Why We Fund Teams And Hire Individuals
Our conception of talent remains incredibly individualistic despite most of our achievements being collective
There's a curious disconnect between the founding stories of successful corporations, artists or scientists, and the entirety of the pyramid that works below them. At the top there is clarity on the benefits and joy of working together, making you unstoppable! Below that there's individualised role descriptions and malleable teams.
The founding mythos of silicon valley often revolve around a pair of founders - the Steves (Jobs and Woz), Gates and Paul Allen, Moore and Noyce, Larry and Sergey. The list is endless. To the point that in tech the received wisdom was that unless you had a cofounder you couldn't even get funded. I know folks like Mark Suster have warned against it, but it still seems the norm.
Outside tech, there's Procter and Gamble (yes, they look exactly how you'd imagine), Ben and Jerry (also, yes) or Wells and Fargo. The legendary friendship and partnership of Buffet and Munger is another paradigmatic example of this in action. Michael Lewis even wrote an iconic book about Tversky and Kahneman, exactly such a friendship and collaboration, and one that helped both of them generate the best work of their illustrious careers by some margin!
Even rabbinic studies use co learning methods called chavrusa to study, learn, analyse and debate Talmudic texts together. While at its core it can just mean a study-buddy, the idea seems broader than that.
If you flip over to Hollywood you see the same thing. Whether its Brian Koppelman working with David Levien, or the power duo D&D adapting Game of Thrones on HBO, Paul Anderson working with Philip Hoffman, Tim Burton and Johnny Depp. The list goes on.
But what emphatically does not seem to happen is the hiring of teams, or even coworking-pairs who were pre-existing, finding places to work together. The atomic unit is an individual, and despite the increasing complexity of the world and the workplace, we're sticking with that.
Even if you go back to the very nascent realms of building an organisation, we overwhelmingly prefer discovering existing teams (every VC firm ever) compared to say Entrepreneur First which tries to match up strong co-founder teams. The EF underlying thesis seems to be that it's possible to create strong teams than just discover them.
But this is the exact opposite once you go a level below the co-founder, co-creator level.
Below the CXOs, we overwhelmingly prefer the EF model, where we think it's far easier to build teams than even try to find already existing, already working, teams.
Isn't this weird?
In fact, what EF did seemed heresy because they took what most organisations were doing most of the time in most places and applied it to the 0.01% that wasn't.
Why is it that apart from literally the most fragile, highly demanding, highly contentious stages of starting a company, we don't think about applying the same principles to other parts of hiring? It does seem to be a curious blind spot.
Bear in mind, we don't treat the rest of our lives this way. We have friends and we have best friends. We have people we love to hang out and play with. For those of us interested in team sports, there are people we love to have on our team. And not just because they make us better, but also because we enjoy it more.
Even in professional sports, there is the quintessential question of the great pairs who make lasting impacts together. Here, take a list of 30 of them to start with.
But somehow, in the world of business where we spend >50% of our waking lives, this recedes into the backdrop. It occasionally emerges as a soft form of mentorship or an anaemic form of apprenticeship but rarely is it even acknowledged.
In fact, I can only really think of Marc and Ben at a16z who have consciously spoken or written about their partnership in this fashion. But even they don't seem to hire GPs in dynamic pairs to replicate their fantastic magic.
We seem to have decided that the things that are incredibly hard (starting a business) requires two or more minds in close collaboration more often than not. But somehow the equally hard problem of continuing to innovate seems relegated to the even harder task of rearranging pyramidal organisational boxes.
There is a great New Yorker article about the power of working in pairs and the wonders it can achieve. It references the seminal book on this topic, Collaborative Circles by Michael Farrell. He references a bunch of places where folk work creatively and collaboratively to great effect. Monet and Renoir for instance worked together to create Impressionism, and Picasso and Braque collaborated to create Cubism! Same on Lennon and McCartney for song writing.
As you might note, these are all artistic collaborations. Is there something special about art that enables this feeling a little bit more? Perhaps, but despite mixing up apprenticeships and collaborations, the subject of that articles notes how a pair can often be more powerful than its individuals.
And this shouldn't surprise us! After all, any Coasean knows that this is sort of a shorthand for businesses overall. We all try to be part of a super-entity that tries to generate more than us individually could, and that stands true for the gargantuan world of government as it does for the lowly startup.
The article however showcases Jeff Dean and Sanjay Ghemawat, pair programmers extraordinaire at Google, who had seminal input (together) in many if not most of the company's aha! moments. Not content with speeding up Google's queries many 1000x over time, or optimising so that searches doesn't fail randomly, or ensuring the company didn't fall over under the weight of the data they collected, or creating MapReduce, it goes on and on about their achievements.
Like a couple, Jeff and Sanjay tell stories together by contributing pieces of the total picture.
And in the interview they both expound on the benefits of their particular style of working.
“I don’t know why more people don’t do it,” Sanjay said, of programming with a partner.
It bears mentioning that this isn't one of those innovations that Google dreamt up. Sanjay and Jeff talk about how they had done it since before they joined Google, in their previous stint at DEC.
And it's not just at Google. Pair programming seemed to be wildly popular, at least when it's done. Though this data is from 2000 96% of programmers apparently loved it. And isn't it a bigger mystery that despite 96% of supposedly your highest paid and most valuable and most volatile employees embracing it, you're not seeing more of it?
There has been little more researched around the fringes, but mostly to help build tables like this:
Mostly they conclude that there are pros and cons and you shouldn't use it exclusively. Well, yup. But you'd think that not exclusive wouldn't necessarily mean non-existent. And yet, there aren't very many of these pairs in the world. Which is a rather big mystery.
