2022 01 18

Today I spent some time working out a hierarchical plan for the semester. My main goal is to have a good paper to submit to the NeurIPS 2022 conference. Hierarchical planning is something that I’ve seen in a few books. For sure in one of Cal Newport’s books and more broadly discussed as a method of business strategy in The Art of Action (as an aside I think this book offers a really compelling view on how to run large organizations successfully, which is unfortunately harder to take away to an individual as you can’t let groups of people below you in the hierarchy figure out the details - you have to do it all!). Hierarchical planning is when you start with your end goal and work the tasks backwards in order to determine what you should be working on today. I found this to be pretty hard to work out and maybe that is just because I am new to planning in this way. To be honest most of my planning historically has been pretty gut-based. Sure, this is using some complicated neural processing of our brain to tie our vast arrays of desires and objectives together and produce some sort of output as a “feeling”, but gut-based approaches aren’t easily systemized or easily reflected on to know what aspects of the decision making was good or not. To me hierarchical planning offers the sanest and simplest approach that leads you to your objective. For me the trickiness is knowing what exactly the steps should be in the independent parts as this is a research project, not producing something like a physical good where there is a fairly clear process. I think doing this planning would be helpful to do with someone else as they will provide feedback to your plan and what steps don’t seem as clear - unclear steps would indicate that you skipped steps in the breakdown and thus added too much difficulty into a single step.

The books The Mysterious Island and The Martian both provide examples of characters that solve problems in this way. In both cases the characters are stranded somewhere and have roughly the goal: survive. They come up with plans that lead to this outcome and then work from first principles to improve their situation. Their first principles being basic necessities of life such as food, water, and shelter, and then improving from there. One thing that I did find useful from the hierarchical planning I did was to identify useful first principles for myself. To produce a good research paper you need to have something good to present. To present something good you need to understand the greater context of the research field and what work has been done and how your work fits into that world. To do this you need to read and understand a lot of existing research. So, a first principle here is to only read something if you will take notes on it. If you don’t take notes (or whatever method you use for understanding new material) then you will only succeed in convincing yourself that you understood the work. I’m great at that game! Writing out this plan helped me recognize this fairly obvious detail that to be serious about having a good paper, I need to be serious about having a good information intake system. It is for sure laborious, but so is starting a farm of potatoes on Mars or building tools from scratch on a remote island. You have to do some dirty work in order to satisfy your basic needs! This sounds potentially miserable, but the hidden benefit of taking care of your first principles is that you can have a sort of compounding effect on your abilities. With the basics taken care of you can start to get concerned with higher level things like how to communicate with Earth from Mars, or in my case thinking about the winds of my research field and what areas/questions feel unexplored, etc.