I go to a lot of conferences.1 I even go to a reasonable number of conferences on the science of science. And, yes, I am about to suggest more conferences that should exist.
Conferences are key to the practice of science. There is something magical about conferences that can’t really be captured in (yet another) Zoom meeting; it remains hard to mingle on Zoom. I’ve met collaborators at conferences; sometimes it feels like the only time I reliably get to go out with my friends is when we are all at the same conferences. That is not to say that conferences have no downsides — they are expensive, time-consuming, and often not very accessible2 — but I think the positives make them invaluable in the current scientific ecosystem.
But this isn’t really a post about how great conferences are, or to run a great conference. Rather, I think there are two conferences that should exist and do not.
I outline them below, as well as why they’d be valuable.
Salon de Refusés
High-risk science does famously badly at grant review. In biology, the NIH refused Katalin Karikó’s grant applications to study mRNA multiple times. In physics, LIGO was viewed as an expensive long shot and nearly died many times.
But high-risk science can also provide outsized rewards. Katalin Karikó’s work on mRNA underlies the Moderna and Pfizer COVID vaccines; LIGO was the first instrument to detect a gravitational wave. It is the risk-reward tradeoff – only if you are willing to take big swings can you make big innovations.
I see some parallels to the beginnings of Impressionism. In 1863, only a certain type of art was deemed acceptable by juries to be shown at the Paris Salon. New, riskier art was rejected; the Salon was not here for innovation.
A group of visionary artists staged a revolt. They demanded the right to show their paintings as well, and they too held a Salon. The Salon de Refusés — the ‘Salon of the Refused’ — focused on new, revolutionary styles of art and featured artists like Courbet, Whistler, Manet and Pissarro. Their “rejected” art is now vastly more famous and influential than the accepted paintings.3
Science should have a similar Salon. We should have a gathering for high-risk science, the kinds of projects that have a 90% chance of failure but a 10% chance of wild success. Such a Salon would bring together risk-taking scientists with risk-tolerant funders, and try to find a way to make more LIGOs and mRNA vaccines happen.
Logistically: at the Salon, scientists would outline their “refusé” idea: the core question, data so far, and the key barrier that spooked funders. Attendees would network and discuss others’ proposals. Funders interested in moonshot ideas would also attend. Together, scientists and funders could discuss how to make a project happen - be that full funding, potential low-cost pilot studies, or cross-disciplinary partnerships.
In the long-term, we hope that Salon projects would lead to new, exciting science. I think it would probably be a bit much to ask that a single conference save the career of the next Katalin Karikó, but it could begin to show that high-risk science can also produce high rewards.
This would encourage other funders — outside the Salon network — to take on more risk.
FailureCon
Laboratory science involves a lot of failure. You try a new method; it doesn’t work. You tweak a protocol; it still doesn’t work.4 Much of this failure is never published; after all, there are very few Nature papers about all the methods that didn’t work.
But sharing knowledge of what doesn’t work can be as valuable as sharing knowledge of what does. Without knowledge sharing, one can imagine different labs all trying the same thing, and all independently finding out it doesn’t work. This is… perhaps not the most efficient way to do science.
FailureCon would attempt to take failure out of the shadows. It would be about presenting all the things that didn’t work — and what one can learn from those failures. One might even include the things that sometimes work - the experiments that only work from one set of cells, or only one grad student can get it to work.
Ultimately, by discussing failure, FailureCon would seek to shorten the “learning loop” from “WHY DOESN’T IT WORK” to “ah, that’s why it didn’t work”. This isn’t entirely a new idea; USAID did “Fail Festivals” about failure in international development, but I’m not aware of this happening much in the hard sciences.
And certainly, presenting on all the science that didn’t work is not without risk; I suspect many early-career scientists would be hesitant to sign up for a slot about the ways they’ve failed. In order to derisk this for others, FailureCon would be headlined by notable scientists able to talk about their failures.
Current and former DARPA program managers would be good candidates as speakers. Since DARPA expects many projects to fail, it’s hardly a career risk to admit out loud that many projects did, in fact, fail. Later-career scientists, with tenure, might also be able to speak on some of the things they tried that didn’t work. They might also be able to share ways that failure early in their career sparked other ideas.
Failure leading to later success may seem far-fetched; it’s rarely that useful if, say, you cause a “small explosion” in your lab.5 But failure — or perceived failure — can also illuminate new facts about the world. Let’s take Mary-Claire King as an example. As a graduate student, she was convinced she’d failed when her comparative protein analysis showed 99% genetic similarity between chimps and humans. Her advisor suggested that perhaps her work was right. Her “failure” had shown that chimps and humans were much more closely related than previously believed— more similar, in fact, than many species of mice are to one another.6
When an experiment “fails”, then, it is sometimes because our model of the world was incorrect. The failure itself is the scientific breakthrough — that something about the model that led us to think this would work is incorrect.
FailureCon would be at least an attempt to shift the conversation towards the Thomas Edison way of doing science — “I have not failed. I’ve just found 10,000 ways that won’t work.”
Thanks to Sarah Constantin and Jean-Paul Chretein for their brainstorming help during the RenPhil retreat; thank you to Eric Gilliam and Andrew Gerard for their comments.
Possibly too many.
I am immunocompromised and go to a lot of conferences. I mask and I spend a lot of time avoiding people that are visibly sick. Even with precautions, I still get sick from conferences, and then I spend three weeks being catatonic on my couch. Not my favorite way to spend three weeks, tbh.
The best-known piece displayed at the Salon de Refusés was probably Manet’s Le Déjeuner sur l’herbe.
Ask me how many mirrors I’ve burned holes through. Actually, please don’t, SLAC might revoke my qualification as a Qualified Laser Operator.
The citation here is illustrative rather than exhaustive; as far as I understand, many chemistry grad students have accidentally blown something up.
This story is related in A Flower Traveled In My Blood, which references King’s work on maternal DNA inheritance.
love love love. Unlike you, I kind of hate conferences. But I would gladly attend both of these!