The tension of age

The sort of post that rubs everybody the wrong way

In many domains, people assume that more experience is always better and that there is a hard floor on how much experience you need to lead a large effort well. Sometimes those assumptions are encoded in explicit requirements on credentials like degrees or years of experience; sometimes they’re more implicit.

In other domains people believe that young people can be incredible leaders and accomplish shocking things. Today, you mostly find this mindset in Tech/Silicon Valley and adjacent fields like AI research; in the past fields like physics were dominated by the young. (Some people take this attitude to the extreme and assume that if you haven’t made it by a certain age you’re basically worthless.) 

Tanner Greer captures the contrast between the two mindsets well in a recent essay:

The technologists of Silicon Valley do not believe in authority. They merrily ignore credentials, discount expertise, and rebel against everything settled and staid.  There is a charming arrogance to their attitude. This arrogance is not entirely unfounded. The heroes of this industry are men who understood in their youth some pillar of the global economy might be completely overturned by an emerging technology. These industries were helmed by men with decades of experience; they spent millions—in some cases, billions—of dollars on strategic planning and market analysis. They employed thousands of economists and business strategists, all with impeccable credentials. Arrayed against these forces were a gaggle of nerds not yet thirty. They were armed with nothing but some seed funding, insight, and an indomitable urge to conquer. And so they conquered.

At Speculative Technologies, and especially in our Brains program, we find ourselves straddling these two worlds.

On the one hand, we have a lot of Silicon Valley DNA: we are convinced that people’s abilities are not defined by their degrees, the gold stars on their resumes, or years of experience.

On the other hand, we constantly interface with worlds that don’t share this attitude: academia, traditional philanthropy, and Washington. We can’t make people successful on our own: together with the ambitious scientists we work with, we need to convince governments or philanthropists that they will be successful in order to get the resources to execute. While there are individual exceptions, these worlds tend to prioritize credentials and experience.

(Certain government ARPAs, despite having many program directors who were no more than five years out of grad school, now has doubled that requirement: expecting program directors to have at least ten years of experience after earning their PhD — that would rule me out and I basically have one foot in the grave by Silicon Valley standards.)

At our Brains demo day in April, many guests from government and philanthropy commented on how impressed they were with the ideas and presentations for how young the fellows were, despite most of the fellows having experience levels that would put them in an upper percentile of experience at many startups. The other reaction was concern that their organizations would have difficulty hiring those impressive fellows because of their age and experience. 

Aside: These credential and experience requirements create tons of adverse selection effects, especially for government (ie. salary capped) employees who need to move to Washington DC. 10 years out of a PhD, talented technical people are generally at peak earning power and often have young children, making a paycut and relocation particularly unappealing. As a result, the demographic of ARPA PMs is heavily biased towards people who already live in DC (eg. government lifers) or folks doing one last adventure before retirement.

The result of interfacing with credential-focused worlds is the frustrating fact that when we’re accepting people to the Brains program or hiring program leads, we can’t just go on the quality of ideas and raw talent alone. It would be heartbreaking to work with a promising Brains fellow who then couldn’t get hired by anyone we work with because of their age. It’s similar dynamic to how seed stage VCs need to make investment decisions both on the inherent quality of a company and on how other VCs will perceive that quality. It’s uncomfortable to say out loud but it’s true.

External pressures aren’t the only reason for a tension about age, though. We’ve also observed that below some level of experience, people are less likely to successfully run a coordinated research program. 

Leading coordinated research programs demands a different set of skills from starting a startup – a set of skills that are hard to acquire without experience in the salt mines of original research. Grappling with systems underpinned by the laws of the universe for years without any certainty that you will succeed creates an intuition that is hard to replicate otherwise. This intuition is critical for leading research programs: thinking through research ideas, acting efficiently, and interfacing with other researchers. You don’t need a PhD to build this intuition — I would expect an 18-year-old who had been working in a lab since they were 13 to have it — but it does take time and often mentorship that are rare outside of a PhD. 

A successful research program leader needs to combine research intuition with the hustle and operational skills of a startup CEO. These skills are also hard to acquire without experience. Not impossible, but hard.     

All of this means that we turn away a lot of ambitious people who are straight out of college or even grad school. Doing this makes me (who has drunk the Silicon Valley kool-aid and would love to give everybody the benefit of the doubt, regardless of age or experience) feel like a giant asshole.

Perhaps this focus is secretly a boondoggle: instead, the real question is whether starting a whole program is actually the right next step towards someone’s goal. Often the answer is no, regardless of age!

So many things can start as amorphous projects before being shoved into an institutional box. It’s shocking how much research one person – especially if they are young and don’t have a lot of other responsibilities yet – can get done. Yes, eventually spending millions of dollars to build a lab for a new sort of model organism might be a critical next step, but is it the next step right now? Have you talked to all the people you expect to use it and deeply understood what would need to be true for it to be useful? Dug into efforts to cultivate new model organisms that failed? Hung out with people who have done it before? Doing all of the first things first will – ahem – not only build experience, but help make sure that when you do start a large program, you’re successful because you work on the right things and don’t waste time. 

To some extent, this post is meant as an explainer to the ambitious young people I talk to who seem awesome but just aren’t good fits yet. To some extent it is a confession of Silicon-Valley-coded sins. To a large extent, it’s a nudge for the world outside of Silicon Valley to be less credentialist!

One day, I want Spectech and the broader extra-academic research ecosystem to provide homes for all ambitious research misfits, regardless of experience. Until then, the tension will remain.

Thanks to Jessica Alfoldi, Adam Mastroianni, Stefan Leslie, Doc Brown, M Pang, and Antonius Gagern for reading drafts of this piece and giving feedback. 

Mini-appendix: Because it’s topical, you might ask the question: “What about AI? Won’t having the equivalent of decades of experience at people’s fingertips change the equation?” I don’t think so: a big part of the intuition that experience builds is for calling bullshit on approaches and ideas. Program leaders will need to call bullshit even on human-level AI.