better in making sense to the contestant.It was confusing.

better in making sense to the contestant.It was confusing.

Hi, thanks for this great test.
I liked the puzzles a lot, although for the first time I had problems with the solution code: I really had to search for the very hiding letters of the alphabet (in Yajilin for example my loop goes once straight over the letter so that it got invisible). And to be able to read the Kakuro-clues I had to install another printer.
Nevertheless a very good test.

Great test Tawan. Your puzzle are really enjoyable, and the idea is very elegant! Thanks for your puzzles!

Really nice test. My only problem is that I solve on computer using paint and stuff, so in the fusion ones I had to use two different tools and keep switching, which put me in a state of confusion throughout the test :P Hopefully I'll get a printer soon...

Beautiful puzzles (the kind of twisted classics I wish the WPC had).

Horrible answer entry mechanisms. I had to look back at the pdf in almost all cases where gray letters were used.

Very nice test!
Can someone send me the entire solution of Searchdoku?

I had the same problem. I thought it was only an issue with my printer, but apparently I wasn't alone.

I only managed to solve 10 puzzles (got stuck on the majilin...) but they were very good. A lot of them combined types that I'm better at with ones I can barely do at all, which was a turnoff for the test because I didn't have time to practise, but with my rate of solving I didn't even get to them.

Good test, thank you.

No more participants at this time. But the score page is not frozen yet. As many players mentioned, answer keys were problematic. So we need some extra time to revisit all wrong submissions. I expect that the complete results will be out in 3 hours from now.

Please check attachment.





Attachments
----------------
SearchDoku.png (80KB - 3 downloads)

You've missed marking the key word BADMINTON which, without E's and not outside the grid, needs to be one of the sudoku diagonals. The all E exterior made for a very interesting puzzle; one of my favorites on this test.

Yes, I missed marking BADMINTON

Tawan has sent me the solution booklet, but this puzzle is missing there. This image is from my solving, and hence the missing line on BADMINTON.

Bit of story there. The original SearchDoku was too cumbersome and less interesting. I was discussing with Tawan to drop the puzzle from the set (the set as a whole looked very difficult to me even otherwise). But then Tawan came up with this very interesting puzzle within couple of hours, and there was no way we were excluding the new one. This one uses both the puzzles (Word Search and Sudoku) nicely. There is a clear logical path as motris shared. One of my favorites along with Anglers and Fishermen at War.

My only concern was (/is) language-neutrality of this puzzle. Just for completeness, the original puzzle used Swiss words.

Detailed results now available at - http://logicmastersindia.com/M201111P/score.asp

motris is one only player to have solved all 24 puzzles and he takes the first position. MellowMelon and xevs both have solved 22, they take the next two positions. The scores are very close after that.

243 players (a whopping 52 from the USA) participated in this test, and 194 players have got non zero scores. Thank you everyone for participating.

Thanks Tawan for an entertaining set of puzzles.

About answer key issues - The most apparent problem was the loop puzzles in Set7. With light grey letters being overwritten by the loop, many players missed couple of letters, especially I and J. As it can be seen from the score page, there are lot of manual overrides for these 3 puzzles.

I don't think the answer key itself was a problem there. We had used it rather successfully in 2010 Decathlon. The marking of the letters were the real culprit.

Phew, Puzzle Fusion is over.
Deb probably hears this after every test but I owe a huge thank to him for running Puzzle Fusion and spreading it across many forums. I was surprised by the number of participants since this was held very shortly after the exhausting WPC/WSC.

I’m happy to see that no puzzles broke, although several did just hours before the starting time. As Deb mentioned, the original Searchdoku had a Maori-named New Zealand Birds theme which made the grid rich in vowels and most placements were done on ‘feel’ than logic. I surprised myself at how better this one worked out.

About the disastrous answer keys, I really didn’t want to use length of line segments in row/column (in Japanese Puzzle Land) since this is very prone to guessers. I agree with Deb that the real culprit here is the faint lettering and how it is badly placed in the centre. Several people asked for claims in the Akari (namely the column 4 light bulb) that we really couldn’t give points for. Of course – that row was chose on purpose.

In all, this has been a great experience for myself and will be a huge step in improving my future in puzzle construction. I thank everyone who took their time in taking this test.

Thanks a lot for the puzzles, I had a lot of fun (and did quite well). There were a number of puzzles I could mention… The Majilin had a great logical solve, using both rule sets and their interactions. LITS2 is an interesting variation, the puzzle worked really well. I remember being happy I chose to do the Kakuro, a type I don't usually enjoy much. Most likely, there's an easier that better solvers would have spotted, but I found the way I broke into that Kakuro very satisfying.

The answer key problems have been mentioned. Probably, small thin black letters in the corners of cells would work a lot better.

One other thing: I felt that points per minute varied extremely between the puzzles. This is certainly in part because I'm better at some types, but I had the impression that even disregarding that, there was a large variance. What do other solvers think?

Cheers
Rob

Thanks Deb! I had illumation about the word badminton only yesterday evening before sleeping! :)

I noticed this some during the test, after thinking I'd sunk way too much time into the Kakuro/Kropki part to finish, but I did other parts much faster to get back to the right pace. Still, my last two puzzles - the Nansuke and the Akasuke felt the hardest of the whole set and took me almost 25 minutes. I'm not surprised those are the puzzles a lot of the 22/24 finished solvers had left to go.

