Introduction

A lot of EA advice seems to presume that the reader is already a top-tier student or young professional who just needs to have their endless font of potential pointed in the right direction. This is not most people. This is not even most EAs. This is certainly not me.

I was not a particularly impressive student. I left my undergraduate institution after five years with a shaky B average and no publication record to speak of. My existing career plans were torpedoed almost immediately after graduating when I began to get involved with EA. By the time I figured out what I wanted to be doing with my life, I had to claw and scrape and jump on whatever bus was heading vaguely in the direction I wanted to go.

I wasn’t in that bad shape. I could code, kind of, and my mediocre state school had an excellent physics department. But I just didn’t look that good on paper. 

If you are an undergraduate or recently-graduated student and this sounds like you, and you aren’t really sure where to go next, consider working as a research assistant at a university. 

Why might I want such a role?

• These roles are unusually easy to get, relative to how much they can benefit you. You are not expected to have qualifications.
  • For many roles, simply knowing basic programming is enough to get your foot in the door; a few weeks of self-study in Python will probably suffice.
  • There is no lengthy application and interview process. Every research role I’ve been in, I got simply by directly contacting a PI I was interested in and asking to work for them.
• If you already have an undergraduate degree, many pay decently well. 
  • I made well over the median American income as a research assistant simply by having a bachelor’s degree.
  • It isn’t going to be amazing money, but it’ll be enough to live on.
• Given the ease of access, these roles are a cheap test of fit.
  • If you’re considering a research career, this is a must. 
• Simply being in the university ecosystem will gain you immediate access to a network of connections. 
  • Your PI will be chief among these, but simply being an insider matters a lot.
  • These connections will matter most in academia, but they’ll matter outside it, too.
• If you are interested in a PhD, this is the most reliable way to get into a prestigious PhD program. 

If these roles are so great, why are they so easy to get?

• Labs really, really need labor.
  • PIs in many fields spend most of their time administrating. Someone else has to do the actual science.
• The labor required is generally highly specialized. That means you generally aren’t expected to have any qualifications.
  • You will be hired based on your enthusiasm, interest, and willingness to learn.
  • Simply by going out and asking for a job in a lab, you’re already demonstrating at least some of these qualities.
  • Though you will be working on fairly specialized problems, that doesn’t mean you won’t learn any transferable skills. Data analysis and science communication, at a minimum, are transferable skills. If you’re very young and have never held a paid position before, that’s incredibly valuable by itself—both for professional and personal development.
• PIs’ careers live and die by the quality of their students
  • Both by their publication rate, and the rate at which they graduate.
  • As a PI, a good student elevates you, but a bad student is a disaster.
  • PIs care a lot about selecting good grad students.
• The PhD interview process provides only low-fidelity signal
  • Undergraduate students who look great on paper do not reliably make good researchers.
  • Even the best signals of grad school success are pretty bad. (correlations <0.3)
  • A single interview is not enough time to get an idea of somebody’s potential for success as a researcher.
    • A work test is not perfect, but much, much better.
    • This is exactly what research assistantships are.
    • You can benefit from this even if you have no intent to do a PhD
• Bear in mind: these roles pay well below industry rates, offer no benefits, and no job security. They aren’t that great. But they are a much better option than unpaid internships or master’s degrees, or other mediocre-paying jobs that don’t meaningfully advance your career.
  • If you can easily get a much better job, then this post probably isn’t for you.

Why wouldn’t I just do an internship?

• Ease of access. I’ve applied to and been rejected for dozens of internships. I’ve never failed to get a research tech position just by going out and asking for one.
• Availability. There are only so many internships in any given area of interest, but far more academic labs.
• Compensation. Only some internships are compensated. Research assistants generally are.
  • Some PIs will try not to pay you. Don’t work for them. It probably doesn’t matter how interesting their research is. PIs who can’t attract funding are sinking ships.
  • Don’t work for course credit, either.
• You should probably not do an unpaid internship.
  • For psychological reasons.
  • Unless you have a really, really good reason.
  • Even if you can afford it.
  • You will probably do better work if you are being paid for your time.
  • I find that people simply try harder and put more effort in when they are being paid.
  • I speculate it works like this: We subconsciously assign value to the way we spend our time based on how much other people value it. $10 of effort per hour is less output than $30 of effort per hour, but $10 of effort per hour is still vastly better than $0 per hour. If someone doesn’t value my time at all, I subconsciously also value it less.
    • This will likely be reflected in your output and subsequently your reputation.
    • Maybe you have superhuman motivation and won’t be affected, but I wouldn’t count on it.
    • This process is, hypothetically, hackable. But it’s probably better to just get a job that pays you for your time.