And it's also a mystery because when you speak with the best CEOs, they do mention this strategy explicitly, though only for enterprise sales. They say that the way they try to make sure the team works well is, say, first to hire the best VP sales they can find, and then the VP Sales goes on to hire the best from the previous teams she's led.
I think Ben Horowitz even has a part of his book where he talks about the necessity of this line of questioning, because that's what separates a superior VP Sales from an inferior one.
And yet, and yet, and yet, we still don't see much of an appreciation for the fact that folks who love working together should probably be kept together, except in the most cursory sense.
Otherwise there's no reason why in this multi-trillion dollar economy filled with hundreds of odd peccadilloes (nap pods shaped like cocoons, bouncy balls for thinking, pirate ship decors for meeting rooms, and all that's just at Google) this isn't more prevalent.
As seems to be normal nowadays, Stripe seemed to have led the way. They actually ran the experiment, and seemed to get kudos from pretty much everyone (and a whole bunch of skeptics). They had a Bring Your Own Team idea where they tried hiring a group of 2 to 5 people.
Any group of 2 to 5 people can apply as a team to Stripe, through our application form. Make sure to include resumes or CVs for each person, indicate which role each person is applying for, and a brief description of how you all know each other or have worked together in the past. Links to (or attached samples of) things you’ve built together are especially helpful. We’re expecting teams to be primarily software engineers, but we’d love to see well-established collaborations between engineers and designers, managers, or product managers.
Wanna guess what happened?
After spending about a year on the BYOT experiment, we’ve decided to sunset it. In short, it didn’t work; we didn’t hire any teams as a result. There were a few reasons for this, which we learned along the way. First, coordinating multiple career moves at once is even more logistically tricky than we realized. Many candidates who considered BYOT ended up applying as individuals. Second, much of the interest in BYOT came from talented teams in countries where Stripe isn’t yet ready or able to hire full-time employees.
Sadly it fell apart. But at least part of the reason is us. I can complain about why the demand side of the problem is lacking, but we haven't fixed the supply side problem much either!
The unfortunate part of it is also that we constantly talk about the idea that working together from an early age but completely skip over the part of what to do once you find folks that you work very well with. We don't even realise it is an option, much less encourage it. Yes, you do need to find someone compatible, but it shouldn't be a surprise when you do because you never thought of it. Just imagine what that would be like for marriage.
Even the Avengers happily teamed up with existing teams, the remnants of the Guardians, Hawkeye and Black Widow, so it’s not even as if our cultural mythology is completely opposed to the idea. It just doesn’t seem to happen!
There are plenty reasons why this is difficult on the demand side. Companies have a hard time finding one, you want us to find two? Every conversation I have with a manager or a CEO has had the same refrain - hiring is broken. This seems commonly accepted. Companies look for talent, wanting to hire superstars, as do recruiters, as do those individual superstars themselves who want to find work. But somehow or the other those two don't seem to be able to easily find each other. So finding two rather than one is far more than 2x the trouble.
Well, maybe. It might not be the perfect solution for all hiring, but considering a) startups and major tech companies are in a desperate war for talent, b) absolute top performance is supposedly worth 10-100x of a regular performer, wouldn't the strategy to identify the best and bring out the best in people be at least worth considering?
Also, the much maligned acquihire is a sort of bandaid solution that solves this exact problem and that does seem to happen more often. It surely isn't much of a leap to imagine that you don't need to suffer a few years and fail before this becomes a viable option!
And if the answer is on the demand side, that's more likely. After all, why aren't there more pair who like working with each other and therefore look for roles together? From asking around, I'm not sure more people reading this newsletter would be able to easily answer that question either.
At least part of the reason seems that nobody seems to even think of this as a viable possibility. And as a result, even if you do have the chance of building a budding collaboration with someone, it easily gets shut down unless you're the one starting the company, or working autonomously.
As a HN commentator pointed out:
I was on a team that tired this - we Formed, Stormed, Normed and Performed at $big_company then all wanted to leave at once after a change in management. We touted ourselves around as a ready-made team and nearly got hired by a couple of places but it didn't happen.
Amidst all the glory that cofounders and co-collaborators get, we seem to forget that it's not just the scientists and CEOs that work together. In fact it's everyone else in the giant organisations beneath them who need to rely on each other much more.
We seem to be implicitly advocating for a philosophy that building is easier than buying when it comes to high performance teams.
Our conception of talent remains incredibly individualistic despite most of our achievements being collective. Just like in knowledge production, where scientific knowledge is increasingly produced by larger teams, it would probably be helpful to start thinking beyond our atomic conception of talent.
It's not about losing individuality, but being open to the idea that our best work is done with others. That you don't need to want to start a company or direct a movie to necessarily work together with folks with whom you can do your best work.
By having too narrow a lens we miss out on the joys of producing things collectively. As Guzey said:
I no longer believe that it’s possible to achieve extremely high productivity sustained over long periods of time working on difficult projects alone, so now I spend the majority of my working time co-working with my friends over video
Occasionally, when it does seem to happen, the outcome sounds idyllic. Later from the same HN thread.
I did have two acquaintances from long ago when I was a humanities major at school who did this. They were roommates in college, both worked for the same lab, then worked at the same tech company after college.
They later moved to California, continued living together and working on the same teams at work. Eventually, they got separate houses but still work together as a high functioning mini-team in their own shared office at a huge SV company last I heard. It probably raised eyebrows at interviews saying they wanted to work together, but they excelled technically and in the long run it seems to have been a significant career advantage that raised their productivity and possibly even rate of learning.
This is what we're missing out on!