One of my favorite people to listen to (or, in this case, read) is Patrick Collison – widely known as the CEO of Stripe.
Patrick is one of the most thoughtful people I’ve heard speak on various podcasts, and his ability to hold complicated networks of bidirectional causality and multiple layers of abstraction in his head at once – while footnoting himself extensively in his normal speech – is a feat to be admired.
Patrick recently published an article in The Atlantic called “Science is Getting Less Bang for Its Buck”.
And, if you want to go really deep on it, you can listen to my favorite podcast (EconTalk) where Russ Roberts and Patrick break down the article in more detail.
The fundamental thesis of the article is that the individual productivity rate for scientific discovery is going down – meaning that it is requiring more people and more effort to produce scientific discoveries that are also seemingly of lower quality than in the past.
If we assume that this is generally true (and – as a non-expert in this area, I find the case made quite compelling), there are a variety of potential causes of this phenomenon.
I see significant parallels to long-term athletic development and business growth; “What got you here is not what will get you there.”
After athletes have used up their “newbie gains,” it requires much more intelligent and dedicated training to continue to make progress.
Athletes also tend to spend a lot of time on performance plateaus where they feel stagnant in their abilities.
Then, they suddenly have breakthrough performances where they reach a new echelon of capacity.
If I were interested in speaking in hackneyed and trite phraseology, I may say something like “we’re building a bridge not a road.”
Meaning that, when building a road, each piece of incremental progress results in increased ground covered. However, when building a bridge, incremental progress does not necessarily result in increasing the distance someone can travel.
Until suddenly it does when the bridge is connected and you’ve suddenly connected two landmasses.
Does this mean that productivity rates will ever return to what they were? Not necessarily.
At some point, we may reach new innovations that open up entirely new layers of abstraction on which individuals can begin making discoveries (think of the confluence of factors that facilitated Instacart being a multi-billion dollar company like smartphones, GPS, and a cultural understanding of the “gig economy” – as opposed to Webvan which was a notorious dot com bust posterchild in the early 2000s).
What could these new layers of abstraction be? Certainly, the ability of machine learning (ugh “AI”) could facilitate an ability to model the complexity of emergent systems in biology, weather, economics, and social sciences in heretofore unseen ways that allows us to get a stronger foothold into the inherent chaos (meaning “extreme dependence on initial conditions”) of these systems.
If I had to put my money on any sort of path forward for major scientific discoveries, that’s where I’d be looking.
Now, that doesn’t mean that the individual productivity rate will rebound – it may in fact continue to be trending downward, but the possibility of new layers of technology and an understanding of how to model chaotic phenomena may result in individuals being able to harness these new technologies and bust through a productivity “plateau” [or potentially “downward sloping steppe” based upon the data in Collison’s article].
And, if we couple an increased productivity rate with Stripe’s mission of creating an economic infrastructure that brings the entire world’s population online in such a way that there is a much larger pool of innovators in a position to potentially make scientific discoveries, we may just save the world after all.
Well, until the universe wheedles, whimpers and whines to an untimely heat death.
