Doing research on the side while working a full-time job

Due to the COVID-19 pandemic and associated economic decline, I wasn’t able to get a research position for after my PhD. I’ll be starting as a patent examiner in August. My preference would have been to work in a research position for 5 to 15 years before retiring early and then doing whatever research I want to. While I will continue looking for research positions, given the current state of the economy, I recognize I might be in this job for years.

Consequently, I’m looking for best practices if one wants to conduct research on the side while working at a full time job. Several people at this forum are probably in a similar position and can offer advice based on their experience.

Here’s an outline of my current plan:

The most important thing may be to scale back my research goals. My time now is fairly limited, so I’m planning on having only two projects until next year. (Might be a good idea to limit your number of active projects anyway!)

Once I’m able to, I’ll switch to the “4/10 plan”, a work schedule that has four 10-hour days per week. I find that I work best if I dedicate an entire day to a particular task. This will allow me to dedicate 1-3 days per week to research rather than 1-2.

The USPTO actually offers a fair amount of time off (26 days of my choice per year for new employees, plus 10 federal holidays). I’m planning to use this time to attend conferences, visit collaborators, and otherwise do research. Overall, in combination with the 4/10 work schedule, I’ll have 180 days without work per year (vs. 128 if I work a conventional 5 day/week schedule).

I’m planning my schedule to save as much time as possible. E.g., I’ve long been a fan of commuting by bike as a way to get exercise in an otherwise unproductive time. By using parallelism, batching, and planning one could save a fair amount of time in the long run.

I’m at an age where many people start having kids. I’ve decided to forgo this as I can’t justify the time expenses.

Above I wrote about plans in terms of time management, but I suspect that managing my motivation is going to be similarly important. A major concern of mine is that working 40 hours a week is going to tire me out and make me not feel like doing research. I’m not sure exactly how to handle this aside from giving myself adequate rest.

Any advice or experiences are welcome.

1 Like

All the best… :slight_smile:

Great topic @btrettel, I’m glad you’ve started a discussion on this.

I should start by saying that I don’t think the community has really developed a set of best practices yet - despite independent research dating back to antiquity everybody’s approach seems to be different and there seem to be various successful models. The key question has always been how to earn an income while doing independent research as most salary grants require university affiliation. Your choice of doing research on the side of another job seems to be the most common and reliable option these days, but I’ve also seen people getting funding from philanthropy (from either crowdfunding or private individuals), venture investment, project grants and consulting (IGDORE is currently in the process of starting up our research consultancy).

A few comments on the personal situation described:

  • Patent examining seems like a good full-time job to have in parallel to research (and of course, you are in good company),
  • If you are planning to continue applying for mainstream University positions or fellowships then I’d advise focusing your initial research at easy ways of getting onto publications as a middle author, probably by assisting other researchers with parts of their projects (if you have any unpublished thesis chapters/experiments you should also publish those). One or two middle author publications/year should keep your applications competitive for the next few years, and although a career break is usually viewed negatively on academic applications I expect there will be greater acceptance of this during the post-COVID recovery,
  • Even if you end up in a situation where you are no longer focusing on getting papers to support applications, I’d recommend continuing to maintain some collaborations with University researchers. I think this is important for keeping your research standards on par with mainstream academia,
  • Depending on your work style, you could consider working a few hours on research a few evenings a week rather than trying to concentrate all your research onto a couple of days (perhaps on days you know your day job will be light). See here for a description of this strategy by somebody I regard as having done very productive research while also holding down a job,
  • Planning to regularly work 7 full days sounds like a recipe for burn out. Take some time off each week!
  • Do you like going to conferences and visiting collaborators? Travelling costs time and money. Since moving to Brazil I’ve found that I can effectively collaborate remotely via email and skype. The in-person serendipitous connections of conferences are lost, but I’ve also made valuable connections through being active in the right forums, slack/facebook groups, etc. If a group doesn’t exist that fits you interest you can always make one. (I think that the Ronin Institute is also experimenting with virtual water cooler session for their researchers),
  • In terms of motivation, remember that the benefit of being an independent researcher is that you can focus on the parts of research you like and/or think are important, rather than what is required for career advancement as a University employee. If you don’t like doing peer reviews then decline them. If you just want to get your work published then submit directly to specialist journals rather than chasing impact factor. However, I would suggest setting clear goals for what you want to achieve with your research, otherwise, it’s easy to lose focus.
  • Do literature or computational based work rather than experimental research if possible. It generally seems impractical to buy and maintain research equipment as an independent researcher. If you do need to do experimental work then this is the ideal case for collaboration.
  • Outsourcing can be a good idea for some tasks even if you have to pay for this yourself. I’ve hired technical proof-readers on Upwork recently and it’s saved me days of time (although my spelling and grammar is terrible, so others may not get as much mileage). I’d never have considered doing this while working as a postdoc.

