Illegible Medicis And Hunting For Outliers
On black swan farming, VCs vs grant-making, the spectrum of illegibility and part III on our new institutional order
Why change the world? Change worlds!
One of the most interesting developments of the past decade has been the extraordinary success of YC. In the world of investing it has created an impressive counterpoint to the perfect big-game hunting of Berkshire Hathaway. Where Buffett focuses on having a wide margin of safety and ensuring adherence to his #1 rule of “never lose money” by only investing in winners, YC focuses on having a wide margin of risk and ensuring they didn’t miss out on any outliers at the cost of plenty of losers.
The spectrum is also the spectrum of legibility vs illegibility. Warren Buffett’s investment thesis is, at its heart, highly legible. Be within a circle of competence that is extremely tightly drawn, with as few question marks on the asset as possible, long holding periods with compounding doing its magic. Even at its genesis, with Ben Graham’s teachings in mind of cigar butts, the answer was still to be legible - look for those opportunities where the metrics outmatched the markets, and place your bets there.
With YC its the opposite. The focus is on the highly illegible opportunities. On betting on founders and on their ability to achieve the impossible. Oftentimes achieve that impossible outcome which they themselves might not be able to visualise at the outset. Of doing what Paul Graham calls black swan farming to ensure they capture all the outliers.
Even a decade ago the YC way of thinking was close to anathema. Despite the existence of venture capital as an asset class, and despite its prominence in the world, and despite the 2014 book from Peter Thiel Zero To One, the ideal of adhering to a power law philosophy is deeply unsettling for most investors.
If you’re used to being right for a living, then being wrong on purpose is jarring even if that is the right thing to do. But that is indeed the world we live in.
We have, however, learned from the existence of this new strategy. In fact enough of us have learnt enough things that we look at the existence of this strategy as a way to change the face of large swathes of previously existing institutional knowledge.
Most of the modern world was built in the shadow of needing to create and drive big changes. And to make it happen we built up our institutions to help guide our paths. They were the mainstay of our success for so long that the very notion that they weren’t prebuilt part of the societal fabric seems ludicrous. They’re part of the scenery, and expected to be functionally infallible.
Think of it this way. The current institutions we have erected, whether in academia or government or private, are focused on getting things right. When they don’t, we pillory them. When they get calls wrong, whether about wars or markets or the behaviour of dictators abroad, we thrust their entire class of expertise under a sceptical microscope.
And oftentimes, rightly so.
And yet. And yet. When we get upset at the bureaucratic meanderings of our various three-letter-agencies we are, in essence, arguing for a world where these agencies are able to take decisions that are a tad more risk-friendly, or at least less risk-averse.
Since this is a hard reform to push through in aggregate, what we have seen instead is the flourishing of new institutions who are now able to support this very need. Who are able to make the illegible needs of our complex modern age their north star. Who are able to move fast and make mistakes and learn and adjust and evolve.
Once upon a time, I wrote about how we needed many more Medicis, lamenting at our collective failure to grab the fruits hanging so low as to be knocking on our collective foreheads. This was a mere 6 months ago. At that time this seemed a forlorn idea, only embraced by iconoclasts like Peter Thiel.
But instead, it was as if I was lamenting the lack of speciation right before the Cambrian explosion. I'm glad most of you took my advice, that too before I made it, which is gratifying.
Let’s look at the starting roster.
Emergent Ventures (the OG), and Fast Grants
I’m stopping here, but the list is ever expanding!
There are two common themes here. That’s speed and autonomy. They mostly act under the assumption (the correct assumption it would seem from a betting lens) that they can identify talent, not bug them excessively, and leave them to do their thing. Instead of imposing rules and strictures and guidelines, they focus on letting the innate megalomania do the work of focusing their research.
First, new organisations usually will have higher productivity on the margins (with higher volatility), simply because they’re playing a different game than others. Starting from a blank sheet and a narrow sense of purpose is a hell of an advantage.
In most things at scale, there’s usually the possibility of improving the performance if you radically narrow the scope. This is the secret behind startups, behind new institutions, behind new approaches even.
