In Founding vs Inheriting, Balaji Srinivasan lays out an argument for a meritocratic future that goes roughly like this: the act of building allows one to understand the fundamental nature of a problem, and thus in the long run builders will win out over those who simply manage existing institutions.
This is, I believe, largely correct, however his theory of progress underappreciates status as a fundamental north star of human behavior. In my view, models of institutional incentive design more accurately predict failure than his founding versus inheriting framing.
One of his core examples is the New York Times, an inherited institution according to his theory that should have been unable to innovate in a time of crisis. The Times, however, completed its transformation to a digital-first publication in 2020 while drastically reducing its reliance on advertising revenue and substantially increasing its subscriber base. The general theory of media tech is that the advertising model is antithetical to quality journalism, and that subscriptions or revenue derived from consumption is the healthier route.
Consider the most recent tech darling in the publishing space, Substack. Their technology and business models are relatively quotidian: email lists and subscriptions. What accounts for Substack’s success is not that it’s a particularly good publishing company, rather it strips away the unappealing parts of traditional publishers and sells access to high status individuals.
Despite being owned by one of the greatest builders of all time, the Washington Post didn’t out-innovate the Times through 2020. From the outside, there’s little that distinguishes their journalistic products, nor their underlying technology. What Founding vs Inheriting misses is that these organizations don’t sell reporting — they sell prestige and status, and by extension trust. The pattern we should desire to break is not the inheritance structure of the institutions, which at any rate are run by non-familial editors. Rather we should aim to reorient consumers towards a different idea of truth if our aim is to create dynamic institutions.
As an aside, it’s not clear that we categorically want dynamic institutions. Family run businesses are more likely to survive through economic shocks. We may want a kind of stability and long-termism in some industries best encouraged by family ownership.
It may seem cynical to think that the Times sells status and not journalism, but I think this is a necessary function. For instance, I know nothing about the Suez Canal, and as far as I know none of my friends are experts either. So when a ship wedges in and causes an international disturbance, I have to trust that somewhere in the content generation chain I consume, somebody has done some reasonable amount of diligence on the matter. How am I able to correctly pick an expert to trust if I know nothing about the domain? The Times provides a handy shortcut solution by wedging itself, like the Ever Given, into the stream of other high status institutions that fit my intellectual countenance. The East Coast institutions are superb at this game.
Suppose, though, if we wanted to reorient my notion of status away from Pulitzers and towards journalistic truth. Does this necessarily require a builder at the helm, as Balaji argues, or can we modify the status structure to encourage better behavior? Suppose as consumers we were to demand our news institutions publish “what we got wrong” updates, in the same way software engineers run sprint retrospectives. We should expect that institutions concerned with truth should have a truthy feedback loop, and what we would like is for the strength of this feedback loop to confer status. To an approximation this is the function of a journalism degree (Columbia Journalism School is another example given as an inherited institution), which seeks to train future journalists in a flavor of critical inquiry. Public trust in this process, however, has eroded over time and seemingly along partisan lines. This is a problem widely acknowledged in the news industry, and organizations are taking the study of the problem seriously.
A read-write truth seeking culture, to riff on Balaji’s “read-only” metaphor, is dependent on a feedback mechanism. We have a few examples of how this might be done: peer reviews, experimental replications, retrospectives, the Hegelian dialectic form. What we should aim to do is raise the status of feedback and the norms of science.
The broader point is that old institutions are not ipso facto stagnant. DARPA is run by a political appointee and as an institution hasn’t appreciably evolved, and yet this massive $3.4 billion government organization has been able to fund astounding breakthrough research because of its unique structural meld of academia and VC. The design of the institution itself can make or break its success, and if we want dynamic institutions then we should design for variation and experimentation, but disruption need not be the only target. Famously, Toyota designed the NUMMI plant collaboration for quality production and then hired the same workers who under GM were among the lowest performing in the country. The result? A reoriented incentive structure with the same staff became the stuff of management legend.