[I ended up writing rather more comments on your situation than I expected and have to go, but I’ll probably add some reflections on my own experiences as a consultant and independent researcher later.]

I’m really interested to hear from any other forum users who also have experience as independent researchers. If we get a few interesting responses we could even put together a 10 Simple Rules for Independent Research style article :smiley:

1 Like


Thanks for your thoughts, particularly on my personal situation. I’ve been busy preparing for my new job and PhD defense so I haven’t had a good opportunity to reply until now. I’ve thought about your thoughts on and off since you posted. If you have the time to write more on your own experiences, I’m interested.

The comment by Peter Hurford is the sort of information I’m looking for. I agree with him that I need a certain amount of time to get started. But I doubt that I’d be able to do deep work at night after working my day job; that day is spent. So I think it would be better to reduce the number of days that I do paid work. Today I started thinking about working three 13 hour 20 minute days per week. I started looking into the research on alternative work schedules. To my surprise, it doesn’t seem that working a “concentrated” schedule has negative effects (the total hours per week is kept constant), though I’m not confident about the quality of the studies.

Still, perhaps I can categorize tasks into shallow vs. deep and do the shallow tasks on weekdays? If I only have about 2 hours, I can’t get much done unless it’s something shallow like transcribing data from the literature (which is so shallow that I can do it while watching TV, etc., though I worry I introduce errors doing that).

I am concerned about overworking myself. I don’t intend to work 7 days a week. I find much of my research energizing, but not all. So I’m planning a fair amount of rest time to balance all the work. The amount needed and type of rest, however, is not clear for me at present.

With respect to publication to help get a traditional position, if needed I can get 3 more papers from my dissertation. I also have several incomplete projects, at least one of which can be completed without too much trouble. (The vast majority of these are single author papers.) I’m not too worried about publication if I want to get a research position, but I think this is worth thinking about.

You’re right that travel is expensive. I’m planning at most one conference a year at this point, maybe zero now that I think more about it. I’m mostly ignored by others at conferences, but I find them useful to meet people. It’s not clear to me what could replace conferences for me aside from contacting researchers individually. The few online groups on fluid dynamics focus almost exclusively on education rather than research. Perhaps if I started my own this situation could change? In response to COVID, I’ve been thinking about contacting a professional organization I’m a member of about organizing a virtual seminar series, but I haven’t done that yet.

My research so far has been theoretical and computational, using primarily data from the literature. But I am realizing that I can do some limited experiments, if I keep the costs down. I consider myself an expert on scale modeling, which can cut costs by a lot. I also find that simple methods often are sufficient. In January, I built an experimental setup intended for my PhD research for about $500 (of my own money). My advisor was unimpressed with the experimental method and recommended some fancy setup with a laser. But my setup is actually far more precise than what’s in the literature and easier/faster to operate. Some laser-based technique would be more precise, but it’s precision that I don’t need. (We both decided against putting the existing data in my dissertation, as I intend to collect a lot more before publication.)

It also helped a lot that I did a fairly comprehensive literature survey, so I knew that the experiments I did weren’t duplicating much already done. Aside from some basic replication to make sure that my setup and/or the literature are consistent, I’d prefer to focus my experimental efforts on getting actually new data.

If the cost of conducting an experiment on my own is below travel costs, then visiting an experimental collaborator can be avoided. I hadn’t thought about this until just now. I had been planning to visit a colleague to conduct some experiments in the future, but I’m going to see how expensive it would be to do those experiments on my own now.

