Specialization: Part 1

PHILLIPS_screwdriver_and_screwI didn’t do that well in physics. It was one of the engineering requirements for a CS degree, but I wasn’t that interested, pushed it off as long as possible, didn’t spend much time on it, and immediately forgot most of what I’d learned moments after taking the exam.

Fast forward a bunch of years, and I was working at a video game startup in LA trying to debug the physics in a car driving level. I tweaked the code this way, and the car went through the walls. I tweaked it that way, and the car floated through space. There’s nothing quite so demoralizing as having no idea what you’re doing, not having a pathway, and not even feeling like you have a pathway to a pathway.

Ironically, a couple years later I was working at a different video game company, optimizing and refactoring a much more sophisticated physics engine being used in a series of best selling driving games. I hadn’t built up a massive store of physics skills in the intervening years – on the contrary, I was then, and remain to this day, something of a physics dunce. The difference is that at the second company there were two physics PhDs working on the physics problems[1], and I was there to handle software engineering concerns. It worked out great! They had the right experts working on the right problems.

Everybody is a genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is stupid.
– Attributed to Albert Einstein

In First Break All The Rules, the authors discuss the difference between skills, knowledge, and talents. The idea is that skills and knowledge can be taught, but that talents are some combination of interests, habits, and personality – “recurring patterns of thought and behavior” – hardwired neurally during childhood and adolescence. Talents are the things you do on your own, whether you’re supposed to or not. It’s what you do on weekends, and late at night, what you think about at odd moments, the subject of the specialized news you read on the internet, what you get excited talking about, your passion. No one has to convince you to do these things – you just do. If you won the lottery, you’d still do them, in one form or another. You’re neurally wired to.

You might have a talent for researching genealogy, learning foreign languages, or confrontation. It could be organizing parties, negotiating prices, solving visual puzzles, or managing complex social situations. No one has all talents, and when you marvel at just how good someone is at something you know would drive you mad, you’ve just identified their talent and your gap. Of course, even someone with a specific talent needs to develop the requisite skills and knowledge, and this is what we all do. Artists doodle. Programmers tinker. Bakers bake. And as we do, we get better.

There’s nothing magical about this. In Babel No More, Michael Erard explores the story of Cardinal Mezzofanti, a 19th century prelate said to have spoken dozens of languages:

At the end of his story, however, he finds a surprise in Mezzofanti’s archive: flashcards. Stacks of them, in Georgian, Hungarian, Arabic, Algonquin and nine other tongues. The world’s most celebrated hyperpolyglot relied on the same tools given to first-year language-learners today. The conclusion? Hyperpolyglots may begin with talent, but they aren’t geniuses. They simply enjoy tasks that are drudgery to normal people. The talent and enjoyment drive a virtuous cycle that pushes them to feats others simply shake their heads at, admiration mixed with no small amount of incomprehension.
– The Economist, The Gift of Tongues

They simply enjoy tasks that are drudgery to normal people. That’s the secret. You love doing things that would bore me to tears, and vice versa. Which is why if you don’t have a specific talent, it’s very hard for you to compete with someone who does. If you just aren’t that interested in obsessively memorizing chess openings, it’s going to be hard for you to beat someone who spends all their spare time at it, because that’s what they love to do.

On the flip side, a non-talent is an area in which you’re singularly lacking in aptitude. Giving people work that aligns with a non-talent will result in failure. We’re all familiar with the stereotype of the star engineer who fails after being “promoted” to management. This makes intuitive sense – while some people make the leap, technical and managerial talents are mostly orthogonal, and there’s no reason to believe that a talent for shutting out the world and solving logical puzzles should coexist with talents for project management, recruiting, effective 1:1s, and political savvy.

The core point of First Break All The Rules is that great managers align people’s work with their talents. This is a pretty banal statement, an old chestnut we hear all the time – basically the work version of the parental advice to choose a major based on the homework – not the classes – that you enjoy. It’s also critical – people whose work aligns with their talents turn into stars.

Although a useful framework for thinking about performance, there are two main dangers with this whole line of thinking. The first is that you can use it to justify your biases: “People in demographic group X have talents A and B, but not C. Person Z? Oh, they’re the exception that proves the rule.” This is the kind of toxic “reasoning” that went into the so-called Google manifesto.

The second danger is not knowing exactly what talents a job requires. It’s easy to create a picture in your head of the “ideal” manager, engineer, PM, whatever – but like the average human form, there’s no such thing. I know highly effective technical managers who are brilliant engineers and architects, others who are great project managers, others who excel at recruiting and training. As T-shaped generalists, they can take on and do well in most technical management roles – but find a job in which their specific outsized talents are the critical ingredients to success, and they’ll be stars. Sometimes you know in advance that you need a specialist, but a lot of the time it only becomes clear in retrospect – either because they’re doing something unexpectedly amazing – or because they’re only performing acceptably on critical emergent needs.

So where does this leave us? We have a conceptual framework that we can’t use because it will only serve to reinforce our biases, and it’s hard to know how to identify the optimal talents for a role anyway. Oh well. Thanks for the pointless blog post, Dan.

OK, not quite. As long as we’re on our guard, I think we can learn some useful lessons.

The first thing it tells us is that people are different. This is so trite as to be a bit embarrassing even to mention, and yet most people walk around looking at people with different talents, and just see broken versions of themselves.

Second, there are some things that you will never, ever be able to do well. No matter what classes you take, what books you study, how long and hard you try – you will always be mediocre at best. If your job – or advancement – requires these talents, then you’re going to need to think hard about how to delegate, obviate, or redefine. Or, indeed, whether you should be in the role at all. This is a bitter pill to swallow, and the opposite of what all those uplifting after-school specials about the power of dreams taught us.

Third, that people can learn skills and knowledge, but they’re fundamentally not going to change who they are. Same goes for you. Stop trying to change, and start trying to improve. I have a talent for learning languages, but it took me a long time to develop the necessary vocabulary memorization skills. You may have a talent for solving coding puzzles, but becoming a senior engineer takes years of diverse experiences to develop the requisite skills and knowledge.

Lastly, and to the point, it explains why the search for the generalized 10x-er is a failed exercise. It’s the match between the person, role, and environment that creates magic. This will be the subject of the second half of this post.

Enter your email address to follow this blog and receive notifications of new posts by email.


[1] It turns out that solving for the generalized collision of two cylinders requires solving an eighth degree polynomial. My suggestion – why not use “hotdogs”? (swept spheres) – got us down to a linear equation :)

[2] No one reads the footnotes – EXCEPT YOU, YOU BADASS! Bonus Saturday Morning Breakfast Cartoon for you.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s