mith wrote:yzf's solver has four extra eliminations (which get cleaned up with basics anyway):
Three kinds true base cands share same cover house in cross lines: r4c1<>7,r4c6<>8,r5c1<>7,r5c6<>8
Ok, thanks! I suspected as much. Those are of course valid, though a bit redundant, taking both naked and hidden eliminations at once. I'm just wondering why it doesn't list -7r9c1 and -8r9c6 as well, as those cells get locked too. We have four true base candidates and eight S-cells, which means that each S-cell gets filled with a base candidate (2x4).
(As a side note, I really hate that style of listing eliminations. I find it very slow to read quickly compared to the -7 r4c1 (or -7r4c1) style. For
yzf: I would recommend adding the latter as an option. And white space too.)
I've commented on this before, but just to make it totally clear - I am nowhere near being able to comfortably spot these things manually at this point.
Who is? This one was easy because it was in the "default" spot. Btw, can you give a hint about the SExocet? (Btw, I don't think it can be shortened to two letters because 'SE' is reserved. Perhaps three first letters?
) I briefly tried to find it, but didn't succeed. Which base cells is it using? (You can put it in a hidden comment if others don't want to see a spoiler at this point.)
I make heavy use of the various solvers in deciding what to post, checking for alternate paths, that sort of thing
And we appreciate. These are quality puzzles, and somewhat different from what we've been used to. It's good to leave the comfort zone every now and then.
At some point I'll take a break from generating and posting puzzles and focus on improving my solving instead.
Probably a good idea (as long as someone keeps posting puzzles!). I'm pretty sure those activities have a positive feedback cycle. Unfortunately I don't know anything about creating puzzles, which probably limits my solving skills a bit. I think it would help with spotting things like complex uniqueness patterns, symmetries, and how the givens can hint to what patterns might be available, etc. It would be awesome if someone wrote a crash course on puzzle generation, especially with those kinds of perspectives included.
Personally, I find basic fish a lot when I solve, but I'm usually solving electronically with some way to highlight all instances of a digit and full pencilmarks - manually solving on paper, not so much.
Yes, it's definitely easier with digit filtering. Still I rarely spot "real" fishes unless they're very obvious, because I usually move on to chaining right after checking for uniqueness patterns.
In that context, I often spot fish before I see a hidden pair (say), just because it stands out visually to me.
Interesting. Even with digit filtering non-chain fishes rarely stand out very easily to me, although I've improved a bit lately (your puzzles have helped). As for hidden pairs, most of them are actually easier to see without candidates. With pencil marks I usually see their naked counterparts first, even if the latter are much bigger.
That's good, because they've been rarely seen here otherwise. That said, puzzles depending on lots of basic fish steps are are quite tedious for a manual solver, so I consider them more software-oriented (which is totally fine -- nothing wrong with catering to them also). Puzzles solvable in a single step (or few) with a complex or kraken fish would be more interesting (for me).