Kerry JK

@Dean speaking here as a learning constructor (not yet at the accepted stage, but far enough along to get a good look under the hood - check out the excellent blog series linked at the bottom of the Wordplay column for a trip down the rabbit hole), I've come to suspect this is a side effect of the various software packages we use to put grids together. In my case that would be the free online app Crosserville, but they all share datasets. Put simply, software is an invaluable tool that speeds up the process of finding interesting paths from an initial concept, but also a false friend that can dogwalk you into particular sets of fills that everyone else is using at the time. There was a phase last year when I saw the same filler words and clues being repeated on successive days, knew how annoying it was as a solver and resolved to do whatever I could to avoid it in my own work. My current relationship with Crosserville's suggestions has a strong element of Man Vs Machine and I've noticed a real difference between grids I made mostly on paper and those I developed electronically. When and if I get to the point where an actual editor works with me, I will discover whether they substitute my inventive but rubbish filler clues for stock fare, which I could see happening in a busy boiler room. Though at this level it would be like entering a Michelin star kitchen and being made to heat up ready meals.