Wadler's Blog

Saturday I had the pleasure to attend Off the Beaten Track, a workshop of POPL 2014.  Every single talk was introduced with the phrase 'And now for something completely different ...'

One talk, by Nada Amin and Tiark Rompf (delivered by Nada), argues that 'papers should expose the sausage-factory of designing calculi, and the minefields in the landscape'.

Ever since Euler (at least), papers in mathematics tend to present a polished solution at the cost of hiding the insights that led to the solution's discovery. Only rarely does one see papers that describe an approach that failed, even though, arguably, knowing what not to do can be as important as knowing what to do.

This leads me to make a suggestion. Every paper is expected to contain sections, where relevant, on
design, implementation, performance, theory, and related work. We should also include, where relevant, a section on 'false starts': research directions that failed to pan out. That is, 'false starts' should be on the checklist of what to cover when first organising a paper. Papers with such material exist, but they are rare; we should make them a common case.

What are your favourite papers that clearly explain a false start?

Labels: Computing, Programming Languages