Nasenbär and jsmif, thanks for the feedback. Great that you speak up, so I know what kind of puzzles you want, and I will try to accommodate you.
If some wants harder puzzles or puzzles requiring special solving techniques, please let me know, and I’ll se what I can do. I have generated puzzles requiring
lots of Special Fish, Wings and more than
50 chains (thereof approx. 30 XY-X chains) to be solved.
EDIT: Easier puzzles (Naked and or Hidden Singles only) can also be arranged. But a puzzle with Naked Singles only is only meaningfull when solving with pen and paper!
During the generation process I can take out a puzzle requiring Tough or Moderate solving techniques only as an A version and let the process continue generating a harder puzzle as a B version.
As you hopefully have seen, I have introduced two ratings, one based on required solving techniques and one based on my personal
subjective rating.
How a rating relates to the required solving techniques you can find here.
Nasenbaer wrote:This one took longer than most of Ruud's clueless (but that might have been influenced by the soccer game).
Is probably the case here

. The hardest technique this puzzle requires is NAKED Pairs (no HIDDEN).
The problem is that you IMHO cannot tell how difficult a puzzle is based on the required solving techniques only. If that was the case SCS #3 should be more difficult than SCS #2, which my personal experience and the feedback do not suggest. Here a comparison based on the required solving techniques listed by JSudoku:
SCS #2 SCS #3
398 Naked Singles 364 Naked Singles
142 Hidden Singles 172 Hidden Singles
66 Intersections 73 Intersections
5 Naked Pairs 5 Naked Pairs
When generating a puzzle for the audience here one also has to make sure that the puzzle is somewhat tenacious and does not collapse into Naked Singles only as soon as all the "tricky" solving steps (Naked/Hidden Subsets & Basic Fish) have been done. As Pete correctly has pointed out most Sudokus, especially Samurais found on the internet seem to collapse into Naked Singles only as soon as their one or two "tricky" solving steps have been found. The tenaciousness of a puzzle seems to be related to the number of Hidden Singles and Intersections and that they are evenly distributed among the grids.
Perhaps as a test balloon I should generate a Clueless with Naked/Hidden Singles and Interactions only, but a lot of Hidden Singles (>250) and Interactions (>100) and see how that turns out.