Failing Faster in Science

Why fundamental science and medicine should start failing faster, a la the Lean Startup method.


I’m write now on a plane (and finished it up on the return flight), I’ve jotted up these notes this past week. (So this is totally a blog-rant by a dude couped up in a small cabin.) And it’s related to what I wrote in my senior thesis. From the little I’ve seen, it seems too much trust is placed in fundamental science and medicine. Science and medicine has progressed more slowly than I think it has to, and I think too many errors have been overlooked along the way. I think there are some lessons we can learn from both skepticism and the startup scene fail-fast concept to catapult us into a new error of “progress”. That said, who the heck really knows—so take this all with a grain of salt.

A Too Long Intro

Brief language clarification: When I say “fundamental science” I am referring to basic core principals, primarily theories that serve as backbones for science (and here, I include, physics). I’m talking about things at the level of Darwinian evolution to basic cell theory to wave-particle duality in quantum mechanics. In terms of medicine, I am referring to its average usage.

A few things precipitated this post:

Theory: My initial skepticism was bred over time, starting several years ago with my interest in health and politics. I wanted to learn from different sources (for both), and I basically became skeptical about most of the claims, but still see value in a more circumscribed (not in a truth-based or ideological sense). I then sought to learn more about skepticism, setting up a 1-1 in college (highly recommend Outlines of Pyrrohinism by Sextus Empiricus, Cicero’s Academica, Montaigne’s Defense of Raymond Sebond in his Essays). That class, especially, only furthered my skeptical-bent, showing me that I wasn’t just crazy. Or at least, there were other crazies that have thought like this from 2,000 years ago until today. My senior thesis was basically me talking about all this stuff.

History: I love history, it’s especially interesting to hear how people operated in such different paradigms. Pre-scientific revolution especially, what a different world. Yet it makes you wonder what are the chances that the paradigm we operate in is accurate? And beyond fundamental science, looking at the history of medicine (Biography of Cancer was an excellent read), it makes you wonder where we are even close to the mark. If there is anything that is right it’s that we seem to be almost always wrong. (Paradox and problem of induction.)

Current events: And now recently, specifically in relation to medicine, and other experimental-based fields, there is a pushback. Confirming the initial theory and historical observations, seeing in real-time how we are wrong most of the time we think we are right (broad generalization of course, but I've noticed The Economist has covered the issue the past few years. You can also check out this andthis and from the Edge's 2016 question).

Last disclaimer: While I am a bit tied into the startup scene, I think a lot of ‘startup wisdom’ and the culture in general can be a bit overblown. It has become commoditized (putting a ping pong table in the office to seem startupy misses the whole point and is incredibly ironic). But something that is nice that has come up is Eric Ries’ Lean Startup fail-fast idea (for a quick synopsis, see “Why the Lean Startup Changes Everything” by Steve Blank.

Fundamental Science

We will ultimately never have a complete understanding of the universe. Our own understanding of the world is bound by our human minds that come with its own peculiarities.

There is no objective ’truth’. The very fact that we are human (and not some sub- or super-spieces alien). We are fundamentally limited to our own neurology in ways that we will never understand. (Even if that neurology can be ‘hacked’.) What there is only what we know relative to what we can understand.

We need to approach fundamental science with the lean startup model. These pulitzer prizes and science-hero worship of people like Einstein, and tesla, while great minds, the words they uttered are just as bailable. It has taken us years to get over the fact that Einstein might have been wrong. Of course we need to build on people’s work. But I would think it might have taken a lot longer for Einstein to build on Newton if he was entirely convinced that Newton was right and had no holes.

At least for the real basic stuff learned in science in high school and college, it seems experimental mistakes and mishaps often lead people to realize the fault in a theory. And that’s all fine and well. But I wonder how many others overlooked those errors, rather than being open to a possibility of a flaw in a theory.

No one is perfect, no theory is perfect. And that will always remain true as long as we are humans and not gods.

That’s not to say to reject frameworks or paradigms out of hand when someone like Einstein comes along. But the acceptance shouldn’t be accepting that it is true, but accepting that this is the one that appears to advance things, and is currently most useful.

While a startup isn’t find the solution to fundamental science, I do think that the approach in the lean startup model, acknowledging at a basic level that the idea and product will evolve, is a healthy one. It recognizes that the current product, while might be at its current best given what you know, can always be improved. And the goal is to improve and iterate with feedback as quickly as possible. Instead of assuming the idea/product is the ‘right’, a basic assumption of the lean startup model is that the product will always have flaws, and thus always have too m for improvement.

And it brings a more humbled approach. No one uses the product thinking “this is it”, and nor should we think that with scientific theories.