Anyhow, it’s getting late for me so I too must go. If anyone else has thoughts on this, I’d appreciate hearing them! I’d also be interested in identifying successful independent researchers who simultaneously worked full-time jobs, like Peter Hurford.

1 Like

@btrettel: thanks for this. very inspiring to read it. i’m rooting for you. :slight_smile:

1 Like

Back to looking at this now, @btrettel

Today I started thinking about working three 13 hour 20 minute days per week. I started looking into the research on alternative work schedules. To my surprise, it doesn’t seem that working a “concentrated” schedule has negative effects (the total hours per week is kept constant), though I’m not confident about the quality of the studies.

I expect this will depend a lot on the individual. I know from personal experience that I can do good work for 12+ hours a day, but working after 10PM almost always leaves me slow to start the next day and I can’t continue this for more than around 5 to 8 days or my work quality will start to go down quickly. But I’ve also seen people who are left frazzled for a day or two after working one 12 hour day. So self-experimentation with different work schedules seems important. I speculate that weekly consistency becomes more important to maintain productivity the more you schedule differs from the standard 8 hour - five day work week.

Another thing that I’ve heard can be useful to experiment with is how you relax. I know I personally relax more by going out in nature than watching TV on weekends (although I do the latter more regularly…), so if you want to optimize productivity while working a demanding schedule, optimizing your down-time may also be important. Lynette Bye’s blog has several articles about self-experimentation in the context of work schedules (and other aspects of productivity) that expand on this idea, and I recall that one of the people she interviews discusses experimenting with leisure activities as well.

I consider myself an expert on scale modelling, which can cut costs by a lot. I also find that simple methods often are sufficient.

That’s cool! Can you recommend any good introductory material for scale modelling? Do you usually scale down or up? Most of my work now is in computational optics and I’ve never really thought about scale modelling, but it would be an interesting extension. The only scale modelling I know of in visual science is actually quite impressive, in the early 1980s, a Russian group scaled up the dimensions of photoreceptors x15000 and did optical studies using microwaves! (abstract link)

Aside, you mention deciding not to have kids in your first post. Previously I also thought that having kids would take up too much time, but reading the book Selfish reasons to have more kids provided an interesting perspective and left me more open to the idea. While I still don’t feel that I want to have kids much more than before, Caplan does at least make a good case that the minimum cost (in time and energy) of raising kids is much lower then most people assume, and further, half-hearted parenting isn’t likely to adversely affect your kids’ long-term outcome.

I agree that Peter Hurford is an exemplar of somebody doing research on the side of a full-time job. I think quite a few people in the Effective Altruism community do, or have done, research on the side of a full-time job so there may be more good examples there. I guess a caveat is that a lot of EA research is generally not academic, in the sense that most researchers don’t have PhDs or publish in journals, so examples of good practice there may not always generalize to independent academia. Ronin researchers have discussed part-time academic research itself as fractional scholarship and how it can fit into a broader academic ecoysystem, but I just don’t think we have many good examples to start generalizing from yet.

Personally I don’t think that I, myself, am a good example of somebody doing research on the side for two reasons.

Firstly, since moving to Brazil I’ve mostly done independent scientific consulting which has been very flexible. So if I’ve wanted to set aside some time to work on my own project then I can usually do so, often at quite short notice. This has been great for me but doesn’t really reflect the stricter demands of a full-time job.