I think this is my breakdown by section based on my answer test timing and knowing I worked some on LITS^2 before skipping and coming back:

Section 1 - 80 points - 22:52 or 3.5 points per minute
Section 2 - 55 points - 10:48 or 5.1 points per minute
Section 3 - 75 points - 25:13 or 3.0 points per minute
Section 4 - 55 points - 16:17 or 3.4 points per minute
Section 5 - 50 points - 7:10 or 6.9 points per minute
Section 6 - 55 points - 12:43 or 4.3 points per minute
Section 7 - 65 points - 11:10 or 5.8 points per minute
Section 8 - 60 points - 11:45 or 5.1 points per minute

I didn't break out specific puzzles and I'm sure the Anglers (48 seconds for 12 points per minute) and Masyu (except for answer entry, also about 1 minute) were my highest value solves but you can see the Tapa/Minesweeper was the easiest section and Akari/Nansuke was the hardest section for me with about 2.3x difference in point value. Because some puzzles rewarded intuition more than logic, having such variance is not particularly unusual although I might have changed the value of some puzzles. Only one solver did the Akasuke (uvo's wrong answer is close though), so I might venture that at least that puzzle was harder than just a 40 given the really low success rate.

Does anyone know the break in for the Akasuke?

Well, next time I´m going to make sure my printer is working properly. Most my puzzles missed a grid. I couldn´t see any of the kropki circles or futoshiki symbols.
I managed to solve enough considering the missing grids. The fisherman at war was really fun and reminded me of the anglers/pentominos puzzle in your other test.
I also really liked the Nansuke opening, but that was as far as I got. I couldn't finish it in the test, I didn't see a way to continue past the opening.

I'm all in favour of Maori-named New Zealand bird themes. Kiwis in particular come to mind...

My intended break-in for Akasuke was that 2003 can only be placed in either of the two right vertical rows. Then 0103 and 1103 needs to be placed in 2 of 3 possible horizontal rows (not the two columns where 2003 could fit since that would make it impossible to fill in 2003) - and there would only be one possible combination to fit the two hanging 3s. From there, the next leap is looking at 0200, 0201 and 0202. At which point the grid would have 2 of 3 places where there's a 020X pre-determined. Afterwards, using simple Akari rules, everything should fall neatly.

After solving the centre, the Nansuke's focus was that all 5-digit numbers start with 5,7 or 9. Also to be noted is that all 5-digit numbers starting with 7 has 8 as their third digit. Knowing this, simple logic is enough to finish Nansuke.

With the Searchdoku, after solving the Wordsearch part of the puzzle, you will notice BADMINTON is missing and since the sudoku grid is surrounded by E's, the BADMINTON (with two Ns) must be placed diagonally. Using crossings from the Wordsearch (especially TEAMMATE), the puzzle simply becomes a sudoku.

The 6 highest rated puzzles were all from the fused types, which is great news to me, because that means the fusion was enjoyed by the solvers.
Thank you once again.

I think the first part of this description is slightly wrong, at least from my solving experience. The either/or on 2003 was also my first step, but then I can't follow up as you say to place the other entries with 3's. There aren't three posssible horizontal rows to take the 0103 and 1103, just two with the second digit 1 impossible in two of the four horizontal choices that can take a trailing 3, and you can't put both of them in these horizontal rows as they share too many squares. Instead the second of 0103/1103 has to be in one of three VERTICAL spots on the left of the grid or in the middle of the grid. I don't think you can even say that one of them has to be horizontal, to be honest, as there are valid arrangements that use the top-left vertical, the middle vertical, and one of the two far right vertical, for all of 0103/1103/2003, only failing when you pack in the rest.

I didn't find any trivial logic to eliminate all these possible placements of either/or pairs, but I did think the easiest situation to experiment with to eliminate one spot was the top-left most vertical entry, which with a horizontal option forces itself to be a 1103 as it only leaves a spot for a 0103 going in just one place. Fortunately for me, this ended up being the correct situation to get to an answer. I've struggled in looking back at it since to prove it is the only one, but satisfying the 2's is indeed the most challenging part. I just don't see what the "next leap" is actually meant to be with them, unless there actually is a good way to get the 1103 and 0103 down as I guessed them to be.

In other words, I think I got a little lucky, and this puzzle probably is the hardest of the set as there is no trivial first deduction.


Edited by motris 2011-11-22 9:28 PM

In terms of issues such as the legibility of text, both for entering solution keys and also the tiny kakuro clues (which I too couldn't read that clearly once printed, even though I used a laser printer), now that there are so many excellent puzzle competitions on LMI might it not be possible to summarise for authors what text sizes and shades of grey "work" and which don't? Also important are page sizes and margins - this test had huge white margins which were unnecessary and made things smaller than they needed to be, and I've also noticed page sizes varying between authors (perhaps Letter versus A4, but I haven't checked to see).

Maybe, when a page is printed without any scaling, there could be a suggested minimum size of clue text and maybe even a minimum size of cell for sudoku-esque puzzles? And page sizes could be chosen so that they are scaled equally when printed on both A4 and Letter? (i.e. pick the minimum of the two sizes in each dimension so that whether you print 'to fit' or 'at original size' you will get a test that prints at a consistent size and is not cropped). In that way tests would generally print with content at the same size no matter where in the world the solver is. It's easy to do this if you're aware of it but possibly something that puzzle authors don't always think about.

Edited by Gareth 2011-11-23 4:04 AM