Introducing ‘Minimum Research Outcomes’
‘If you don’t plan properly, you will do the wrong work. It’s important to make a careful design before starting to make things.’
‘Just do something! If you start with planning, you’ll end up with a load of design documents in a filing cabinet, but you’ll never make anything!’
These two arguments have been contending within the software community for many years. Have a look at the Agile Manifesto for part of the story. A useful discovery has been that incremental, step by step delivery, with useful outputs created along the way, means that you can be sure your project will indeed eventually deliver something!
The vulnerable period, therefore, is the bit at the start of your project, before your first, smallest useful result. This first useful result is called a ‘minimum viable product’, or MVP. After that, you can gradually improve on something, fixing a bug here or adding a bell or whistle there. But until you get to your MVP, things are scary.
What does this mean for our interdisciplinary research project, which needs to combine a reflective, patient, scholarly approach, with producing real outcomes, historical analyses realised as computer programs? We cannot afford to think carefully about our approach for four years, then spend the last year writing up!
We are experimenting with an approach we call ‘Minimum Research Outcomes’: we have written simple, short design documents for each of our first outputs, to take about the first nine months of the project. We hope that these will get us quickly to the point where we have done something useful, and to prove that our approach is working.
We know that we’re cutting a lot of corners in doing this. The minimal methods are full of problems, interesting biases we know we need to address, and choices we’re not sure are properly justified. Once we know we’ve really made something, then we can start gradually improving it, filling in the missing parts. But we hope that by starting small, we can be sure our approach is going somewhere.