Secondly, I feel that while I’ve been quite good at research idea generation and proposal writing in parallel to consulting work, I haven’t really been able to make progress on research projects when I’ve tried to do that in parallel. A loose hypothesis I’ve formed around my own work habits is that I’m limited to working on two separate long-term ‘jobs’ at a time (maybe three if I’m lucky) before I start to feel overwhelmed. While doing scientific consulting I’ve worked with short-term (less than a week) and long-term (more than a month) client projects and found that I’ve set each long-term consulting client’s project apart as a ‘job’ in my mind (being on the IGDORE board also gets a job slot). I’ve found that I can take a few days to do a short-term project or some of my independent research while also continuing two or three long-term ‘jobs’, but if I then try to regularly come back to my independent research over a period of more than a few weeks it moves into the ‘job’ category and I feel I have to cut something back. I’ve then generally prioritized consulting work (it pays the bills), but this has been frustrating as I’ve set aside research projects that I thought were timely and had a lot of potential. For me, this seems to be an issue of mental compartmentalization, as I don’t feel particularly limited by the number of projects I can put within any given job, so it would be preferable to find a mental strategy that compartmentalized all my long-term client projects under a single consulting ‘job’, but I haven’t found a way to do that yet. As it is, I’m trying to wrap-up my current long-term consulting projects and will then limit myself to working with one client at a time to allow space for both IGDORE and my own research work.

Anyway, it sounds like you probably wont be in the situation I described above, but I guess it could still be worth thinking about how to mentally compartmentalize different aspects of your ‘jobs’ if you end up felling like you are working on too many things.

1 Like

I just recalled Cal Newport’s book Deep Work which I read earlier this year and recommend (Peter Hurford mentions it as well). One point that reinforces my comments above are that he recommends not doing pseudowork during your leisure time. Data entry in front of the TV would probably fall into that category for you, and I often find myself mentally writing email responses when I should be relaxing.

Another point he mentions is productive meditation (where one thinks deeply about something, possibly while engaged in another, non-mentally demanding, activity). I find that I have really good insights when thinking about my research while I’m doing aerobic exercise at moderate intensity. So I guess this could be quite useful for you to multi-task some aspects of your research.

Gavin, I thought I replied here earlier (particularly with respect to scale modeling) but I guess I had not, so here’s a delayed reply.

The USPTO’s patent training academy has some weeks where it’s basically impossible to do 8 hours each day because they don’t allow you to work on certain Fridays. (I don’t know why that is.) I’ve tried 8, 9, and 10 hour days in the past two months. So far, for me, working 10 hour days has not had any perceptible negative effects for me, aside from increasing the probability that I stay up later than I should doing chores or whatnot because “it has to be done before tomorrow”. I’ve slowly gotten better at streamlining or eliminating my chores to avoid that. I am practically unable to do any research on days when I have a 10 hour schedule, aside from jotting down random ideas I have to do later. I doubt I could do a 13 hour 20 minute day simply from a logistical standpoint. Also, more practically, as I recall, the USPTO does not allow one to work more than 12 hours a day (3 day weeks are against the rules), and while I’m in the patent training academy I am not allowed to work more than 10 hours per day.

Can you recommend any good introductory material for scale modelling? Do you usually scale down or up?

Any decent introductory fluid mechanics textbook will have a good introduction to scale modeling. This is the book I used as an undergraduate. You can buy old used versions for far cheaper than the latest version. Chapter 7 is on scale modeling.

So far I’ve only scaled down because of space and cost limitations (can’t do full scale). But there can be good reasons to scale up. If the phenomena is too small to properly interrogate or fabricate a model then scaling up can be a big advantage. The Russian research you linked to suggests there were issues fabricating a true scale eye model, so they scaled up instead. This also apparently required scaling the “light source” so that it now is microwaves instead of visible light.

In terms of mental compartmentalization, one thing I like about working for the USPTO is that at the end of the day I don’t think about my paid work much at all. I’ll talk about it sometimes during dinner but beyond that, it doesn’t cut into my time. I also don’t have to spend time finding clients, etc., as my work is assigned to me automatically and there’s no shortage.

So now that I’ve worked for the USPTO for nearly two months, I have a much better idea about how the job could work for independent researchers. Here are some comments and observations:

It’s important to keep perspective. If I were a full professor, the statistics I’ve seen suggest I’d get only about 10 hours per week of my own research done:

That’s while working a roughly 60 hour week. By working 40 hours a week and doing 10 hours a week of research, in some sense that would be doing better than an average full professor. (I recognize that a large fraction of the time spent by a professor is directing the research of others, but I don’t want to be a manager. I want to do the research myself.)