If talent has anywhere close to a downward sloping line when compared with, shall we say, bureaucratic encumbrances, stands to reason for at least a portion of that talent would do better when encumbrances are gone.
It’s easier to perform revolutions at the margins than on averages. But if you pull on ends of the curves enough, the whole curve might shift.
Second, the organisations have the merit of illegibility, but that’s also its drawback. The very fact that they aren’t rule-bound to do things just so means that they are able to take risks deemed inappropriate or ill-advised for the larger behemoths.
The merit of legibility is that it's agnostic to the individual, and the merit of illegibility is that it's agnostic to the system.
The very malleability to the details is what makes illegibility great in circumstances. If I trust that you'll do the right thing, I don't need to worry about you having to explain to me how you'll do the right thing.
(A fascinating trend on all these organisations by the way is that they explicitly call our how they aren't replacements to the well functioning existing ones, but a parallel route. Like a cycler lane besides a clogged highway. This amuses me because the normal rhetoric is almost exclusively on how the existing institutions are useless.)
So far, so good. The move towards increased illegibility, increased autonomy (for some) and lessened bureaucracy is a fantastic thing because it'll loosen up innovation for sure.
The key to a power-law organisation is the simultaneous realisation that:
Most of your returns will come from a very small sample of your actions
Most of your actions will not be fruitful
Most of the time you will not know if you’re succeeding or failing
In the world of startups there’s a metric that’s standardised we can all use to find our way around it. That is money. The world of nonprofits and grants and research organisations don’t really have it.
Some might, in a sort of tangential way. Like if one of the New Science scientists ends up getting a Nobel Prize a decade down the line, that’s a pretty neat indication that the organisation is power-law valuable, even if the rest of its scientists ends up only treading water.
(Side note: I don’t think one superstar and rest treading-water is the right modal outcome anyway, since geniuses come in genius clusters, and genius clusters evolve through internal positive feedback loops.)
And in order for this to be true in the world of giving grants, we’d need one of the grantees to become the next Beethoven, the next Linus Pauling, the next Gates, or the silver medal equivalent thereof.
But part of the grant-giving is lopsided today. They are results-oriented. They have identified the big wins they want and then are trying to fund various approaches to solve them. Whether that is safe geoengineering or AI alignment research, we already know the answers we want. We just need to get there. This would be like YC being pretty opinionated about wanting to see a new payments company emerge and funding variants until they hit Stripe.
It gets better at undirected research - whether that is nanoelectronics or molecular machines, or productising the benefits of crispr to all other diseases.
Now what if the world you're planning on bringing about isn't there already? That is, you're engaged in a creation exercise, not a selection exercise. This doesn't mean that you'll hit all winners, because it's not like it's easier to find that one special person who'll become Elvis than to create rock n roll! That would be just uncounted hubris.
You will end up co-evolving with the landscape. If you're lucky then the market grows bigger as time goes on, and you get to watch your winners grow ever larger!
The emergence of this new class of illegible institutions to drive change is a sea change in how we deal with driving change not immediately measurable by a dollar sign. The sheer number of ways we could declare success is in itself sufficient to confuse the issue, but us attempting to do this is fantastic.
Will this result in a hub and spoke model? Where the large existing institutions safeguard the larger parts of the effort and protect the downside, and smaller institutions act as variance seekers to capture the upside.
Or will this result in a changing of the old guard altogether? The way of smilodons and their enduring fangs? The creation of a minotaur, a hybrid designed to be effective but sterile, or just mammals popping forth after the terrible lizards had their due?
The next step is to create sparks to fan the flames and build the fire. The thesis the new organisations all have is that the world of ideas was being ignored in favour of the world of legible execution. In a pyramid world there's idea generation at the top but everywhere else there mostly is execution and oversight.
Regardless of the poetry the structure of the landscape shifts profoundly once we take the power-law seriously. Once we realise that the illegible is where the alpha is, we need temples of a different sort of help ask for the favour of the gods. Because if you don’t try to change direction, you may just end up where you are heading.