One year of PaperPict: what we learned about figures
Eleven months, six tools, and a lot of figures. Some of what we learned we expected. Some of it we didn't.
Read article →The experiment
This was an informal test inside our own team, not a study, so treat the numbers as illustrative. We took a five-step mechanism figure and asked a handful of colleagues with research backgrounds to apply a typical reviewer request: rename one protein, change one arrow from activation to inhibition, and switch the whole figure to a colour-blind-safe palette. Each person did it once on a flat PNG with an image editor, and once on the layered SVG in PaperPict.
What we measured
On the flat image, people took somewhere between half an hour and an hour, and a couple gave up on patching pixels and regenerated the figure from scratch because the patch looked worse than starting over. On the editable figure everyone finished in two to three minutes, and nobody regenerated anything.
Why the gap is so large
A flat image makes you reconstruct everything around the change: which font was that label set in, what exact colour was the arrow, how do you remove an arrowhead without leaving a scar. An editable figure already knows all of that, so the correction is one property on one object. The numbers are rough, but the shape of the result won't surprise anyone who has revised a figure by hand.
The hidden cost
Corrections that are expensive get skipped. More than one person admitted that on a real manuscript they would have argued with the reviewer rather than make the change. When the edit costs two minutes, the argument is not worth having, and the figure gets better.
What this means for a lab
Multiply by the number of figures, revision rounds, and co-authors. The time saved is real, but the larger effect is that figures become living documents that improve through review instead of artefacts frozen at first submission.