In the past two months I actually have done some good research. I’ve kept track of approximately how many hours of research I’ve done since starting, and it’s about 36 hours’ worth so far. While not that impressive on its face, I was far more efficient than normal during this time as measured by my own subjective evaluation of the progress made and also more objective measures like the number of pages I wrote down in my theory notebook. Judging by notebook pages, September was one of my most productive months of all time! I recall going through my records and I probably worked roughly an order of magnitude more on research in terms of hours on every month where I was similarly or more productive. I attribute this partly to luck — I’ve hit a “rich vein” of research so-to-speak — and also to effective management of a part time schedule. I’ve read a fair bit about “creative incubation” and I think I made very effective use of this in September. Basically, I accepted that I would not be able to work continuously and I designed my research schedule around this. Typically I would work until I get stuck on a problem that I don’t know how to move forward on. Then I carefully document what the problem is. So far I’ve frequently had either an epiphany that solved the problem or (more common) by the time I returned to the research, the answer seemed obvious (i.e., there was no epiphany). There’s a lot written in the creativity literature on this under the title of “incubation”. Now, this strategy only works when one is bound by creativity. It won’t work on “grunt” work, like the reprogramming of something that I’ve been planning for a long time. I know what I need to do there, but I’m waiting for the right opportunity (e.g., I run out of creative ideas).

In terms of library resources, the USPTO does decently. The USPTO has subscriptions to many journals, including some that I did not have access to at the University of Texas at Austin. But in general the coverage is limited to technology-related areas, and also more limited than a university’s coverage. This page lists many of the resources they have. I should note that this is not up-to-date, e.g., contrary to what is claimed on that page they do not have access to any American Institute of Physics publications to my knowledge. The USPTO also does have interlibrary loan, however it is strictly for work purposes only. Realistically, I’ll be making a lot of trips to the Library of Congress once they open again to get papers that the USPTO has no access to. And I will be trying the interlibrary loan service of a local public library, probably at cost, to get whatever the Library of Congress doesn’t have that can’t be purchased. I recall a discussion I had with gwern on Hacker News about public library interlibrary loan, and he seemed skeptical that it would deliver here, but I don’t see any reason why not.

During the USPTO training I got the impression that you need to ask for permission from the USPTO to publish. I asked and got the impression the process was a formality. They don’t seem too concerned that you publish as long as no one thinks that the work is endorsed by the USPTO or the US government. They didn’t even seem concerned that I publish something which could possibly be patentable; they made it clear during the training that I can’t own the rights to a patent (one of the downsides of the job), but I apparently can make a “defensive disclosure” that would prevent anyone from getting a patent on an invention. This is good because one advantage of the USPTO is that I get the keep my own IP rights. If you are a paid employee of most universities, government agencies, and companies, you probably sign away the rights to practically whatever you think of on work time or even your own time.

In terms of schedules, I previously mentioned that you could do a compressed schedule of 4 days per week. What I didn’t know is that the USPTO also has a part-time option that maintains full benefits (though you’ll pay more out of pocket for the benefits). You can work a minimum of 16 hours per week. At present I intend to try to switch to part time as soon as I’m able.; you can only make the request once you’re a GS-11. I’m frugal enough that that a 50% pay cut would be acceptable to me. There are a limited number of part-time positions available but I’m not certain that they’re all taken at present.

I’d be interested in a discussion of possible part-time jobs one could take if they want more time to dedicate to research. I think the USPTO would be a good option here given the pay and benefits.

Also, while not of particular interest to me, after a certain amount of time you can go remote while working for the USPTO. I know this would be a big motivating factor for many people. If I stay at the USPTO long enough, I might use the remote option temporarily to visit collaborators for longer periods. I’ve talked to one guy who moved more permanently to collaborate more closely with a previous research colleague of his, so this sort of thing already is happening.

Two months ago I was not particularly enthusiastic about working for the USPTO, but after getting some perspective about the “right” choice (becoming a professor) not being that great in terms of time, learning about the option to go part-time at the USPTO, and a bunch of other smaller things (benefits, location, on-site conveniences, etc.), I’m more confident about working for the USPTO long-term now. It’s not perfect but I think it’s doable, and it might be a better deal than academia for most.

1 Like