Welcome to the confession booth.
Stuart spent years giving away tens of millions of dollars to fix the way science gets done, and found that scientific research often doesn’t replicate. He now runs a think tank building a better and more open science. We ask him what he learned on the inside.
In this episode, we explore why philanthropy can take risks nobody else can, why it mostly doesn’t, and what this costs us.
Note: The Laura and John Arnold Foundation was renamed Arnold Ventures in 2019.
We talk about:
“The loophole in life”
What’s it like when everyone in the room wants your money? [3:50]
Philanthropy’s absent market pressures & the effects on its speed [5:31]
The unmoving, rock layers of process [7:06]
Strategy and its theater
When is due diligence an institutional comfort? [9:25]
Noticing people & patterns as a portfolio [11:00]
The limits of evidence-based giving
Effective altruism and the questions a randomized trial can reach [11:54]
How does natural gas trading translate to grantmaking? [13:50]
Failure
One grant he regrets [15:07]
Reputation is a fragile return [15:59]
Trad versus modern philanthropy
Warren Weaver at Rockefeller, molecular biology, and the green revolution [17:59]
Why adventurousness comes and goes in waves [19:00]
Why VC portfolio logic doesn’t apply to philanthropy [20:55]
AI and the pace of science
The limits on AI speeding up drug development [23:10]
AI & the regulatory arms race [25:25]
When to be patient and when to be fast in philanthropy
Timing curiosity & caution [26:14]
Latent ambition
Making space for illegible people [30:00]
How post-war government funding built modern science research infrastructure [31:41]
Metascience offices: who will fund them? [34:19]
From $50k to $5M: the story of the Center for Open Science [36:20]
Our favourite lines from this episode include:
“Some of the highest-impact grants that I made were not born out of some grand plan: it was just noticing people that are ambitious and don’t have the money.”
“We’re steering money towards things about which we can develop evidence: but that might exclude lots of other good ideas.”
“Philanthropy doesn’t have to make a return. It doesn’t have to respond to voters … They do take risk, but not nearly enough.”
About Stuart: Stuart Buck is Executive Director of the Good Science Project. As a Vice President at Arnold Ventures, Stuart funded renowned work showing that scientific research is often irreproducible, including the Reproducibility Projects in Psychology and Cancer Biology. As a grantmaker, he helped launch the Center for Open Science, Vivli, the Meta-Research Innovation Center at Stanford, the Stanford Center for Reproducible Neuroscience, the Yale Collaboration on Research Integrity and Transparency, and the Evidence-Based Medicine DataLab at Oxford. He was instrumental in creating the TOP Guidelines, the world’s most widely-adopted standards for scientific publication. His grantmaking was featured in Wired, the Economist, the New York Times, and the Atlantic, among many others.
If you want to go deeper, here are the references from this episode:
Essays:
Stuart’s “Metascience Since 2012: A Personal History” essay, which describes the grant he regrets [15:15]
Ruxandra Teslo’s “Intelligence is not the main bottleneck” [24:10]
Concepts mentioned:
David Shaywitz and colleagues in Nature Reviews Drug Discovery on why AI’s clinically relevant impact in pharma R&D remains “disappointingly limited”: thread here [23:53]
The metabolic pathways map Stuart describes [24:29]
Center for Open Science, founded by Brian Nosek in 2013 through a grant from Arnold Ventures [36:55]
Chesterton’s Fence [31:25]
The White House OSTP report “Science: A New Golden Age”, which recommends metascience offices at NIH and NSF [33:53]
Our own “A Taxonomy of R&D Orgs: What Is New, What Is Missing?” is cited in this report at footnote 65
Folk Science, an Expeditions Fund project we funded [29:19]
