Right now, India is celebrating Diwali. It is the yearly extravaganza of the country, where almost everybody goes back home. But, I’m here in Germany. And, I’m in no mood to work. So, I shall indulge in some wishful thinking this time, in my diary entry.
What should a hypothetical school, that prepares one for research, teach? In no particular order, this is what I would have loved to learn:
Everyone who does serious scientific research, knows about Wikibook, and a lot of trial and error. I can make a simple document, and even make my resume with LaTeX. But, I still cannot make proper research papers. I’ll just have to look around for an existing layout, and plug my content in it.
I wouldn’t want anyone to spoon feed me LaTeX, but I would surely love to better understand the basics of this software before I embarked on more adventurous experiments.
Speaking of experiments, most of the engineering degrees include a handful of courses focused solely on lab experiments in Physics, Chemistry and a little bit of Biology. We do a lot of experiments, fill up endless journals, and if we have a good teacher, even understand the tools and sensors we use. But, I honestly do not remember anyone explaining why a certain experiment is the best approach to prove (or disprove) a certain theory.
During research, one can validate theories by building mathematical models, or running experiments. Given the complexity of some software systems, it may not always be possible to build a good mathematical model. And, even if a model can be constructed, it must be validated in an actual system. And, this is where experiment design comes into picture. I am currently working on a very basic hypothesis in the fellowship, and testing it with some industry wide benchmarks. Some knowledge, beyond basic common sense, would have made it a little easier to cook up the right experiments.
One of the core things I learnt from Feynman’s biography was to understand a system, by building up my own interpretation. In doing so, I often use analogies in completely different subjects, to understand the system I am studying currently. Also, to borrow from the Ladder of Abstraction, I should choose analogies that can help me step up and down whenever necessary. I find that my analogies sorely lack this quality and hence, aren’t always a good fit.