You are here

Ridiculously long answer keys

The answer keys for the 2nd Round is simply too long, and furthermore it had no basic checks of any kind. Please have something like that implemented, and/or making the answer key more reasonable to enter.

Edit: I partially retract my statement. I did screwed up on the Kakuro on 1 digit. My point in general about the answer key being too long still stands, however.

Yes the answer keys were too long and not appropriate to puzzles. Try to do them easier. I think that the keys could be more universal. For example the puzzles with painting some line like Masyu, Slitherlink, Snake can have the answer key same maybe "the longest sequence in all row from top to bottom". From the GP 2 I have about five types of answer keys not eleven like you. Despite the answer keys I enjoyed the competition and puzzles in Slovak landscape :) Good job Matus and Matej!!

Thanks for the feedback.

I experimented with using an answer key for Snake similar to that used for Slitherlink and Masyu; however, the problem was that there are too many "close but incorrect" answers with the same answer key for Snake. This is because of the nature of the puzzle -- a Snake puzzle can allow for sections of the snake to move left or right without violating the constraints too much, whereas a Masyu will completely fail its constraints if you try to move sections of the loop one space in any direction.

In general the large grids made the puzzles more interesting but made it very hard to have an answer key that actually reasonably checked for puzzle correctness without being long.

I can assure you that next month we have much smaller grids and more manageable answer keys.

I second that the answer keys were chosen inappropriately. Each of the puzzle had its own way how to encode the black/white - occupied/free cell, which - I guess - brought in lot of point reductions (including myself), just because the player was not careful enough to transfer the correct solution to the expected answer key.

TiiT's picture

The other thing is the length of the answer keys. If you spend 75 minutes for puzzles and 15 minutes to typing in answer keys, then the proportion of submitting answer keys is more than 15%. So, your result is 83,3% of puzzle solving skills and 16,7% of typing in the answer keys skill. Some people are faster, some slower in typing in numbers and letters and counting cells. It doesn't seem right to me.

TiiT: I'm afraid that this is a problem of those grids like Wei-Hwa already mentioned...
You can solve nice grids (inside interesting shapes) but with those difficult answer keys or you can solve more common grids with much more easier answer keys...

TiiT's picture

I just think that there was a chance to optimize this. Why do you need numbers from every column in kakuro? Why do you need the longest gap between 2 shaded cells from every column in letter hitori? Same thing with every circeled puzzle and maybe some others too. Although some of the keys were reasonable too.

You should have seen some of the answer key systems on the earlier phase -- Kakuro just asked for two rows and so the key was twice as long. Anyway, if we hadn't asked for a number in every column, I bet instead of seeing complaints that the key was too long we would be seeing complaints that "oh, I missed one of the circles, can I get partial credit"? I think the lesson here is that large grids do not do well with the online answer-key system. Month 3 is fine, but I hope that Month 4 and 5 do not do large grids -- and if they do, I will try to figure out how to make shorter keys.

Is it possible that the Kakuro would've been solvable with a circle in each row instead of Column? Ok, maybe not Kakuro, but I think this holds true for say a Hitori. I personally thought having the same key for each row gave around the same information, but maybe I didn't pay enough attention to that.

And for Kakuro, maybe marking the entirety of 2 columns instead of 2 rows would've been ok. In general entering a sequence of digits is probably easier than comparing each column (Obviously one can represent the numbers below the column, but I'm never a fan of shifting thought process to an answer key during a solve). Again, I think that it's really difficult to see if there will be unsolved bits.

The best solution would be to not have irregular shaped grids, for the GP (regardless of puzzle quality). They are better left for the online contests that do not have a goal to encourage beginners. Or at least, if you do have irregular shaped grids, try and make them smaller in size (I realize that probably wasn't possible with the theme here). Like Tiit said, the percentage of time spent on keys should not be that high, and I'll add another aspect to it - If some keys don't take long (e.g. Masyu, Irregular Battleships) and some take extremely long, then puzzle selection comes into play based on keys rather than solving preference.

I agree with not having irregular shaped grids. However, it was pretty obvious that there was a theme going on in Slovakia's submission, so it was very hard for me to say "no, you must start over with smaller grids"... although I did think about it once or twice!

The theme was impressive, and yes it'd be too hard on anyone to tell them to redo the contest considering how young the GP is. I just think these things can be a lesson for what restrictions to add to future GPs. Again, I think it was a good idea (and really good execution) for a general online contest. Just not a contest where one of the goals is to encourage beginner participation, and where the importance of it all intensifies the scrutiny.