Samuel Bird

Is there a right number of projects to balance? (for me)

The problem of how many research projects to work on in parallel has loomed often in my mind during my time as a PhD student. It’s hard to assess what the optimal strategy is, so I’m going to put down some thoughts on the matter to better think it through.

At the beginning, I found myself working almost solely on one project, and at the time I felt often a sort of yearning for more projects. I think partially this was out of insecurity, or a fear that ideas are hard to come by (which I no longer think - coming up with ideas is a skill that requires practice and experience, while really good ideas probably require a mixture of that, iteration, exposure, and luck) Partially it was out of a desire to explore more though, to see more different kinds of research, and discover what I’m most interested in. Finally, there was also a feeling that I quite often had slow-feedback large simulations running, and so there were periods of time I could quite happily spend working on another project.

Turning to my environment, I encountered very different pictures depending on where I looked. Some PhD students in different fields seemed to have upwards of 5 projects ongoing at any time, especially in certain fields, while other PhDs and Postdocs would be focusing on only 1 or 2 projects at a given time. It was also typically quite difficult to figure out exactly what it means for somebody to have N projects. It’s unlikely they are the primary researcher on all projects if N is large. In fact, some groups seem to function in a way where everybody is working as part of a larger collaborative machinery. This seems to depend a lot on the field and on the specific group dynamic. My impression is that some students join a group and then projects are assigned to them on a conveyer belt depending on the collaborations that arise from the PIs conversations. This can surely be good, since you can get a lot of exposure to different people and (maybe) different topics. The likely downside though is not having the freedom to pick your own projects (and the necessity of learning this core research skill), or the time to explore wildly different topics than what your group is focused on.

It seems that non-primary input into a project can vary significantly in what it entails. The general picture I get though is that it’s a lot less demanding than primary research. Perhaps my mental picture of this is flawed though. It’s not so clear to me how a buddy research project works - with two “co-primary researchers” - and how demanding this is. Is it halved? Is it just as demanding as a single primary researcher? It probably depends on the project as always. But I’m interested in large brushstrokes here so I will forget about this for now.

So, the question I’m really interested in is how many projects is it best to work on at the same time, given that you are the primary researcher?

What others say

There are a bunch of interesting takes on this that I’ve read. Here are a few of my favourites.

In his reflections on his time as a PhD student, Eugene claims that he cannot work simultaneously on lots of projects. His rule of thumb is “1 project that you’re the primary writer / programmer / theorem prover on, 1 project that you have large intellectual contributions to but minimal programming time / expected theorem proving”.

My first thought was that this seems quite conservative, and, to be honest, at odds with what I see “out there”. I found it hard to reconcile how researchers who go on to get tenure track positions, and who typically have an abundance of publications (generalising somewhat), can actually be operating like this. For example, during the time of his PhD Eugene seems to have participated in about 20 papers (roughly, maybe I misinterpret exact numbers?). Granted he is in all sorts of places on the author list, but he has a fair number of first author (and presumably some joint first) papers. So a skeptical part of me wants to question whether or not he would have been so successful had he taken his own advice? However, it is clear that, had he not been involved in many large collaborations (most of the papers have at least 5 authors), his output would not appear so prolific so I think his point still stands - I’m sure some of the projects fall into a third category of “projects where you are neither the primary researcher nor make a large intellectual contribution, but rather a smaller contribution” (This might also relate to the third key reason he makes (see below)).

That being said, I agree with his premise. He astutely identifies three key reasons why we trick ourselves into taking on many projects. The first two are the desire to work with a particular researcher (a kind of FOMO), and variance reduction (what I call damage control).

These are batted away like a pro. We may want to work with a particular researcher, but if that work only gets $1/n^\text{th}$ of our attention, for an $n$ too large, then it’s unlikely they’ll be impressed. Similarly, having 4 projects in case one of them fails sounds fine in principle, until you ask how much more likely they are to fail due to the fact there’s 4 of them? I think it’s easy to underestimate that severely - less time, attention, and motivation, as well as a higher general level of fatigue, all accompany a larger project load, and these factors can likely lead to a bunch of projects being abandoned. Think of it this way: if you work on 4 projects in parallel, but limited attention only allows for all 4 (optimally) to be finished in 2 years, this is much less enjoyable than working serially on 4 projects, finishing each in 6 months, and you have less to show for it in the meantime.