Why shouldn’t I just get a master’s degree?

• They’re scams
  • They’re how universities make extra money
  • Even fancy ones
• Potentially it can even be a red flag, depending on the field
• You’ll be more productive if you’re not spending time hitting fake targets in classes.
• You can get most of the benefit of a master’s degree while being paid, as a research assistant.
  • You will get just as much research experience
  • You’ll still be building your resume
• Unless you know with high confidence that you have to get a master’s degree, don’t get a master’s degree
  • Even then, are you sure you can’t go for a PhD and “master out”?
• Rule of thumb: Don’t go to grad school unless someone is paying you for it
  • Your job, a doctoral program, doesn’t matter.
  • Even then, are you sure you need one?

What if I’ve already decided I want to do a PhD? Shouldn’t I just go ahead and apply?

• If you have already decided that you should do a PhD, it is within your interests to do it at the best school you possibly can, for at least three reasons:
  • Brand Name: No matter what you want to do with your career, the brand name of your PhD institution really matters.
    • In academia, if a middling school is known for a really outstanding department, that might be worth more than a good school with a middling department.
    • In government or other heavily bureaucratic non-academic institutions, the value of the brand name only matters more, because nobody is going to be looking at your thesis topic.
    • In industry, you quite possibly don’t even need the PhD in the first place, and can acquire the relevant skills just by working as a research tech.
  • Talented Collaborators: Ultimately you will rise or fall on the quality of your research. But elite institutions are good at concentrating resources, including talented collaborators. Being surrounded by smart people doing interesting things increases your own chances to do interesting, valuable things.
  • Flexibility: A large, well-funded institution is less likely to be constrained along pretty much any axis you care to name. If you aren’t completely certain what your research direction should be, this is especially important.
• It is difficult to access elite institutions [citation needed].
  • Unless you are obviously head-and-shoulders better than all the other applicants.
  • Or you flawlessly optimized your student career for the right kind of signals.
  • You probably didn’t do that. If you did, this post is not for you.
• However, see above: the interview process is a poor signal of researcher quality. PIs would much prefer to take on a student they already know and trust over an unknown quantity with a fancy CV.
• PIs in the same department talk to each other. If you work for one PI and then decide to switch to another lab, a good word from a colleague matters a lot—much more than a letter of recommendation, which are often not even written by the PIs themselves. (I’ve written several of my own letters of recommendation. Get used to this.)
  • PIs generally do not sabotage their students. Unless you did an absolutely terrible job, you will probably get a good recommendation. Simply being a known quantity is the important part.

What if I'm still a student? Shouldn’t I focus on my grades?

• No.
• If you are in the US:
  • Your grades matter in the sole event that you are trying to go to law or med school.
  • They kind of matter for certain PhD programs, but unless the department has sole discretion over which students PIs can bring on—and this is rare—they won’t stop you from getting in.
  • They don’t matter anywhere else.
  • Get the minimum acceptable grades you can and invest your time elsewhere.
• If you are in the UK:
  • My impression from speaking with UK-based students is that grades matter marginally more in the UK for employment
  • But I cannot imagine that connections matter any  less.
  • Your strategy ends up being the same—get the minimum possible grades, and invest your time elsewhere.
• In basically every conceivable way, you are better off focusing on a high-skilled research job than on grades.
• A bachelor’s degree is basically a requirement for entry into many high-impact sectors. A few years of your life are going to have to be spent getting it. Don't spend more of your time on this than you absolutely have to.
• For the record, I think that general partying and living it up are also useful things to do during college, if for no other reason than it's useful to get it out of your system now rather than later.

What about fellowships, honor societies, and awards?

• Probably not.
• A lot of these things are useful solely for signalling your class suitability for elite positions.
• This guide assumes that the reader has already not invested in conventional signals of academic prestige. If you already look very impressive, then your strategy is probably different.
• On one hand, prestige is a self-sustaining cycle—many prestigious things (positions, grants, etc) are awarded on the basis of being judged already prestigious enough to have them. On the other, it’s all inherently kind of fake—personally, I find it psychologically costly to engage in fakery.
• If you grew up being taught to do these things, and it’s not psychologically costly for you or you even enjoy them, then sure, why not. Prestige never hurts.
• But eyes on the prize—it’s always better to actually do stuff. Prestige is useful, but at some point you have to cash it out.

Okay, you’ve convinced me. How do I do this? 

