Spectech Lessons and Updated Hypotheses from 2024
Things we got right, things we got wrong, and things we didn't know we didn't know
In the spirit of being an institutional experiment, we wanted to share some of the key takeaways from 2024: what we got right and what we are updating from our hypotheses going into the year, some lessons from 2024, and then outline some big hypotheses going into 2025.
In addition to our actual outputs, we hope that through the feedback loop between meta-scientific ideas and executing on those ideas, we can pave the way for other institutional experiments. I realize that each of these points wants its own memo unpacking it. I hope to create those over the coming year.
Below is a summary of our takeaways, divided into hypotheses from last year that we want to double down on, beliefs that we have updated or were wrong about, and new hypotheses going into 2025. I expand on each one further down. (As a gentle nudge that we are subscriber- and donor- supported, the updated and new hypotheses are paywalled for now.)
Double Down:
There are far more ideas that don’t fit into existing institutions than a single organization can handle.
Governments run into fundamental tensions around ambitious research.
Working with Bureaucracies is incredibly hard for a new organization.
Materials and manufacturing are an incredibly impactful place to focus for new institutional models.
Universities have developed a monopoly on pre- and non-commercial research.
Exclusively working with external performers in the 21st century is severely limiting.
Updated/Wrong
Corporate research has been gutted.
It is harder to fund 1-off ambitious research programs in materials and manufacturing than we thought.
There are a lot of subtle annoying things about nonprofits.
New
Universities need to be unbundled.
PI-based funding is holding back progress.
As an organization, we need to figure out how to use AI well.
There are a lot of systemic things widening the ‘valley of death.’
In many domains, IP is net negative.
Double down
I think most of our hypotheses going into 2024 were spot on. I’m not going to touch on all of them, but I’ll highlight the most important ones that I want to double down on and the additional evidence for them.
There are far more ideas that don’t fit into existing institutions than a single organization can handle. Throughout 2024, we constantly ran into people with ambitious ideas that don’t fit into normal institutional boxes: some of them were stuck, some of them were struggling to cram their ideas into normal-shaped boxes, and a small-but-increasing number are forging their own path. We acted on this one by running the Brains Accelerator, publishing the Research Leader’s Playbook, and starting a growing network of people running out-of-the-ordinary research organizations.
Governments run into fundamental tensions around ambitious research. We spent quite a bit of time exploring possibilities for government funding and were mostly disappointed: everything from the important work being too use-focused for basic research grants and not specific-application-focused enough for applied contracts, to falling outside rigid categories, to grants that can only go to accredited universities or startups. Furthermore, working with the government eats time that a small team can barely afford: it takes many hours to even sign up to be eligible to receive contracts and we literally had the discussion of “this contract isn’t big enough for us to go for because it doesn’t cover the additional person we would need to hire to handle the reporting.” These government constraints are coupled to another lesson from last year that I want to double down on: Working with Bureaucracies is incredibly hard for a new organization.
Materials and manufacturing are an incredibly impactful place to focus for new institutional models. 2024 convinced me this is more true both on the impact piece and the need for new institutional models. (See here and here for much more on that.) Based on these pieces, I now run into so many people with ambitious ideas in materials and manufacturing who are like “yup, neither academia and startups are any good for making these things happen.”
Universities have developed a monopoly on pre- and non-commercial research. 2024 hit us over the head repeatedly about how difficult it is to avoid interfacing with universities when you’re doing any sort of research that isn’t obviously a business and how much bureaucracy and weird institutional incentives that brings. Experts? University. Lab space? University. A government or foundation wants to set up a new research-adjacent initiative? It will inevitably be anchored to a university. This needs to change. Two anonymized anecdotes on this front:
I talked to [a state government] that is working to create a new [hype-y topic area] hub. The hub is of course run by [a prestigious university]. I asked about when I should expect updates. The state official said “no idea, we’ve been waiting months on the university.”
An acquaintance who had spent years working in several departments of the US federal government went to [a prestigious university] to start a new initiative there. He rage-quit a few months later because it was even more bureaucratic.
Exclusively working with external performers in the 21st century is severely limiting. We did a lot of work trying to coordinate work between different organizations. I need to start tracking how much time is wasted haggling over non-existent IP, trying to align incentives and timescales, all of which is compounded by how so many organizations (even startups!) are basically run by lawyers.