Help science go faster
Why you should build scientific instruments
1. A different taxonomy of science
One of the things they don’t tell you about science when you get started is that there are a lot of different ways to do it. Your academic advisors assume you’ll sort yourselves out: the mathematically inclined go into physics, those with no regard for their personal safety go into chemistry. But beyond an unhealthy affection for integrals or a belief that free soloing should be safer than your day-to-day, how do you know what you should do? What you’ll actually like doing? A synthetic chemist and a theoretical physicist do such different work that you’d be forgiven for not believing they are part of the same scientific enterprise. You might have figured that out as an undergrad, but what you might not have realized is that the differences within a scientific discipline can be larger than the differences between disciplines. A physicist at a synchrotron and a polymer physicist live entirely different lives, even though they are both “experimental physicists.” Being an observational astronomer is completely different than being a string theorist, even though neither ever sets foot in a lab. The difference between all these physicists is primarily their approach to interacting with the world. When you start your training as a scientist, you choose an area of study and start developing an approach to interfacing with reality. There are many such approaches, all important, and all different. Here are some that come to mind: computer simulations, observational field work, experiment design, analysis of existing datasets, technique development, algorithms development, applied math, theory, operating large facilities to make complex measurements, building new instruments, and using existing instruments to make measurements. You can find many of these approaches across all scientific disciplines, and moving between approaches can be just as challenging, if not more so, than moving between disciplines. Choosing the correct approach is vitally important for your career satisfaction because it is a question of temperament. Choose well and you may find yourself constantly delighted by the things which first drew you to science. Choose poorly and you may find yourself aggravated and bogged down with daily drudgery. I’d like to map out all these approaches at some point, but for now, I’m just going to talk about one of the approaches I’ve focused on during my PhD: instrument building.
2. What are scientific instruments?
Science is a way of answering questions with empirical evidence. To do so, we need to make observations and build models within which those observations fit. In physics, our models are built mathematically, like the equations we use to describe gravitation and electromagnetism. A new observation can confirm or disconfirm existing models. We can use our senses to make an observation, or we can use a scientific instrument – a tool for making observations, often measuring things outside our perception.
In my research, I want to understand the fundamental principles that govern magnetic materials. I can’t see or feel magnetism, and even if I could, I’m not fast enough to sense it as it changes. To study these fundamental principles of magnetism, I built an ultrafast x-ray spectrometer that can measure the magnetic moment of multiple elements simultaneously on femtosecond timescales. I can “watch” the material’s magnetism change in the time it takes for light to travel the width of a human hair. This new instrument lets me see what I otherwise couldn’t, like a telescope that lets us see the structure of faraway galaxies, a microscope that lets us look at viral proteins, or an oscilloscope that lets us see electrical oscillations. We even use a common suffix appended to many instruments: -scope, coming from the Greek skopos (σκοπος), meaning the watchman, or the target.
If the goal of science is to expand our scope of possibility, the goal of instrument building is to expand the scope of science.
3. The scientific advantage of building an instrument
Imagine you’re Antoine Van Leeuwenhoek in 1676 and you just built a sweet new microscope. You’re ready to look at some plant fibers and minerals in more detail than anyone before. So you pop a sample in the microscope and – what the… what are those things? The Jains postulated the existence of microorganisms in 600 BC, but no one had ever actually seen them. The discovery of bacteria revolutionized our understanding of biology and medicine, and such a discovery was simply impossible with the unaided eye. Similarly, Galileo’s telescope allowed him to validate the Copernican heliocentric model and conclusively demonstrate that the earth revolves around the sun. Some of the most important scientific discoveries were enabled by the development of new instruments, and if you develop an instrument that allows you to see what others can’t, that significantly increases the likelihood that you’ll see something others haven’t. The primary advantage of building a scientific instrument is that you’re more likely to make a truly unique discovery with a unique instrument.
