How to Identify Underrated Talent

1 February, 2023
Decision-making; Startups

Identifying underrated talent is an edge.

It’s much like finding underpriced stocks: being simultaneously non-consensus and correct gives you a competitive advantage.

Forming these opinions is as much art as science. It’s a skill that’s trained. 


Through pattern recognition, practice, and developing your intuitions.

I’ve spent the past 18 months learning to identify underrated talent with Next Chapter, where we bring together the most talented young people in the world.

To identify underrated talent, we want to find people who have accelerating growth curves. I’ll first explain the philosophy of growth curves. Then I’ll break down how I assess growth curves. 

Part 1: Understanding growth curves

My goal is to popularise the language of growth curves. That starts with definitions:

Distance refers to how much someone has achieved to date.

Speed refers to how fast a person is growing.

Acceleration refers to the rate at which a person’s speed is changing. Positive acceleration looks like 1 unit in year one -> 2 units in year two -> 4 units -> 8 -> 16 -> 32 … you get the point.

On the chart below, we see that acceleration > speed > distance.

The implications?

  1. Initial distance doesn’t matter in the long term
  2. Speed is crucial. People who improve fast have steep growth curves
  3. Acceleration matters even more than speed. “Accelerateurs” continue to speed up. They are ‘exponential improvers’. They compound knowledge.

To find underrated talent, identify ‘accelerateurs’. 

Part 2: Assessing Growth curves

Now for the tricky part: actually estimating one’s growth curve.

I’ll step through:

  • 2 preliminary thoughts
  • 6 questions to assess talent
  • How to think about work samples
  • How to think about brainteasers and IQ

2 preliminary thoughts

First, a warning: be wary of excess credentialism. 

Focusing on credentials is a good way to assess ‘rated’ talent, not under-rated talent. Credentialism misses outliers. It biases distance over speed.

As a corollary, when presented with two seemingly equal people, often the one with the least label-oriented background is more impressive. This suggests they succeeded in spite of having fewer credentials and a less privileged upbringing.

Second, a guiding principle: actions speak louder than words. 

Look at someone’s prior actions (not jobs). 

Where have they shown agency? 

What do their side projects say about their curiosity and drive? 

Do they ask lots of intelligent questions? 

Where have they rejected societal defaults?

Another useful predictor I’ve noticed is using LinkedIn, Twitter, and GitHub in high school. Only a curious, ambitious person tends to do this.

From here, it’s question time. 

6 growth questions

1) “What are your goals? What are you willing to give up to reach them?”

People with high acceleration have ambitious goals, often which require prioritisation and sacrifice

As Ray Dalio says:

“You can have just about anything you want, but not everything you want.”

An extreme example of prioritisation was Bill Gates:

“I never took a day off in my twenties. Not one”

Consistency matters since it opens you up to compounding knowledge.

A related question is “how successful do you want to be?” This question is unexpected which makes it surprisingly difficult to fake.

2) “What were you like as a child?”

I’ve noticed an uncanny correlation between childhood behaviours and later success. 

As children, the successful were often entrepreneurial, action-oriented, curious, tinkerers, competitive, and often had an obsessive hobby that they would do for hours on end. 

A question I ask is “how do you compare to your siblings or friends?” The comparison is often more telling than just getting the person to talk about themselves. 

3) “What’s something you have worked on with the least ROI?”

Working on something with low ROI is a sign of intellectual curiosity and work ethic. It shows intrinsic motivation.

I’ve seen a trend where a hobbyist (a person who creates out of passion) transitions to a tinkerer (a person who creates to solve a problem only they have) to an entrepreneur (a person who creates to solve problems they share with lots of people).

4) “What’s an uncommon book (or person) which has influenced you the most?”

This assesses intellectual curiosity and drive for self-improvement, which correlates with steep growth curves. Most wildly successful founders or investors are voracious readers.

The question also reveals the person’s interests and Girardian models of desire. 

Asking “why” as a follow-up is particularly useful here. Talented people have justified opinions. They can articulate why they like something.

5) Question: “What do you do in your downtime?”

Downtime-revealed preferences are telling. You are looking for people who spend their downtime tinkering, learning, building, or upgrading themselves. Top performers prioritise intensive self-improvement over more ‘normal’ hobbies. 

An interesting follow-up is “What are the open tabs on your browser right now” (h/t Tyler Cowen). The problem with this question is that the variance for a single individual is too high – a single person oscillates between no tabs open and hundreds.

6) Question: “How do you upgrade yourself in your day-to-day life?”

Top performers focus on improving the rate at which they can improve. Look for strategies and systems to perform at higher levels of efficiency.

To assess this, three questions I like:

  • “How did the way you did your last job change between the beginning and end? 
  • “What would be your advice to yourself if starting again?”
  • “How have you changed over the past year?”

How about work samples?

Work samples are good but not great. They are particularly good at assessing distance. But they have three key problems:

  1. They are gameable – people can collude with others
  2. Experience is a confounding variable – work samples overweight distance and underweight speed
  3. ‘Time spent’ is a confounding variable – if you tell someone to “take 2hrs”, some people will take 20hrs. Perhaps the most talented people don’t have 20 to spare

Overall, I think work samples are good ways of assessing mid-level hires. I think they are limited in identifying accelerateurs and founders. 

How about brainteasers?

Brainteasers aim to assess “horsepower”, or IQ. 

IQ is necessary, but not sufficient. 

I’ve noticed that IQ also correlates (imperfectly) with curiosity and a desire to learn. IQ can therefore be proxied by the person’s (a) responses to the 6 growth questions, and (b) questions back to you.

That said, well-constructed, left-field brainteasers can be a great test of “horsepower” and creativity.

Let’s look at an example: 

“In 20 years, you wake up and there’s no such thing as a bank. What has replaced it?”.

Low-horsepower people get often revert to answers that are attached to the dogma of the present: “alternative lenders, BNPL, fintech, crypto”. They revert to availability bias rather than first principles. Their answers are “skeuomorphic”. 

Stronger thinkers might go back to a “jobs to be done” framework. Or they might first ask “what does ‘no such thing as a bank’ imply about the society we are in?”

Overall, I think brainteasers are helpful but have a weaker correlation with growth curves than the 6 growth questions.


My goal is to popularise the language of “growth curves” and “accelerateurs”. It’s impractical to assess talent without a common vocabulary. 

Once you understand growth curves, the questions and principles I’ve provided help determine the curve’s shape.

The people who succeed aren’t normal: they are obsessive readers, listeners, askers of questions, absorbers of knowledge, triers of new things. They have to do things to get to sleep.