Anyone who values their career (and stress levels) wants to work in an exceptional team.
For managers in the knowledge economy, the ability to hire extremely talented people is a skill worth learning.
And no, we’re not talking about the 'technique' of ‘We’re looking for someone with 5 years experience, who’s already completely proven in the role we want to hire them for’. Please join the back of the ‘world’s most unimaginative hiring managers’ queue.
The back of the ‘world’s most unimaginative hiring managers’ queue.
In this post we’re going to look at a concept that helped make one financial trader (Nassim Nicholas Taleb) a multimillionaire, and can also help you hire the best people. Essentially it’s about understanding the power of ‘extreme’ data points.
To pull this off, you’ll need to feel very comfortable going against conventional wisdom, and often having people think you’re wrong. But don’t worry, just like Taleb, you’ll have the last laugh!
Normal and extreme variance
First things first, let’s look at what we mean by normal and extreme variance.
There’s a good chance that you’re a knowledge worker in the software space, and therefore you know that the world of work is subject to ‘extreme’ differences.
That’s why you can find tech teams of 20-50 people churning out poor quality software at a snail’s pace (costing millions of dollars). Meanwhile, at the other extreme, the world’s best startups might be just two or three engineers, with little or no funding, producing great products in a matter of weeks.
With so much at stake, how do we find exceptional people?
Pick your top 3
From the list below choose the top 3 things you would be most interested to know about when speaking with potential future colleagues (e.g. when hiring for designers, developers, or product managers).
Note: Your personal info/data will *not* be collected, so please use the form as we'd love to see what people choose.
Now that you’ve chosen your 3 top lines of enquiry, let’s head to two magical places that will help you hire the best of the best.
Average and extreme outcomes
In his book the ‘Black Swan’ Taleb introduces the fictional countries of ‘Mediocristan’ and ‘Extremistan’.
These two places are subject to very different phenomena. It’s important to know which of those countries you’re in [metaphorically speaking] when examining data.
Let’s imagine you round up 100 people in your local area and line them up by height order. You’ll find you have a nice normal distribution (or ‘bell’) curve.
But what if we go and fetch the tallest person in the world? They stand a lofty 8ft 2 in, or 251 cm tall. You’ve got a big outlier, but your bell curve is still intact.
Let’s line up the same 100 people in age order. Your local town’s bell curve might skew slightly depending on demographics (e.g. towards school age kids, young professionals, or retirees), but you’ll still see a bell curve like pattern.
Now, we go and find the oldest person in the world (116), and add them at the end of the line. Again, you’ve now got a bell curve with a big outlier.
Now let’s get those same 100 people lined up by net worth (these are imaginary people remember, don’t try this at home, unless you live somewhere that’s incredibly open about personal finance), again we’d probably have a normal-ish distribution pattern.
Now we go and fetch the richest person in the world and add them into the mix.
The richest person in the world is worth over $200 billion USD. This time you’ve just found a ‘Black Swan’, and obliterated your bell curve!
Taleb describes a ‘Black Swan’ as ‘a single observation [that] can disproportionately impact the aggregate or the total’. Taleb points out that the first Europeans who went to Australia saw black swans. No matter the number of white swans that had been observed up until that point, it only took a single observation of one Black Swan to change their world view.
It’s clear that wealth is subject to extreme variation in a population (Extremistan), in a way that weight and age are not (Mediocristan).
Just to emphasise this point, in order to find someone who could dominate the age and height line up in the same way the billionaire dominates the wealth one (i.e. represent >99% of the total by themselves), we would need to find someone who was several kilometres tall, or thousands of years old!
...we would need to find someone who was several kilometres tall, or thousands of years old!
What’s this got to do with hiring?
Let’s go back to some of the important things we consider when hiring. How many of your ‘top 3’ come from ‘Mediocristan’? If none of them do, then congratulations!
By focusing heavily on the ‘Mediocristan’ type questions, you’re giving candidates very little chance to stand out as exceptional.
No one has 1000 years of experience in React, Golang, Figma, ‘building B2B apps’ and so on, and pretty much every candidate (no matter how exceptional) will have between 1-10 years experience.
It’s only by delving into individual achievements with those skills, you’ll start seeing ‘exceptional’ data points. Perhaps someone single handedly turning around the fortunes of a project, team, or entire department?
Case in point. If we take a young Steve Jobs, Bill Gates, and Mark Zuckerberg, and compare them with three ‘average’ candidates, using the criteria on the left hand side, the ‘average’ candidates will come out on top.
