Reasonable research progress & useful habits

How to not let yourself down when setting goals

Jan 13, 2026

As I would often find myself setting unattainable goals, this post is half a reminder for myself to refer back to when I ineviatibly say I’ll work on 20+ different things within a week. The other half is advice to others who find themselves not completing everything they’ve set out to complete within some period of time. This is especially useful when working with others in a collaboration, as solid communication about what experiments you expect to perform, what theorems you’ll prove, or what you’ll spend time learning are essential to making fast progress.

General Rules of Thumb

• Have a maximum of one big, attention-intensive task a day—that’s an experiment coded up if it isn’t thousands of lines long, a portion of a proof you’ll need to do, or validating parts of your setup are working as expected. If you aim for more than this, just know that it’ll only be viable in the short term. You’ll find dozens of small things that come up when completing these tasks, mostly debugging, that’ll eat away at your time. Factoring in these issues upfront let’s you be more realistic about the quality of work you get out, instead of feeling terrible about how behind you are.

• Having subtasks to complete are generally fine, so long as they follow a few criteria: (1) if they’re just parts of something bigger, then be honest about how much of a “subtask” it is. It could be that these require a lot of attention—these should either be reclassified, or be noted to have some special importance to them, even if they won’t take up much time. (2) Subtasks should be secondary to the main tasks. I’ve become used to putting down large, daunting goals for myself and cozied up to the much more bite-sized problems, even if they end up hindering what I’m actually wanting to do. If something is truly important, then treat it like it is. (3) Too many subtasks end up integrating to something big. This is fine, so long as you set aside something like a day to do everything.

• Sustainable work requires breaks, within a month, a week, even a day. Solving problems often comes focusing on something entirely different and coming back later with a refreshed mind. Don’t exhaust yourself!

• Don’t have giant leaps between successive steps of a project. Massive steps are for the big ideas, not the everyday work. If progress becomes overly jagged, it’s much more likely that you’ll slip on something you would’ve avoided by taking things more carefully. In the worst case scenario, you’ll think something doesn’t work, when it really would have with just a bit more forethought.

• Remind yourself of the end goal. It’s easy to be gradually making progress in what seems like the right direction but end up nudging yourself off the path you wanted to explore. Make sure to take steps back occasionally to see what the point was, and adjust if needed.

• Get stuck often, but don’t stay stuck. Immersing yourself is a time and effort-intense process, one that we’re all too familiar is ultimately worth the price (unless you’ve grown up with AI chatbots being the norm in which case I’m not sure what to say). What’s much harder to both say and do is getting yourself out of it—not just by solving the problem, but by getting outside help. It might happen slowly, but eventually you will be able to help yourself just as easily.