I got a little sick last night when I heard about a mistake, but I'm fine now. There is a long history of wrong puzzles.

First of all I would like to say a few words about some previous posts.

@ Fred76
Yes, seems it will be a skyscrapers month, but this is coincidence. Sudoku City was originally planned for July, but in June we had Logidoku so it is rescheduled for August. Deb wanted to avoid two consecutive sudoku tests based on logic puzzle types. The reason is that these tests were qualified for UK Sudoku team.

@ rakesh rai
Nice building statistic.

@ motris, neerajmehrotra, swaroop2011, thesubro, Ours brun, puzzlemad, Ziti...
Thank you all for taking the free time this weekend for sudoku and many thanks for the compliments.

Nikola

Now I want to thank everyone who take a walk through our small town this weekend. Congratulation to Yuhei Kusui, Thomas Snyder and Branko Ceranic for topping this test. Yuhei was inaugurated as mayor of Sudoku City.

I must thank to Rishi Puri and Zoltan Horvath, they helped me a lot. I have to mention Rishi’s time. He solved all 12 puzzles in amazing 89 minutes, but in the relaxed atmosphere, so he is only a virtual winner of the contest. The test was planned in two hours, but when I saw his time, I decided to limit the time on 100 minutes.

Once again many thanks to LMI and Deb Mohanty for hosting Sudoku City.

Nikola

Edited by Nikola 2011-08-08 2:25 PM

Nikola, Thanks for building a nice city and allowing everyone to roam around.

I myself was planning to create a Skyscraper based Sudoku test, but your "City Theme" was mind-blowing.

1)
Yes, the number of zero scores are bit disturbing. Close to 35% players have zero scores in this test. It happens when the test puzzles are tough.
As Ours brun posted there could be various reasons for these zero scores. In this test, many good players started but didn't submit.
greenhorn has already posted why he didn't / doesn't submit his score. I'm not sure if LMI ratings has a role to play as well.
Ideas on how to improve non-zero scores is welcome. (It is not surprising that fivefold is the only test with >90% non-zero scores. Players couldn't have waited till the end to decide whether they should submit or not)

2)
Regarding players trying to cheat the system using 2 logins : we can't be too concerned because in my mind there is no technical solution to prevent this when the test is running for 48 hours. However, let me assure all that we do our best to find out such cases, and disqualify. We don't post how we detect such cases and who we disqualify, but previous authors (as recently as Something is Missing, Fivefold) can certify that we've disqualified more than one player in each cases.

I think that P07 should be voided. When I realized in the bottom row that it has no solution, I didn´t post any. I don´t think that giving points for incorrect solutions is good idea.

I was very surprised at the results. I’m very happy to win a LMI contest after an interval of more than two and a half years (Mock Test 3 held on December 2008). One of the main reasons for win is that the points of SUM and PRODUCT were relatively low; I’m poor at mathematics...
Nikola, thank you for the beautiful constructions.
Together with wonderful citizens, I’ll do my best for the development of Sudoku City!
(At first, I didn’t plan to participate in the skyscrapers contest in Logic Masters. But I decided to visit that contest as "official duty" )

BTW, I’d like to point out one thing. When I solved 11 grids, there were about 5 minutes left. If I had been a completely rational man, I had hit “Claim Bonus” button immediately and got additional 2 points. (In fact, I felt it was almost impossible for me to solve a 7-point PRODUCT in 5 minutes.) But I hated to abandon a puzzle and selected to try last one, only to lose time bonus points.
We should avoid “Quit early, Get advantage” system, which is a lesson we must learn from Zilina. Thus I think some consideration is needed for putting together partial-solved time bonus system and “Claim Bonus” button.

Anyway, the next puzzle test is designed by the Japanese puzzle makers. I hope many people can enjoy our high-quality puzzles!

Edited by Yuhei Kusui 2011-08-09 8:29 AM

Congratulations once again for winning. Even Branko is in the LMI podium after a long time.
I find it interesting to see new winners, although motris is always a favorite to win every test he plays having won 7 of the 11 Sudoku tests he has played.

The "Claim Bonus" button is new, and the functionality probably needs few refinements.
But, I'm guessing that you suggest we shouldn't have had partial bonus points in this test.

Our partial bonus system in WSC Philadelphia was tied to a belief that the solver had made an honest effort on any puzzle they had a mistake in. They might have lost the puzzle points for a mistake, but they had spent enough time to get far enough on it. I would certainly lean in this direction. I still think the best system for an online test uses some amount of instant grading and reporting at the moment of claimed bonus which might tell solvers with a typo that they have 1 puzzle wrong and would tell solvers that are done that they are done. No blank (early quitting) submissions would ever make sense in this system but at the same time typos wouldn't void all possible time bonus.

The way "Claim Bonus" was designed would have required Yuhei to enter solutions for all 12 Sudokus. So in effect, the bonus points would have been only if all 12 are correct, OR 11 are correct + 1 is incorrect. However, there is nothing that would have stopped one to enter some dummy answers for the 12^th to claim bonus.

With instant grading not in place yet, one change we could do is authors/organizers can decide whether the wrong answer is an genuine attempt and give bonus points accordingly.

So I was looking back at this comment which seemed weird to me as I thought the check box at the start meant 0's were now counting against solvers (unless they report technical problems). But it seems from looking at some top players that indeed a solver can start a test, not submit any answers, and not impact their rankings. Which is what you are getting at above with how some good solvers don't report on a test they know they are doing bad on. I didn't know submitting no answers equaled a free pass; I don't think it should equal a free pass. Perhaps we should re-explore the purpose of the opt-in box at the start. If players can go about a test to not report their bad days, then the rankings for some will be inflated unfairly. If players don't want to report their scores, they can choose to compete unofficially and then not submit.

I thought the puzzles were well constructed and didn't notice the omission in the puzzle with the mistake.

My only problem with the contest was that the Odd/Even puzzle didn't allow for people solving online to input anything in the boxes around the 9x9 grid. Not being able to input these numbers turned a 11 point grid into a 30+ point grid in difficulty. If the people who printed out the puzzles don't understand where I'm coming from, try solving the puzzle blind-folded so that the puzzle becomes a test of your memory to solve a puzzle rather than a test of using logic.

That is really a bad miss, and I've to blame myself for that.
Extremely sorry about that, and we just have to be more careful in future.

Sorry to ask a bit late and probably not in the best topic, but I have a request. I solved Sudoku City online and found the 'Undo' button quite useful. Unfortunately I clicked on it too many times when solving the Product Sudoku and I got lost which correct numbers I have removed (I was too tired :( ).
Would it be possible to add a 'Redo' button as well or it would be an advantage for online solvers because offline solvers can't 'redo' erased numbers from the 'memory' of the paper?

Edited by wicktroll 2011-08-17 12:27 AM

I'm not sure if "redo" feature is a good idea. Like you, I've a feeling that it will be more advantage to online players.

You should be using different colors, if you bifurcate while solving online.