by SpAce » Sat May 09, 2020 4:19 pm
Yes, it's unfortunate that there's no standard for the stars. What should it be like if there were?
Personally I think the number of stars should roughly follow SE ratings, with 4 stars being the toughest basics-only level. My local newspaper does it pretty well (as a curiosity, those puzzles are made by the infamous Arto Inkala). Stars 1-4 need only basics, while 5-star puzzles need at least one simple non-basic technique (usually just one). Also the basic techniques required for each level 1-4 reflect quite well their actual difficulty in a p&p context (my timings increase linearly for the first three: <5-10-15 min; after that it depends more on the puzzle and whether I need some kind of pencil marks or not). I think the technique spread is something like this, though it's been a while and I've never really analyzed it:
1 star: full house, box-based hidden single
2 stars: + line-based hidden single, naked single
3 stars: + pointing, hidden pair
4 stars: + claiming, naked pair, hidden/naked triple, (rarely) hidden/naked quad
5 stars: + UR, X-Wing, Turbot Fishes, W-Wing, Y-Wing, BUG, X-Chain, XY-Chain, XYZ-Wing, simple AIC, Swordfish...
(In a software environment with automatic pm, naked should come before hidden of course, but this is the correct order for p&p and other no-pm solving.)
For me, 4 stars is actually the hardest level, or rather the one that is the most likely to require at least some pm. A great majority of the 5-star puzzles is solvable with a simple UR, X-Wing, or Skyscraper, and I can spot those almost always without pm. Not so easy with the hardest basics. Since the basics in the 5-star puzzles are usually easier than in the 4-star puzzles, I almost always complete them faster and without pm. However, this is not a flaw in the star ratings (it's just me).
What do you think? Would something like that work for a universal star system (with stars beyond 5 for more difficult puzzles)? Is something close to it common in other newspapers?