What a delight, Daisy! I was inducted into the (dark) arts of the Graun crossword when I began working in a Hackney secondary school in the 1970s by Julia Schilt, one of our maths teachers. my favourites were Custos and Auracaria, in particular, but then Rufus, Altair and Lavengro also come to mind. As time went on, I didn't have the time and shifted to the Telegraph (easier!).
One of my favourite clues was abcdefg....pqrstuvwxyz (5): answer 'water'.
And, I always liked the apocryphal story about the Cambridge don M R James, of whom it was said could 'do' the Times crossword while his eggs boiled. And M R James liked his eggs soft.
Oh that is a lovely clue. If I could run the study again I would ask a different question: "The more elegant clue?" I suspect your example would come near the top.
Isn't there an issue with this, and any studying items across time, that you only have raters from now, not then? So if crossword clues, or mathematics exam, look easier it might be because they are easier, or because they are more aligned with what those doing the judges are good at doing?
To expand a bit, a simple model would be that crosswords (or maths papers) contain a mix of type A and type B clues /problems (which are, in some deep sense, 'equally difficult'). Those trying to solve them have a mix of practice at type A and type B problems. If then problem ratio drifts over time, and judges in 2026 have more practice with the current ratio, then it will seem like the difficulty of the overall crossword/paper has decreased, when really people are just relatively well aligned with how it is now, and relatively misaligned with how it was.
Does this make sense, and is it a really concern for this kind of cross-time study?
Yes, this is a concern whenever you run one of these studies. Luckily, in maths, the rules haven't changed. Maths today is pretty much the same as maths yesterday, and in crosswords, pretty much the same. The rules have hardly changed for the Guardian since around 1957. Perhaps throughout the 50s and 60s they were a little more anarchic, or, as Ashley, aka Boatman, says, libertarian. The libertarian could well lead modern solvers baffled who expect a certain syntax to a clue, but I would say the vast majority of the clues I looked at would be called fair in a modern sense.
The big change over time was moving away from clues where you either know the answer or you don't, to clues where, even if you don't know the obscure Bolivian poet, you can work them out from the wordplay. I think it is this factor that explains why the crosswords of the 1930s were judged to be “difficult”. For me, the biggest confound in the study isn't the age of the judges. It is that the construct of the crossword changed.
The explicit rules might be the same, but that still leaves room for differences in style (which you seem to admit, and which is surely why we know the names of different setters). If contemporary judges are aligned with contemporary style (and historic judges aligned with historic style) then this provides an alternative explanation for any claim that the test has got 'easier' in any true sense.
Extrapolating a bit, things that might defend against this:
1. Task gets harder [not the case for crosswords or maths]
2. Training is consistent [not the case for crosswords or maths]
3. Cohort effects [e.g. if people who learnt crosswords in the 1990s also find the 2020s crossword easier in the same way as people who learn crosswords in the 2020s
4. Item effects [not sure about this one, but if you could identify, e.g. the type A and type B problems you could then track their incidence over time and how the overall task was composed of them,. but that devolves the problem to a whole subproblem of identifying types of problem rigourously]
There might be others!
Do you have the age data on participants to do a cohort analysis on your data?
What a delight, Daisy! I was inducted into the (dark) arts of the Graun crossword when I began working in a Hackney secondary school in the 1970s by Julia Schilt, one of our maths teachers. my favourites were Custos and Auracaria, in particular, but then Rufus, Altair and Lavengro also come to mind. As time went on, I didn't have the time and shifted to the Telegraph (easier!).
One of my favourite clues was abcdefg....pqrstuvwxyz (5): answer 'water'.
And, I always liked the apocryphal story about the Cambridge don M R James, of whom it was said could 'do' the Times crossword while his eggs boiled. And M R James liked his eggs soft.
Oh that is a lovely clue. If I could run the study again I would ask a different question: "The more elegant clue?" I suspect your example would come near the top.
How about this, Chris/ It's too easy to be granted a top place but: gegs (9, 4)... and so on :)
Isn't there an issue with this, and any studying items across time, that you only have raters from now, not then? So if crossword clues, or mathematics exam, look easier it might be because they are easier, or because they are more aligned with what those doing the judges are good at doing?
To expand a bit, a simple model would be that crosswords (or maths papers) contain a mix of type A and type B clues /problems (which are, in some deep sense, 'equally difficult'). Those trying to solve them have a mix of practice at type A and type B problems. If then problem ratio drifts over time, and judges in 2026 have more practice with the current ratio, then it will seem like the difficulty of the overall crossword/paper has decreased, when really people are just relatively well aligned with how it is now, and relatively misaligned with how it was.
Does this make sense, and is it a really concern for this kind of cross-time study?
Yes, this is a concern whenever you run one of these studies. Luckily, in maths, the rules haven't changed. Maths today is pretty much the same as maths yesterday, and in crosswords, pretty much the same. The rules have hardly changed for the Guardian since around 1957. Perhaps throughout the 50s and 60s they were a little more anarchic, or, as Ashley, aka Boatman, says, libertarian. The libertarian could well lead modern solvers baffled who expect a certain syntax to a clue, but I would say the vast majority of the clues I looked at would be called fair in a modern sense.
The big change over time was moving away from clues where you either know the answer or you don't, to clues where, even if you don't know the obscure Bolivian poet, you can work them out from the wordplay. I think it is this factor that explains why the crosswords of the 1930s were judged to be “difficult”. For me, the biggest confound in the study isn't the age of the judges. It is that the construct of the crossword changed.
The explicit rules might be the same, but that still leaves room for differences in style (which you seem to admit, and which is surely why we know the names of different setters). If contemporary judges are aligned with contemporary style (and historic judges aligned with historic style) then this provides an alternative explanation for any claim that the test has got 'easier' in any true sense.
Extrapolating a bit, things that might defend against this:
1. Task gets harder [not the case for crosswords or maths]
2. Training is consistent [not the case for crosswords or maths]
3. Cohort effects [e.g. if people who learnt crosswords in the 1990s also find the 2020s crossword easier in the same way as people who learn crosswords in the 2020s
4. Item effects [not sure about this one, but if you could identify, e.g. the type A and type B problems you could then track their incidence over time and how the overall task was composed of them,. but that devolves the problem to a whole subproblem of identifying types of problem rigourously]
There might be others!
Do you have the age data on participants to do a cohort analysis on your data?