I know it's important as we age to keep challenging the ole grey matter, so that it remains tuned-up and ready to go. That becomes probably more necessary in retirement, as job challenges (thankfully!) recede in the rear-view mirror.
Soon after I retired, I decided I liked the decoding of "Cryptoquotes" which are published six times a week by my local paper. These are pithy or amusing or revealing short quotes, usually of well under 100 letters, accompanied by an attribution. The trick is that they are encoded. Each letter stands for a different one (today a "T" may used for all instances of "K", for instance) and the code changes with each puzzle. The challenge is to unsnarl things that look like "PXOKA GRTYY" to the actual message.
Since I am (was!) an engineer, it is never that easy! I created a template in Excel that lets me rapidly make trial substitutions in the message and easily change them if they prove wrong. I developed a cheat sheet of common 1-, 2-, 3-, 4-, & 5-letter words (and even some longer), and include stuff I found on code-breaking like the relative incidence of the letters, common doublets, triplets, and double letters, common prefixes and suffixes and contractions, and common words with repeated letters. Then I time my performance. And I think of ways to wring more efficiency out of setting up the cryptoquote. (I imagine myself competing against someone with a pencil, who can begin at once; I must first transcribe the quote to the spreadsheet and then link each letter to the first instance where it is used - which usually takes almost as much time as the solution phase.) And finally, I analyze the heck out of my performance!
For 2009, I decoded 271 cryptoquotes. I averaged 12.2 seconds to solve each unique letter, or 7.3 seconds for each total letter. For a normal quote of 66 letters, that's just over 8 minutes!
Charitably for you, I decline to bore you with the ranges, the standard deviations, or the graphs showing my performance improving over time!