Forage, Then Feast

Around the same time as I decided I wanted to make a living in biotech, my friend Jack decided he wanted to make a career out of designing car engines. He was also a mechanical engineering major, so he was already on exactly the right track. He also made better grades than me and had a few years of experience in the university auto club, so he was a much better qualified applicant for his preferred job. Better yet, the university had already set up internship programs with several major auto companies, so he was using an existing system to get his foot in the door. When he got an internship, he didn’t even have to start out working for free like I did, but instead started at $20 per hour!

When Jack arrived for his first day of work, he quickly realized that his competition for full-time positions after graduation would be stiff. There were a couple dozen other interns there, and they all had the exact same career aspirations as he did. So he worked his ass off, putting in twelve hour day after twelve hour day of doing exactly as he was told. His boss was a tyrant, but he was the man to please. So Jack kept toiling, and gradually became a master of undergraduate-level automotive engineering. By the time he finished his internship and graduated from college, he had become the model applicant for exactly the job he wanted.

When it came time for me to apply to PhD programs and look for a laboratory to join, I was overwhelmed by all the options available. There are scores of high-quality research universities in the USA, and most of them have several laboratories doing some sort of diabetes research. I found about 300 labs I could imagine working for. Rather than try to narrow it down based on whose papers I found more or less interesting, I wrote a personal letter to the principle investigator of every single relevant lab with my resume attached, expressing interest and asking for a job. Most of them never replied and a few told me not to email them ever again, but a few dozen labs were quite receptive.

Once I’d emailed virtually every diabetes researcher on the continent, there were twelve universities that had at least two investigators that might be interested in hiring me. Since I wasn’t really qualified for bio-related PhD programs, I applied to all of them, just in case. As it turned out, I got really lucky. Eleven of the twelve PhD programs I applied to accepted me. I interviewed with ten of them (one didn’t want to buy my flights), and talked to five or six potential advisers at each institution.

My friends and family thought I was being silly (“You got into Yale. Why even look at the other places?”), but still I looked at every possible option. Some universities seemed much more enthusiastic about having me than others, some potential advisers were much more personable than others, and the best options weren’t necessarily what I would’ve expected in advance. At the end of the day, I wound up taking a position at a university that let me rotate through several exciting labs and that offered a generous stipend in a fairly low cost-of-living area. It took a lot of thankless, unpaid work to email 300 potential advisers before I even started looking at graduate programs, but in the end it paid off. Now I can get a high-paying job pretty much wherever I want.

Meanwhile, Jack started his hunt for a permanent position knowing exactly where he wanted a job. He’d already had a job at the company he wanted to work for, but he wanted to move up in the ranks. So he applied for a full-time position at the same auto company where he did his internship, and also applied to a few other auto companies just in case. Unfortunately, all those other auto companies were mostly interested in hiring people who already had some sort of connection to the company (such as former interns), and he wound up only getting an offer from the same company he interned with.

The salary was lower than he would’ve liked, he didn’t much care for the city where the company was based, and he would be working for the same tyrannical boss as during his internship. He didn’t have any other options, though, so he accepted the offer and he’s still working there. Jack may be designing engines for a living, just like he wanted to do, but he doesn’t have much potential for future career growth, he’s not making as much money as a top-of-the-line graduate from a top-of-the-line engineering program would usually expect, and he’s working twelve hour days in a city where he has virtually no social contacts. And he has no other realistic options.

I think that there’s an important lesson in this story. Maybe I just got luckier than Jack, but personally, I think that there’s more to it. The conventional wisdom of “pick one thing and get really good at it” is only good advice if the one thing that you’ve picked in advance actually turns out to be a good decision. There’s no way of knowing that in advance. Sometimes it’s better to try before you buy. This doesn’t mean that you should waffle over your decisions or never commit at all. To the contrary, applying these principles requires decisive decision making. Once you pick a path, you have to go all in and all out. First forage to get to know your options, and then once you’ve found what you’re looking for, feast to your heart’s content.

Moreover, I’m not the only one to come to this conclusion. Far from it. In the book Biodesign: The Process of Innovating Medical Technologies, Stefanos Zenios, Josh Makower, and Paul Yock describe the lessons that they learned from their own serial entrepreneurship and from a wide variety of case studies into the practices of successful biotech innovators. They synthesize these lessons into a cohesive, formulaic process for biomedical innovation.