4. The career advantage of building an instrument
If you’re a young scientist at the start of your career, maybe you want to know – should I be an instrument builder? If you’re looking for glamor, the answer is probably not. The scientists who built CERN and LIGO are relatively unknown compared to theorists like Peter Higgs and Kip Thorne who won Nobel prizes based on the discoveries these instruments made possible. Of course, that’s not to say you can’t glamor from a new technique1 – the development of the polymerase chain reaction (PCR), and of the electron microscope both won Nobel prizes. More recently, Stefan Hell, Will Moerner, and Eric Betzig won Nobel prizes for developing super-resolution fluorescence microscopy. For a young scientist choosing their career, there is an important distinction here: developing a technique is not the same as building an instrument. A new technique requires a demonstration to prove its viability, but making something work well enough for a one-off demonstration has vastly different challenges from building an instrument that works reliably and repeatably. A new technique is often more glamorous than a well-implemented one, so if your aspirations are stardom, an exclusive focus on instrument building may not be for you.
Why then do I say that you specifically should build an instrument? Simply because the instrument building skillset is exceptionally rare and valuable. Building an instrument will turn you into a technical wizard – you will know the nitty gritty technical details and technical paraphernalia of your work in a way that few others do, and your troubleshooting skills will be first class. This makes you an extremely valuable asset to any team, since you will be one of their technical best, and labs will be happy to compete for your talent. Building an instrument will also help you develop a deep technical understanding that guides your scientific intuition, steering you away from technically infeasible problems that might otherwise waste months or years.
Though the benefits are numerous, you should pay close attention to the academic incentive structures while you build your instrument. Doing this as a graduate student can be challenging because academia runs on credit and credit assignment for instrumentation is tricky. There are effectively three ways in which you can get credit for a new instrument – (1) originating the idea, (2) building it, and (3) using it to make discoveries. Because most scientific instrumentation is capital intensive, you must be in a well-funded research group to have a shot at building a new one. Good ideas don’t mean anything if you can’t put them together into a real device, and that will cost somewhere from thousands to millions to billions of dollars. If you are a senior student in a well-funded group, you might be able to do some combination of (1) and (2), but new instruments often take so long to build that you probably won’t make many measurements with yours before you graduate. This means that the graduate students who come after you will take most of the credit for any discoveries made with the instrument since they will be the ones actually using it to make discoveries – your role was to design and build it. If the instrument is particularly complex, you might be part of a large team designing and building it, sharing credit partway up the author list. None of this should be much of a problem for you yet, since you can easily leverage your experience into a post-doc position. If you enjoy keeping your hands dirty and staying focused on developing new technologies, you can turn your expertise into a semi-permanent soft money position in a larger research group and avoid the administrative nightmares that come from being a professor. If you want to direct your own research group, you need to make a concerted effort to develop more broad, scientific aims and start making discoveries, not just cool toys. You should also consider getting involved in professional societies, serving as a journal editor, or as a conference chair. This will help you maintain your academic credentials, and these groups appreciate the nit-picky eye of instrument builders.
Outside of academia, large facilities like synchrotrons and telescopes are the pinnacle of scientific instrumentation and need instrument builders to design, build, and maintain them. The recently launched James Webb Space Telescope (JWST) is one of the most complex scientific instruments ever built, and if successful, it will usher in a new era of astronomy. Many companies sell instruments to scientists and instrument builders are the bedrock of their business. These companies make much of the day-to-day work in science possible. If your goal is to make as much money as your software engineering friends, these industry and national lab positions are usually much better compensated than their academic counterparts.
5. The long road to independence
If you’re a young scientist just getting started, being an instrument builder means you are chained to your advisor, even if you have independent funding. I suspect this is true not just for building instruments, but for any laboratory science with high capital equipment costs. I had an NSF Graduate Research Fellowship, which might give you a decent amount of independence in some fields because your advisor doesn’t need to pay your salary. But my experiment cost over $1 million to build. The $34k/year salary provided by the NSF with no research funding wouldn’t even cover 1/10th of my equipment costs even if I stopped eating and moved my bed into the lab. Such a fellowship did not provide any meaningful independence from my PI whatsoever, and I believe this is true for laboratory scientists in all disciplines. Capital costs are far too high for you to pursue your own research without a major grant, and you cannot win a major grant without a major track record. The best way for you to develop a new instrument as a graduate student is to convince your advisor that it’s worth it.