• Assume this process will take several weeks.
  • Not several weeks of full-time work, but reading so many people’s websites is pretty tiring, and you’ll want to spread it out. Aim to spend maybe a few hours a week gathering names.
• Find the biggest, fanciest school in your area.
  • Bigger, more prestigious universities are easier to get hired at, because they have more money. If you intend to do a PhD, this is even more important, assuming you are trying to get into that school.
  • Prestigious, well-funded universities will probably pay you more. More money will make work more motivating, and make it generally easier to live in a city where these schools tend to be located.
  • This might be your undergraduate institution, or a school in the same city.
    • If you live in Boston or San Francisco, this will be quite easy.
    • If you live in Idaho, consider moving—but, remember that some mid-tier schools will have outstanding individual departments.
    • If you don’t want to move, you might be able to get away with working remotely in a computational lab.
  • No need whatsoever to limit yourself to one institution. If there are 3 great schools in your area, you should absolutely be looking at all of them.
• Consider subfield. You might have a difficult time getting paid for theoretical physics, but an easy time in experimental physics (though you will probably publish more in theory, if that is what you want to optimize for.)
• Make a spreadsheet. I like Google Sheets for ease of tabbing between department websites. Include these headings:
  • Name
  • Email
  • Title (Assistant Professor, Associated Professor, Adjunct, etc)
  • Website link
  • Area of research
  • Specific projects/areas of overlap with your interests
• Look for early-to-mid-career professors. These are people who are going to most actively need students. 
  • Assistant and associate professors will be your best bet. Adjuncts often have other priorities, and full professors tend to be less desperate.
  • Start clicking around the ‘Faculty’ heading of the relevant department. You may have to look for professors within several different departments; department appointments are somewhat arbitrary. For example, my current research is almost entirely ML focused, but my appointment is in the Department of Psychiatry.
  • Skim the faculty page and lab website of every PI who looks even vaguely relevant to your interests and goals. Aim for at least several dozen.
• If you have an institutional email, use it.
  • This decreases the chances of getting stuck in a spam filter.
• Email everybody you possibly can.
  • Set aside a whole day for this. Many people find sending e-mails emotionally exhausting, so give yourself time and emotional space for it. This will potentially be one of the highest yield professional days you have, so trust me, it’s worth it.
  • Yes, everybody. Including the adjuncts and distinguished full professors and even postdocs and students. These people will give you valuable information! They can give you the names of particular people you might be interested in speaking with who you didn’t find yourself; they’ll tell you which departments and which programs are best
    • If you’re considering a PhD, this is especially important. Mass-emailing is an extremely efficient way
• Use this template:
  • “Dear Dr. [Lastname],
        I am [explain who you are, and why they should care.] I have [experience/a background/am studying] in [a thing related to what they are doing], and I am interested in [the specific thing they are doing.] I’m most excited about your work in [name several specific areas/projects they are doing], and I would love the opportunity to work in your lab. Are you available anytime this week to chat? 
        Best,
    [yourname]
• Absolutely do not send identical cookie-cutter emails. 
  • They can be similar, but not identical.
  • Some departments have centralized e-mail management, and professors talk to each other. It will be pretty uncomfortable to be found sending identical emails to different people, mostly because it signals a lack of interest in that lab specifically.
  • That being said, plenty of labs are really that desperate for labor, and will not care.
• If they sign off with their first name, it’s okay to refer to them by their first name in future communications.
• Have your CV/resume ready. 
  • They don’t always (or even often) ask for it, but you don’t want to end up in a situation where you are asked for it and have to scramble to edit and update it while staying on top of your existing obligations.
  • Quick responses are a sign of interest, and you absolutely want to signal interest above everything else.
• The interviews will be informal chats. I usually at least skimmed some papers from every PI I talked to, but honestly, a lot of the time this is overkill—lots of them will just immediately launch into a summary of their research and try really hard to sell you on their lab.

My experience

• When I did this most recently, I was a student at a mid-tier university. I contacted everyone at that university’s bioengineering department, which amounted to 12 people even vaguely relevant to my goals. I spoke with three of them, and the only offer I got was clearly exploitative. If this happens to you, look elsewhere.
• I tried again at a top school in the same city. I made a list of 100 people, contacted around 70 of them, and received 40 responses. Of those 40, 20 were interested in interviewing me, and 3 made me immediate offers.
• The lab I picked was the one who offered me the highest hourly wage, and whose PI seemed most excited to work with me.

Caveats

• My experience is solely in the USA. I don’t think the picture is that different in other rich countries, but your mileage may vary.
• Your mileage may also vary by field and department.
  • This strategy has worked well for me in physics and biology.  “Having screwed around in Python for a few weeks” will probably be a less valuable asset in computer science.
  • All bets are off for any of the humanities—but economics and political science are probably your best shot.

Outstanding Questions

• Has anyone tried this and failed?
• What fields are good for this, which are bad?
• To what extent does this apply outside the US academic labor market? What considerations make this different?