Crossword SNITCH

A quantitative guide to the degree of difficulty of the Times Cryptic Crossword

For anyone interested, here is some background on the journey leading to the creation of this site.

Beginnings

Having attempted the Times Cryptic Crossword most weekdays for some years, I began looking at solving times to answer the question: is the Friday crossword harder than the rest? And is the Monday one easier? This seems to be the feeling from my experience and from the Times for the Times blog, although I understand that it is officially denied. Being a certified nerd (with an Electrical Engineering degree and 30+ years of telecom software development experience), I wanted to answer this question with hard data.

I noted that TfTT bloggers tended to use the Time Crossword Club leaderboard to check their time compared to the rest and use this as a measure of difficulty. So I began to collect the leaderboard data and process it in Excel spreadsheets. It needs some filtering - there are clearly some members who post unrealistic times (the neutrinos, as Tony Sever named them), and even for regular solvers there seems to be an odd result (one that's far too short or too long). So I ended up concentrating on a list of 155 solvers who regularly solved the puzzle and seemed to be genuine. And even for these reference solvers I excluded any outlying solving times in calculating an average.

Other than finding a way to remove outliers, the method is reasonably simple: work out an average solving time for each of the reference solvers and compare their solving time for an individual puzzle (if they solved it without error) to their own average time. A score of 100 would be average time for them, 150 would be one and a half times their average, 67 would be two thirds their average time. By averaging the results across all the reference solvers, a score for each puzzle could be calculated.

For the 15x15 crossword the results were reasonably conclusive. The group of Monday puzzles had a score in the 80s and the Friday puzzles had a score over 110. So on this measure the trend was confirmed. You can see the latest data on the Trends page.

Progression

Having answered my basic question, I thought it would be interesting to share the results on:

  1. the actual score for each crossword; and
  2. how hard a particular crossword is, on the day of the crossword itself.

The first is almost required to provide more evidence around the conclusion. If you can see the scores for the crosswords on each day of the week, it's easier to convince yourself of the trend. This is especially true since there is some variability. Not every Friday crossword is harder than every Monday crossword - see the week of 10 October 2016 for an example of where the Friday puzzle was the easiest of the week. But they are harder on average, as the Monday puzzle is easier on average.

The second item grew from my eagerness to see how my experience in solving the crossword compared to the NITCH calculation. It also matches what some of the bloggers do in practice - compare their completed time to that of others. So for each crossword I provide a list of the reference solvers who complete the crossword without error and their time compared to their average.

I had other excuses for continuing:

My wife and children have been gracious in accepting these character flaws and even shown some interest in the results.

In any event, this has resulted in the current site, which updates all the data automatically every 10 minutes. I hoped others might find it useful and feedback from the TftT community has confirmed this. I now tend to visit it daily on my way to the TfTT blog.

Quick SNITCH

For some years people had been requesting a SNITCH for the Times Quick Cryptic. I finally managed to find some time to set this up in September 2023. There may be the occasional odd reference intended for the 15x15 crosswords in the Quick site, so I apologise for these in advance.

Conclusion

I am always interested in feedback and try to respond to queries that people raise with me. The official denial that there is any attempt by the Crossword editors to make the Monday puzzle easier or the Friday puzzle harder is especially interesting, given the SNITCH data. I think my measure is a reasonable one, so should provide a prima facie case to say that there is a trend. And it aligns with our perceptions as solvers. But I'm open to others' thoughts on this.

Is there perhaps some other explanation? Do we as solvers lose energy during the week, so attack the Monday puzzle with a fresh mind but struggle to keep the same sharpness by Friday? Of is there some groupthink at work, where we are led in our performance by our own expections? Or perhaps the editors themselves are the ones unconscious of their actions, unwittingly putting an easier one in early in the week and leaving the more challenging ones till Friday? At least there is now some evidence to add to the debate.

About me

As a Times Crossword Club member (under the name starstruck) I was stalking the Times for the Times blog for some years. I appreciated the comments from its expert solvers to improve my own solving skills. This has happened in stages: completing most crosswords, then completing them without aids, then improving my speed. Around 2015 I started contributing to the TftT comments and I've been a regular (if inconsistent) commenter on the puzzles since.

Being based in Australia (and never having lived in the UK) solving is occasionally made harder by some missing local knowledge. But based on the performance of Galspray (the "Australian Magoo"), I should at least be aiming for a 17-minute average. I expect this will remain but a dream; with a full-time job and many distractions. It doesn't, however, destroy my enjoyment of battling through the daily challenge of solving the crossword, no matter how long it takes.

So if you find this site of any benefit, please consider it as thanks to the gurus of the Times for the Times in sharing their reflections and expertise for the rest of us. You can also make contributions towards the hosting costs at this Buy me a coffee page.