Is there an alternative? You could pursue a low capital line of work – if you are trying to become independent as soon as possible, this is certainly a good idea. If you’re an astrophysicist working with public datasets, you can be practically independent on an NSF grant! But working with public datasets, while interesting, is not building a new instrument.
Is there a way that young instrument builders could bootstrap themselves to success? I’m not sure. In the startup world, it’s easier to bootstrap a software startup than a hardware startup. Some software companies pivot to hardware after they’re profitable, but it’s not straightforward. Making the switch from a low capital field to a high capital field is challenging, and we shouldn’t just let young scientists who want independence leak out of high capital fields. One possibility for the future might be to offer a fellowship that includes equipment costs, to give young graduate student investigators more independence. Right now, the academic path to success for an instrument builder is to spend 6+ years doing what your PI tells you to do for a good recommendation letter, get a good postdoc, do what your PI tells you to for another 2-4 years, and then apply for a faculty position and a laboratory startup grant for $500k where you finally have enough cash to strike out on your own. Reputationally, you will be working with borrowed power, not owned power, for a long time.
Conversely, you will be building owned technical power. While building instruments may slow your career, it also strengthens it. If you can build a three-photon microscope for deep brain imaging, you will be highly sought after. Because new instruments are so advantageous for new discoveries, being one of the few people with the skill to build them is a powerful piece of leverage in a scientific career.
6. The future of scientific instruments
Science expands the scope of possibility while new instruments expand the scope of science, and the future of scientific instrumentation is bright. In the past decade we have seen the successful commissioning of Laser Interferometer Gravitational Wave Observatory (LIGO), which observed the first black hole mergers, and recently the launch of the James Webb Space Telescope (JWST). The particle accelerator at CERN confirmed the existence of the Higgs Boson, the particle that gives other particles mass. The best atomic clocks can measure the difference in the passage of time at two points only a millimeter apart. There are yet more instruments on the horizon: a myriad of companies are competing to build the first useful quantum computer. Lasers are becoming more powerful and will soon reach the Schwinger limit where electric fields are so strong that they produce pairs of particles from the vacuum of space. It is nearly possible to image the entire nervous system of C. Elegans simultaneously. Electron microscopes will be able to image atomic orbitals directly. Laboratory scale cold atom experiments will offer new tests of fundamental physics. Massive synchrotron facilities may be replaced with tabletop x-ray sources.
If you intend to be an experimental scientist, building a new instrument is the best way to hone your skills. There is no greater test of your technical prowess than designing and building a new instrument, and one of the greatest benefits is the development of the hacker ethos required to master the variety of disciplines involved in instrument building2. Once you’ve built your instrument, your options are excellent: stay in academia and leverage your expertise into the most interesting position you can find as a coveted technical expert, or move into industry as a much-better-compensated technical expert. You may discover something genuinely game changing with your instrument, and at the very least, you will have developed two incredible abilities: to turn your ideas into physical reality, and to enable something previously impossible. I cannot think of any abilities more profound and creative than these.
All that being said – this long discussion about building instruments might imply that the only important goal is to expand our perception. This is an essential aspect of science, but it’s not the whole story. Perception is not the same as understanding. Eric Jonas and Konrad Kording wrote a great paper explaining the difference, titled “Could a neuroscientist understand a microprocessor?” In the paper, they simulate a microprocessor at the transistor level and then study the system using modern neuroscience techniques. Even though they can study every electrical voltage at every point in the system (their perception is definitionally perfect, as if they have the greatest microscope of all time), they are entirely unable to explain the behavior of the system as a whole and how those voltages come together to generate the experience of Donkey Kong. Our measurements are only as useful as our paradigms, and instrument building must go hand in hand with theory building.
Instrument building is just one approach of many in science. The more of these approaches you can understand and keep in your pocket, the better, even if just to help you communicate with those who use different ones than you. If it sounds horrible, don’t worry. Building instruments isn’t the only way to do science. But it is one of the ways you can help science go faster, and if that sounds amazing, trust me that it is - and start building!
So many Nobel prizes have been awarded for new techniques that it might be the most reliable way to get one.
At the very least, you’ll dramatically strengthen general skills like machining, welding, CAD, electronics, and programming. Depending on your field, you’ll learn WAY more than this.