Puzzle Description:

Flight feather. Flight feathers are assymmetrical and usually have downy wisps near the base. This one is all black with a white spot near the tip.

#1: Teresa K (fasstar) on Sep 6, 2009 [HINT]

Oh my! That was so challenging and so fun. I couldn't recognize it until I read the description, but I can see it clearly now.

I used simple edge logic to place one black in the top row and lots of dots. But then I had to do some pretty deep edge logic on the upper left to determine an either-or sequence that let me fill in some squares in Row 5. Then it solved easily. But the beginning really taxed my brain, and was making me think more than just a few moves ahead.

very fun! once you figure out the first column it all comes together, but getting there required a bit of foresight for me.

Ditto #1: ("... the beginning really taxed my brain, and was making me think more than just a few moves ahead") which technically means guessing to me.

After I got the 1 black square in R1, I proved through trial and error that C7 was a dot. By doing that, no matter where else the rest of the 4 went, the first 2 in R2 had to go in 2 out of the first 3 squares in that row.

Nice image.

Thanks for entry # 13 so far. Once I placed the 4 on the first column ... easy.

So, I do a lot of edge logic, including some fairly deep edge logic on the left side (checking the 5 in C18 against the two 2s in C16), and I get as far as this:

And then I don't see a single logical step I can take.

The 4 of C1 can't be at the bottom, cause then there would be no room for the 2 of R19. That helped me.

Hmm... You have to observe that R18 and R20 are mostly dots. in R19 we have a dot in C2 and the "2" must be placed somewhere to the right of there. Since there are dots above and below, it must go someplace where there are two adjacent clues with "1" as the bottom most clue number. But there is no such place, so we have a contradiction.

This is at least one level more look-ahead than we normally use on edge logic. Having had it pointed out, I can certainly see it in my head without having to try it and backtrack, but it seems to be kind of on the bleeding edge of trial and error search. Not sure how I'd rule on that.

Look at the 8 clue in column 7. Imagine trying to place it in rows 13-20. If you look at the resulting black cells in rows 14-17, this causes a contradiction on column 3. Using this you can dot R16C7 - R20C7. The same logic can be done on column 6 to make R16C6 - R20C6 dots. It will solve out normally from there.

Jan: To be honest, I did not get as far as you did on the deep edge logic. When I worked forward from your image is when I could see the way forward.

I would probably classify this as one of the hardest logically solvable 20x20s out there just because it is so hard to get started on this one. Nice puzzle!

Found to be logically solvable by Gator.

Gator, you are kind to find this logically solvable. The reality is that I MESSED UP. I messed up in my test solve, misreading the clues for the last four rows, so that I thought the 4 in C1 was excluded there. It isn't. I myself would classify this as requiring guessing, which makes me weep on the inside, but it's honest.

Sometimes you have to go beyond what you would normally consider an "edge". That nice line of dots in C8 provides a great edge to work from.

The first time I worked on this one, I did not see how I could use the 5 clue in C18 for edge logic. Now it is obvious when I worked on this the second time.

And for the record, we will all probably mess up our test solves at some point or another. I know I have. :)

Thank you Robyn for admitting that (#10). I ditto your comment. Even Jan mentioned how much of a stretch it was to be considered logically solvable.

I actually think Gator's trick for getting forward from my snapshot is fine. I agree with him that this is logically solvable, but ridiculously hard. It requires at least two different non-standard applications of edge logic in non-obvious places. They're clear enough once you see them, but hard to find. Clearly a puzzle for master logical solvers (or carefree guessers).

So a definite "difficulty 5." Now *I* have to try to see if I can do it Gator's way! And I'm going to try to figure it out for myself first.

Still... I replaced the darn thing with one that *definitely* is logically solvable, #6695. I tweaked it so you would need some simple edge logic on that one, but it is DEFINITELY SOLVABLE BY LOGIC using advanced logic techniques.

I thought I had cracked it more than once, only to find flaws in the reasoning that got me there. Learned some deep edge-logic tricks, though. In the end, I had to first use Jan's description of his deep edge logic with C16 and C18, then Gator's hint about the row of dots in C8. It is solvable by Jan's and Gator's logic. I agree this is a 5 in terms of difficulty.

Apparently not too many people thought it was that hard.

It's easy to guess. It's only hard for those of us who obsessed with using only number logic and not picture logic.

I don't know if you've seen the movie "Wordplay" which documents the crossword puzzle competitions run by Will Shortz. The competition is all about who can solve puzzles the fastest. I've often thought it would be cool to do something similar with PBN puzzles, but if solving fast was the goal, then the guessers would win every time, unless you made the puzzle non-representational, in which case half the fun would be stripped out.

Maybe you'd have to make a rule that says "No undo". Mis-mark one cell and you're disqualified. Or maybe each mis-mark adds a minute to your official solving time.

I would say the mis-marks would add ever increasing time. 1 minute for the first one, 2 minutes, 4 minutes, 8 minutes, etc.

I've played the online Picross DS battles, but it only penalizes you for a few seconds for each mis-mark. So it is ruled by the guessers (I stopped playing).

That's kind of what I would expect. In designing this site, I considered putting a timer on the puzzle page so people could compare solving speeds. You could have the site track who the fastest puzzles solvers are. But I decided that was the wrong way to go. It encourages a less thoughtful approach to puzzle solving, and it is way too easy to cheat at. On the whole, it would make the whole environment nastier.

But it might be fun to have an occasional timed puzzle solving competition. People would submit puzzles to the timed competition. Someone would pick a puzzle and announce a time when the competition would start. Competitors would go to the puzzle race page, and just see a count down clock till the puzzle start time. At zero, the puzzle would appear. First person to solve it would win (after their scores were adjusted for mistakes). Afterwards the puzzle would be available as a regular PBN puzzle.

Jan, thank you for NOT putting a timer on this site. Most ot the time I solve puzzles a little at a time, in between crisis calls and lots of interruptions, and while talking on the phone to friends. I come here to relax as an escape from all the stress, and having a timer would add a little too much stress, for me anyway. But I can see how a timed competition now and then would be fun. :-)

Interesting concept Jan, lets see if the master of the WCP organizes this one in which there should be NO checker.

wow. that was tough...and definitely brain taxing; but a nice image.

i found it edge logically sound... start with 4

amazingly tough
fabulous solve

I found this pretty easy to get started, but maybe I assume too much when it comes to edge logic. I looked at the 4 and two 2's in the first 2 columns, and the 212 in C1&2, R2-4 looked like the only place to put them. I was able to place both 2's and three pixels of the 4, and that was enough to get me going.

I thought this was rather easy. When you color in R1C3 using logic, no matter what you do with the 4 clue in row 1, R5C6 always is black. It solves easily from there.

logically solved

Goto next topic

You must register and log in to be able to participate in this discussion.