I’m doing a sandwich thesis for my PhD at McMaster. It involves to a collection of related papers rather than a more traditionally structured thesis. Part of the appeal of taking this approach was that I could measure progress, albeit in big chunks.
When a problem can be broken down into known subproblems, you can calculate a rough approximation of the work involved. It might be a science that involves some heuristics, this year I keep hearing good things about using story points, but generally speaking you can do it.
But what about work that, 1) can’t be broken down into a known number of subproblems, and/or 2) includes subproblems that involve non-trivial unknowns (things that either you or your team haven’t done before, or haven’t been done by anyone before). How do you calculate the work involved in situations like this?
A lot of the time you can still “ballpark guesstimate” the work involved with a reasonable amount of accuracy. But with certain problems, especially problems with an unclear end point, it’s hard to even make guesstimates.
The best solution I’ve found is to give up on trying to estimate in the short term, and start measuring something good. The reason you want to measure is so that you can tell if you’re making progress, and have some idea ‘how much’ progress. Once you have a measure of progress, you can reward yourself for making progress, hold yourself to a specific rate of progress, or adapt your approach if you are not. In the case of problems with an unclear endpoint, once you can measure something, whatever that is, it may allow you to make a better guesstimate of what is possible at all in the maximum time you do have available.
And you need to make sure it’s something good you are measuring, otherwise you could be measuring the wrong kind of progress. In my experience it only needs to be something reasonably good that you are measuring.
So with paper writing for example, you don’t really know the exact end point of the paper when you start. And while words written might not be the best measure of progress, it’s good enough in most cases.
With the app I’m working on now, I would like to fill it with a certain amount of content. I can break the content down into a total number of pieces, and then measure the number of pieces I create each week. The rate varies depending on the piece of content, but I can measure that. And if I find I’m exceeding the maximum time I have to complete the task at the rate I am going, then I may have to adjust my end goals (either the maximum time or the total pieces of content).