Start Here¶
Coursework and research are a little different¶
In coursework, someone has usually already done the hard work of deciding what the problem is. In research, working out which question is worth asking can be part of the project.
| In coursework | In research |
|---|---|
| You are usually given a fairly clear task. | You may begin with a broad topic or an existing project. |
| The expected techniques are often taught in advance. | You need to work out which method is appropriate and justify it. |
| There is often a known answer or target. | The result may be unexpected, negative or inconclusive. |
| Your work needs to satisfy the assessment. | Your evidence also needs to support the claims you make. |
| Submitting the assignment usually finishes the task. | One result usually leads to another question. |
This does not mean you should spend the whole semester thinking about the perfect research question before building anything. Reading, building and testing will change your understanding of the problem. The trick is to make those changes deliberately and keep a record of why they happened.
So, where do I begin?¶
1. Find out what already exists¶
Read three to five good starting papers. At this stage, the goal is not to read the entire field. It is to learn the vocabulary, see how other people frame the problem and find the references that keep appearing.
Write down anything you do not understand yet. Ask your supervisor (me) questions.
2. Get something that already exists working¶
If there is existing code or hardware, try to build it, run it and reproduce a small known result before changing it. This gives you a baseline and helps you understand the assumptions and limitations.
Try things because it's hard. Try things because you are curious. Try things because it's fun.
3. Keep a weekly research log¶
Write down what you planned, what you tried, what happened and what you will do next.
You do not need the perfect question in week one¶
When you are completely new to a topic, asking a good research question is a process in itself. You need to explore the problem before narrowing it down, then explore possible solutions before choosing what to build and evaluate.
I borrow the Double Diamond from human-centred design to make that process easier to see.