One of the key features of their formula is that the successful innovator must devise several possible solutions to any problem that they hope to solve. As you work through the details of these several solutions, you’ll inevitably find that some are more difficult to execute than expected, some fail to actually solve the problem for reasons you never would’ve seen in advance, and some work just fine but turn out not to be any better than existing technologies. And if you’re lucky, one of your possible solutions will work really, really well. Devising several potential solutions effectively works like buying multiple lottery tickets. Each idea has a limited downside (the work required to design and/or prototype), and virtually unlimited upside.

The same principle is also widely applied in software development, where it’s called “A/B testing”. The basic premise of A/B testing is that you create not one, but several versions of an algorithm or website, and then deploy them all and measure the results. Some algorithms inevitably wind up yielding better results than others, and it isn’t always predictable in advance which algorithm that will be. It takes a lot of work to devise multiple algorithms for the same application, but as with the biodesign process, and as with the job hunt stories I told above, the costs of exploring multiple options up front are often far outweighed by the benefits of an unexpectedly good solution.

For a real-world example of A/B testing, I need to be able to measure the motion of red blood cells in capillaries in one of my current research projects, and doing all the measurements by hand would be far too slow to obtain a meaningful sample size. So I created a particle tracking algorithm, a cross-correlation based motion detection algorithm, and also an algorithm that would measure the number of “blips” across a particular vessel cross section. I expected in advance the particle tracking approach would work the best. As it turns out, the cross-correlation motion detection was far superior. The cross-correlation algorithm also turned out to have applications for another project that I didn’t even know I was going to be working on. I bought multiple lottery tickets, and I won.

The strategy of buying multiple metaphorical lottery tickets is also the foundation of venture capitalism. Venture capitalists first investigate far more start-ups than they actually invest in, despite the fact that every company they don’t invest in is a net loss of their time and effort. Even then, 90% of start-ups fail. But there’s a fundamental asymmetry here. You can’t possibly lose more money than you invest, but even one successful startup has the potential to wind up yielding hundreds of times more money than you invested. As with my job-hunt strategy, as with biodesign, and as with A/B testing, venture capital is profitable because it’s designed to harness volatility and uncertainty. You can try as many times as you’d like, and you only have to get lucky once.

Of course, once you’ve identified a winning option, you have to go all in and all out. I got to dabble with several laboratories and exchange emails with several hundred more, but in the end I could only do my PhD in one lab. FDA clearance is difficult and expensive enough that you’re not realistically going to be able to bring several medical devices for the same application to market in parallel. Turning an algorithm into a comprehensive software package requires a whole lot more work than just testing it to see how well it works. Once you’re done foraging it’s time to feast.

It’s also worth noting that the “forage” part of this process usually comes at a significant cost. I spent hundreds of hours working for free for labs that never gave me any sort of official credit, and scores more into emailing professors who never even dignified me with a reply. Every engineering analysis performed on a failed medical device concept, and every algorithm written that subsequently fails costs the inventor quite a bit of time and money. Venture capitalists lose enormous amounts of money on investments that go bust through no fault of the investor. The sort of opportunistic non-commitment I’m promoting in this essay shouldn’t be confused with cowardly or lazy non-commitment. Instead of expending your energies on nothing, expend your energies on everything, until you’ve found something worthy of all the energy you can muster.

Now that I’ve made a broad case for opportunistic non-commitment, let me spell out what it means in a concise, formulaic fashion: 1) Always begin with multiple parallel paths to success in mind. 2) Pursue every plausible path to success as vigorously and thoroughly as possible. 3) Decisively abandon the paths that yield poor results, and don’t let it feel like a failure when you do. 4) Once you identify a winning path, drop everything else and go all in. 5) Repeat indefinitely for each new goal in life.

Finally, I want to show you that I put my money where my mouth is. Right now I’m finishing one episode of my career and preparing to advance to the next. My overall mission is to reduce the miseries of chronic diseases. I may be able to do that by advancing medical science, so I’m applying for grants like crazy. I may be able to do that by developing new technologies, so I’m building business relationships with several exciting startups. I may be able to do that by persuading people to live healthier lifestyles, so I’m writing a blog with a mix of attention-grabbing (to gain followers) and genuinely insightful (I hope) posts to practice my persuasive writing. The simple fact of the matter is that each of these endeavors costs me a great deal of time and effort, and yet the vast majority will fail.

But I only have to get lucky once.