Are delays and 'being stuck' normal in PhDs, and how much should I work to compensate?

Question

I'm doing a PhD in Computer Science in France (completed 10 months out of 3 years), and I am annoyed at how slow I am to produce results.

Long story short, I have explored various potential solutions to a certain problem in Computer Vision using Deep Learning - defining the problem properly was by itself very difficult - meeting various dead-ends and delays like buggy open source code and issues to reproduce results from existing papers.

I have produced small-ish results, but I haven't got enough for a quality paper yet (e.g. CVPR). I should be close, but then I've been convinced of that for a couple months.

I work full-time on my PhD (8h-17h), with occasional 1-2h overtime. I would say I am usually quite productive although there are days it would have been more useful to stay at home sleeping. I rarely work during week-ends: it is way more efficient to test / develop the algorithms in the lab, and I am just tired. The exception is when I have something to write, which is not the case right now.

My supervisors are somewhat disappointed. On the one hand, I feel like I could do more, like by working at home or something. It is true that in the beginning of the PhD I wasted some time by leaving work earlier at 4pm (which I kinda regret) and by exploring algorithms that were useless in hindsight. I also feel guilty if I spend my week-end binge-watching movies or playing games on my PC. On the other hand, I am also usually out of ideas or things to try and may not do something useful anyway.

I am somewhat at a loss, hence my question: how far should I go? Should I always try my hardest all the time, like when I prepared exams for engineering schools (classes préparatoires, where free time is an unknown entity)? The thing is that I just do not know how to completely solve the problem I am researching. It is very different from studying for an exam, where I know precisely what to do. If I had the solution, I'd be happy to write the paper...

I always could do something, like looking for new papers to read or studying deep learning. I also sometimes hear about people working until 2am in other labs and publishing papers after 6 months, which does not help. I do not know what I should do, and afraid of regretting my decision either way :|

Answer 1

What you describe seems pretty normal, except for the disappointed supervisors. But maybe your work habits are inefficient. It seems like you may be stressing time in the seat too much. If you don't take regular breaks, get enough sleep, and tend to your health, then your mind won't work efficiently. You will get stuck and pounding harder isn't the way to think deeper. Take a break regularly, especially when you get stuck. It doesn't need to be long or elaborate, but a short walk, run, or bike ride, can restore both your body and mind.

Your mind doesn't just work when you try to force it to. It actually works all the time, even when you are sleeping. Sometimes the break is what it takes for your mind to stop focusing on minutia and start to integrate what you have been looking at.

But, if you are really doing research, and not just going through the motions, then you can't schedule results. They come when they will.

“If we knew what it was we were doing, it would not be called research, would it?” ~~Albert Einstein

Another trick is to talk to people about what you are doing. They need to have some knowledge, but if you just try to explain it you will, perhaps, come to a better understanding yourself. Some people even "talk" to a Rubber Ducky that they keep on their desk. Tell it to the Duck. But people can give you feedback.

The "dead ends" you explored may also actually be valuable, later if not now. They might actually do nothing more than provide a break from trying to push forward. And those breaks are essential.

If you do too many push-ups your muscles get saturated with lactic acid (I think I got that right) and weaken. Something similar happens with your brain. Work hard, but take breaks. For me, random breaks, rather than a schedule, works best. When I'm not being productive, do something else for a while.

But if your breaks take you away from your desk as they should, you can also carry a few note cards, including blank ones. You can use these to remind yourself of partially formed ideas and capture new ones if they occur during exercise. Don't force yourself to work on a break, but have them ready to capture any serendipitous thoughts.