New Paradigms

Since thinking more about paradigms and science in college, I would like to apply this rule. Take any statement in science, and ask this question: What are the chances that this theory holds true and is understood in this way in 1,000 years should science continue to progress? What about 10,000 years, or n amount, as n approaches infinity?

You get the idea. I can’t think of any theories that would pass the test that at least for me I don’t think would be altered in the future. (How so, is another matter.)

One of the theories I hear about most is evolution. Now, I grew up in a home where evolution was a given, and I have a sister getting a phd in evolutionary biology, and my mom was/is a science teacher, teaching evolution. I’ve been exposed to quite a lot of that science since an early age, and there’s no doubt a lot of value there.

But more often than not, I hear evolution used as a crutch to explain any sort of behavior or trait. Why does X do Y? Well…and then comes some sort of imagined narrative. It seems like the explanatory limits to someone operating in the evolutionary frameworks is someone’s fictional narrative skills. No doubt that this kind of pop-science isn’t touted by everyone, but I think the scientific community has given too much causal power to a theory that isn’t even three hundred years old. I say that with the upmost respect for those in the field, and doing good work (like my sister!).

And the more degrees and awards someone has, we should be more and more on the prowl for the cognitive biases and errors in judgement all humans have.

Yet, it’s almost heresy to suggest that evolution might be wrong. I think that’s unhealthy. And the real bummer is that I don’t have the answers, the knowledge, or the tool to know what the answer is or where to begin. The PhD’s that have spent years studying this stuff are those most equipped to tackle this, but they’ll be ostracized, likely lose their funding, if they stray too far from the path.

A startup that held onto an old product for three hundred years would be long dead, probably dead after 6 months and the first round of funding dried up. Capitalism is in that respect a lot less forgiving. Our theories should fail fast, and improve—perhaps there just isn’t the competition required.

It is very hard for me to believe that some of our basic scientific theories will be around in a few hundred years. I think the concept like evolution—and how it is being applied (I took a rather advance class in cultural evolution in college)—is fascinating and super helpful. But it’s a stepping stone, not the holy grail. (As an aside, I don’t think we’ll probably ever get to some sort of final “figured it out“ aspect for the world. Especially with science’s reductionist approach. But I would love to be wrong!)


But this is true not only for broader scientific theories, but in practical medical application as well. Fortunately, medicine progresses a lo faster than basic theories, but still not fast enough. There are many today who are undergoing the practice of blood-letting. Well, perhaps not exactly blood-letting, but some sort of practice that is likely quite harmful.

The notion that medicine is an imperfect practice isn’t a novel idea. Most people go through life with experiences that loud and clear scream of evidence that medicine is far from perfect.

Yet, it took way too long, for example, for the scientific community to get over the idea that antibiotics aren’t the be all solution when you are sick.

When I was a kid, the doctors wanted to prescribe antibiotics to me for all kinds of things, and my mom wouldn’t let them. At one point, I think a doctor might have threatened my mom. Fast forward 10 years, the doctor was wrong, and we are realizing how much overperscription there was.

I think in this case, there are probably a few very difficult matters at play. There’s the legal consideration that being overcautious won’t get you sued, but being under cautious (‘negligent’) can. Then, similar to the fundamental science case, it’ll be hard to start off as a regular doctor when you don’t assume basic assumptions in the field. And then there is the massive R&D behind drugs that are no doubt pushing a lot of the drugs themselves.

Unlike scientific theories that require more intellectual leg work in failing fast, for medical application it requires a boat load of capital. I don’t know of any successful small startup that is a drug development company (though I think YC accepted one working with AIDS), and part of it might just be it is far too costly with the current FDA requirements. Perhaps you simply need an obscene amount of money to put together a satisfactory study. (Although I know a friend working on a startup in the drug trial space, and there’s no doubt going to improvements on that end.)

No doubt a lot of it comes from caution, and the checks in place are well meaning.

But we are going far to slow. And honestly, I can’t imagine in fail-fast, (rather than grow really big and fail later) approach things could be that much worse than some of what’s been out there in the past. But then again, in medicine, “fail” can be fatal, even at a small scale.

Whether in drug development, or in prescription, though—even if it takes time to develop, we should recognize the faults a lot sooner and iterate. And do our best not to get stuck in old paradigms.

If you wanna get a kick of just how off-base medicine can be, any history of medicine will do—Biography of Cancer is a great one.


These are just airplane scribbled thoughts on science and medicine. We have a long way to go, what is excited. Let’s fail-fast, learn, and progress more quickly. Since I had an absurdly long-winded intro, here's an inversely short conclusion :)

I was hesitant to post this as it comes off as super negative, and I’m very much ignorant of many of these matters. So count this as a plane rant from being cooped up for 10 hours.

Rock on,

If you enjoy this kind of stuff, click here to stay in the loop.