The Wall of Ugly Plots
The idea is pretty simple. Throughout our process of data analysis, we tend to produce hundreds maybe thousands of plots.
The starting point is always the same - a question. Phrasing a question that will lead you to know exactly what to type or where to look is the hard part of the job.
Now let’s assume that we are lucky and sitting in front of our computer with our favorite hot beverage and we can focus and make plots. What are they for?
These plots are used for testing, cross-checking, analysing, double checking, comparing, evaluating, concluding, fitting, averaging, spotting one mistake, spotting many mistakes, seeing trends or patterns, predicting behaviours, extrapolating, interpolating etc.
The process is very much iterative. Plots are then used as a basis of discussing, criticism, review, presentation and finally publication.
If you ask any experimental particle physicist, I bet you a good espresso that they know without overthinking what is their favorite plot.
For young(ish) people, it can be overwhelming maybe or exasperating to see how far they have to go from the shitty (pardon my french) first plots that they make when they start their project towards the final sparkly ones that make it to a journal or a conference talk.
We started the wall of Ugly plots together with my former PhD student Vitalii a few years ago. We stick in there some of our most horrific buggy plots, if it were not for this wall they would never see the light of the day. However they do, in fact they sit proudly as a daily reminder of the non linearity and toughness of the scientific process.
If one is an aesthetic purist and is allergic to this concept. Let me say that the best way of not producing an Ugly plot is to never make one in the first place, that’s also the best way of never producing anything at all.
I just found out that an other group of people started a wall of Ugly plots in their office and that made me truly happy.