focus on a software solution. On the other hand, there is the long-tail of problems which are fuzzy, difficult to define and yet, deliver disproportionate value when solved. This is where domain expertise begins to make a difference – and organizations that are able to marshal the right domain experts and get them to focus on these long tail problems that will succeed than otherwise. More on the long tail distribution of problems later.