For a start the ‘average’ tech job candidates usually have university degrees, and they’re up against three college dropouts. Maybe you even filtered out Gates/Zuck/Jobs before the interview stage? 🤦
Don’t get me wrong, all the questions on the left hand side are definitely valid, but it’s only on the right hand side (and other questions like this) that you’ll see extreme variance in people’s responses.
Turning down the 'superstar'
Remember we said you’re going to have to go against conventional wisdom to pull this technique off?
Here’s what we mean by that. Let’s say you’re hiring for a backend developer, and two candidates arrive in your inbox:
Backend developer candidate 1)
A Harvard graduate (or another top university if you prefer). Experience with a big name/well respected software company, plus they have seven years’ experience using Python/Golang.
So far, so amazing.
Backend developer candidate 2)
Someone with no university degree. You’ve never heard of the company they work at (it’s fewer than 250 people). This candidate has 4 years of experience in tech, and just two years’ experience using Python/Golang.
So far, so not amazing. If you’re a ‘normal’ recruiter or hiring manager...
… But just like Taleb and his 'Black Swans’, you’re about to go against ‘the market’, and do what >99% of people would never do. That’s right, after interviewing both candidates you’re turning down candidate number 1, in favour of candidate 2.
How did that happen?
Your questions moved away from ‘Mediocristan’ (2 years vs 7 years experience in backend development, university degree vs no degree) to the right hand column. Using these types of questions we do see extreme variance.
If you’re doing your screening correctly, then far from being just another near neighbour on an imaginary bell curve, the most talented candidates should stick out like 10 metre tall giants!
In this case it turns out the first candidate, even though they’ve worked at some amazing companies, comes across more like a ‘passenger’. They’ve always been part of larger development teams, and it looks like they’ve been pulled along. Their individual contributions seem underwhelming, and they have no obvious passion for their role. Yes, on paper they look decent, but you’d be unwise to build a team around them.
Meanwhile our second candidate started out in customer service and switched to tech support; they then worked their way up the ranks of the development team in record time. They are completely self taught as a programmer (often in their spare time), and demonstrate a passion for customer service and technology (a rare combo).
Their individual contribution to projects shows they’re the linchpin of the team. They’ve been the one figuring how to solve pain points; designing, developing, testing, and deploying solutions that work. You’ve never heard of the company, but you ask questions, do some research, and figure out that the main product line is the real deal (modern tech, lots of customers). You're confident that the candidate will have learned (more than) enough to perform on bigger products and projects.
The trajectory of the second candidate’s career is phenomenal. They’re basically ‘jumping up' a level every 6-12 months, even if that means learning completely new skills or technologies.
Who wouldn't want someone on their team who’s clearly ambitious, customer focused, self-motivated, fast learning, and has an exceptional track record?
That’s the kind of person you want on your team for the next 3-5 years. Using the kind of tips we’ve explored here, you’ll be able to find these people more easily.
In the tech industry, the best leaders will build exceptional teams, and therefore exceptional software. To do this, you’ll need to consistently spot, and hire, exceptionally talented people.
If you fixate on the type of data that would always fit onto a ‘bell curve’ (‘Mediocristan’ type data), like ‘years of experience’, you risk not giving the best candidates a chance to stand out.
You could also be going down a rabbit hole of things that won’t necessarily predict future performance. Whether candidates have worked with X and Y tools already (AWS or Google Cloud, Adobe or Figma), may not have been under their control. Their team or boss, may have decided what tools people used, and that may even be the reason why the candidate is looking for a new role in the first place. A good hiring manager would rather find exceptional talent in the nearest adjacent tool vs uninspiring output with the target one.
Your role as hiring manager is to uncover exceptional data points, as these are some of the best predictors of future performance for candidates.
These ‘Extremistan’ data points can blow all the other information out of the water. You must ensure your screening process allows candidates to give ‘wildly’ different responses to questions. The open ended ‘what have you achieved using X and Y tools?’ or 'why did you choose to use X and Y tools for that project?', are very different from the linear ‘how many years have you used X and Y?’ or binary ‘have you used X tool before?’.
Oh and by the way, if people think some of your candidate shortlists are crazy, then don’t worry, you're probably doing this right! There’s a good chance that your team will be delivering great software in no time; and they’ll be wondering why their team (which seemed so good on paper) is struggling to deliver.
By David Fallon
Founder of Intaview.me
The Black Swan : the Impact of the Highly Improbable. (Taleb, Nassim Nicholas)
New York: Random House, 2007