This post is part of my "research" series.

Some time ago, I started a series of posts on D&D.Sci as a way of familiarizing myself with the theory and practice of statistics. I went through the first half of the problems -and enjoyed myself tremendously- but have since then stopped posting. This does not indicate I have lost interest, or that I intend to drop the project. Rather, I noticed I didn't have a model for some basic problems (mainly: how to detect and disentagle mixed distributions, and how to model problems with missing or corrupted data) and decided to do some research.

I proceeded to look through my personal library for a treatment of these subjects. I found statistics textbooks of various levels and a number of less academic works, but ultimately nothing that discussed these problems to satisfaction. I literally could not believe that the kind of work I was looking for did not exist, which could only mean I was failing to find it. I then proceeded to flounder uselessly for days, in increasing embarrassment and distress.

Eventually, I remembered the advice of an excellent little book: Umberto Eco's "How to Write a Thesis"1. The relevant advice (the book dispenses much advice, and good) is to find books through other books, not just Google. I started trawling for bibliographic references in the books I had available and found many pleasant surprises, including a book titled "Principles of Forecasting: a Handbook for Researchers and Practitioners". It ended up not being the perfect fit that its title suggests, but it was still very useful.

Happy ending, then; so why have I not published more work? Because this ordeal exposed a complete lack of familiarity with the practice of research, which I've been trying to correct for the last month and a half. Working up to a position of knowledge from a position of ignorance inside a known, finite timeframe is a core rationalist competency, and I expect to make good use of it my entire life. So it makes sense to study this topic exhaustively, and to create an written trail of my evolving understanding. Obviously, it would be complete nonsense to try and hold off from publishing ideas or pursuing other interests until I'm finishined, since this is not a project I can finish before I die. So there is no reason I should stop working on D&D.Sci, or whatever else I want to do, while I take notes on how to do research. This realization came to me slowly and piecemeal, hence the delay.

My plans for the immediate future as as follows:

  1. Publish summaries of my readings so far.
  2. Write a preliminary "how to do research" (ie: the steps I know how to do, plus the steps that seems like obviously good advice, assembled into a coherent checklist).
  3. Finish D&D.Sci 4-6. No more detours or fascinating research questions, we can come back to those.
  4. ???
  5. PROFIT!!!

  1. Eco, Umberto. How to Write a Thesis. MIT Press, 2015. Translated by Caterina Mongiat Farina and Geoff Farina. Original publication: Come si fa una tesi di laurea: le materie umanistiche. Bompiani, 1977. ↩