The third key reason Eugene sees for why we trick ourselves into taking on many projects is “a failure of recognising the order in which things occur”. What this means is that really good researchers don’t have a ton of papers because they work on a ton of projects at the same time, but because they worked on a small number (1 maybe) of really good projects, and that these naturally spawn a lot of follow-up work. Therefore, when we see really good researchers and notice they have lots of papers, we should think rather that they worked really hard on a small number of projects and this lead to the later volume. By demonstrating expertise in a nice this way, one can also imagine this gets you invited onto other projects in a more advisory or lesser collaborative role. This is one example of the “third category” of research contribution I mentioned earlier.

Approaching the question of attention more broadly, Paul makes a case that there’s only one thing, a research project in our case, which occupies your shower time. I’ve been told throughout my life how our subconscious is always buzzing away at problems, and this is the reason for the sudden moment of inspiration - shouting “Eureka!” in the shower and so on. Supposing that our subconscious is only going to be buzzing away on one problem, that which most occupies our attention, it’s reasonable to presume then that nearly all of our little eureka moments will be going towards one thing.

The idea is nice, and understandable, which is a good sign. I wonder why it wouldn’t be possible to switch the main focus every week though, and accordingly shift your shower thoughts from one to the other? Of course, in my experience it is not so easy. This might be due to large psychological switching costs in doing so, or an inability of the subconscious to process this change in primary attention so easily in a useful way which allows for effective background pondering.

One commenter describes how they worked on 3 different topics during their PhD and this allowed them to explore different branches and discover one they really liked. They even mention having a friend who worked on 5 projects in parallel. It’s hard to critique without knowing more context. If they were the (or one of the) primary researcher in all projects (I doubt it), that’s a mighty achievement but it’s unclear to me how it would have gone if they had restricted themselves to 2 projects at a time? This also seems to be a different angle on the ‘variance reduction’ reason to work on many projects at the same time which Eugene already argued against.

They write of the advantages of such a style: “one broadens their knowledge (some ideas from one topic may be useful for the other), builds a bigger professional network, learns to work on different things in parallel”. I agree that the skill of learning to work on different things in parallel is a valuable one - there are times in life where it will become necessary and it’s best if you can manage it without losing your mind. As for broadening your knowledge and building a bigger professional network, these seem to me to be forms of ‘FOMO’, and it’s unclear why it wouldn’t be just as effective (or more) to work on fewer projects but get each completed faster. This just seems to be a facet of working on a range of different topics, not necessarily in parallel.

I would imagine that all these benefits can likely be gathered optimally by having just a single secondary project, which we can switch to when we have the free time or want a change of scenery. In fact, agreeing with Eugene, I can’t imagine how it would be feasible to have more than that without incurring damagingly steep fines on attention in each project.

My take

Another aspect of this whole shenanigan however is that I want to spend time learning about new ideas and methods that catch my curiosity. Learning is fun, and exciting, when guided by curiosity, and leads down new avenues of inquiry. How can I possibly know that what I’m working on right now is the best thing for me to be working on? It almost certainly isn’t. One way of finding new questions and ideas is by talking to people, and I think this is very important also. This kind of open exploration is something which I imagine reaps both good practical and spiritual rewards. Therefore, finding a way to fit it into any project management scheme is import to me.

So, with all of this in mind, I’m going to propose my own (deeply unoriginal) project balancing approach.

My plan is to stick to this schema for the coming year and reflect on how it went. Maybe I will fail, in which case it will be interesting to understand why. Maybe I will succeed, but have a sense that I could have done more, or perhaps should have done less. A part of me wonders if such an outline of personal working habits can ever be instructive or useful, though I think that’s being pessimistic. At the very least, by having an idea of what I want my project management to look like, I can better notice when I am abusing the system and perhaps my wellbeing.

I’m very curious to explore other frameworks for balancing projects though. In principle if it is possible to get really good at attention switching and be able to glide between 5 different projects like a swan, and this can somehow come together in a way that means your time with each project is a small portion of sweet golden focus, then great. I’d love to learn that. I’m skeptical.