WPF Puzzle GP 2020

Information and discussion on Puzzle Competitions
Puzzle_Maestro
Posts: 75
Joined: Mon 11 Mar, 2019 5:23 pm

Re: WPF Puzzle GP 2020

Post by Puzzle_Maestro » Tue 17 Mar, 2020 7:41 am

Official results for round 3:

10.(!) Neil Zussman 654
27. James McGowan 579
43. Freddie Hand 531

Congratulations to Neil on a top ten finish! Only four Japanese solvers to beat now...

I cannot make sense of the formatting for the domino puzzles (puzzle 12 below, puzzle 13 above). There is absolutely no benefit to doing it this way and it is guaranteed to trip up almost everyone. I hope something like this doesn't happen again.
Last edited by Puzzle_Maestro on Fri 03 Apr, 2020 1:55 pm, edited 2 times in total.

Feadoor
Posts: 177
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Puzzle GP 2020

Post by Feadoor » Tue 17 Mar, 2020 7:56 am

While on the topic of formatting... It baffles me that the four puzzle types with dashed grid lines (Domino Search, Pentomino Division, Hitori, Statue Park) all used different styles for the grid lines themselves! Four different thicknesses, four different shades of grey... The Hitori ones were so light and thin that they didn't appear at all on my printed copy! And surely at least Domino Search and Pentomino Division, which are both about dividing a grid into polyominoes, should have used the same style, even if the others didn't necessarily need to?

I was a bit disappointed by the Tower Defence puzzle... It's a type with some interesting possibilities (I convinced myself of this by constructing a few examples before the contest) - but the competition puzzle ended up with all circled numbers given, and with most of them being satisfiable in only one way (up to symmetric placements of a few numbers). The points were just a side-effect of its size :(

On the other hand, I really enjoyed the Nurikabe and Nanro puzzles :D

detuned
Posts: 1953
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Puzzle GP 2020

Post by detuned » Tue 17 Mar, 2020 8:24 am

Yes I got caught out by the domino puzzles being the wrong way around. Hopefully they’ll realise I have the right code in the wrong place and give me the points.

From the forums, it seems Kota also has this problem so I don’t feel quite as daft for missing this

puzzlemad
Posts: 902
Joined: Mon 21 Jun, 2010 10:20 pm
Location: Oldham

Re: WPF Puzzle GP 2020

Post by puzzlemad » Wed 18 Mar, 2020 8:41 am

I haven't seen the puzzles but would guess on the formatting that they have been created by different people, possibly with different methods of drawing the grids up.

Feadoor
Posts: 177
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Puzzle GP 2020

Post by Feadoor » Wed 18 Mar, 2020 9:39 am

I think my earlier comment comes across as very harsh and overly critical of people who have worked hard to prepare this contest. If anyone involved in preparing the round reads this, I'd like to apologise, and let you know that I really did enjoy the puzzles. Thank you for your efforts.

detuned
Posts: 1953
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Puzzle GP 2020

Post by detuned » Wed 18 Mar, 2020 3:00 pm

I’m relatively sure wei-hwa draws up all the grids himself for the GP booklets... of course if you happen to have lots of different templates for different styles then maintaining consistency is difficult (although still very much desirable!)

Re the swapping of 12 and 13, there is an update:

https://gp.worldpuzzle.org/content/puzz ... ts-delayed

it’ll probably see Neil knocked out of the top 10 for this round, with Kota jumping ahead.

mwhite
Posts: 5
Joined: Tue 28 Feb, 2017 7:37 am

Re: WPF Puzzle GP 2020

Post by mwhite » Sat 21 Mar, 2020 6:50 am

I managed to catch the 12/13 issue when entering them. Sadly I missed that the Hitori needed the second shaded block as I didn't read the instructions as carefully as normal because of the Open. It would have been my highest score on WPC too :oops:

Puzzle_Maestro
Posts: 75
Joined: Mon 11 Mar, 2019 5:23 pm

Re: WPF Puzzle GP 2020

Post by Puzzle_Maestro » Mon 23 Mar, 2020 7:47 am

Official results are out for round 3. Neil holds on to his number 10 spot.

Puzzle_Maestro
Posts: 75
Joined: Mon 11 Mar, 2019 5:23 pm

Re: WPF Puzzle GP 2020

Post by Puzzle_Maestro » Sat 04 Apr, 2020 8:31 pm

The instructions booklet for round 4 of the Puzzle GP can be found here: https://gp.worldpuzzle.org/sites/defaul ... IB_v01.pdf.
Apparently the author goes by the name of authornames, but I expect it is Deyan Razsadov.

I've noticed that this round has striking similarities with a previous round by the (presumably) same author, which can be found here: https://gp.worldpuzzle.org/sites/defaul ... Round6.pdf. Some of the puzzles which appear in both are Four Winds, Snake (Ends), some form of Skyscrapers, Arrows, Magnets, and Japanese Sums (Hitori). Might be very useful for practice, if you did not participate in that round.

Puzzle_Maestro
Posts: 75
Joined: Mon 11 Mar, 2019 5:23 pm

Re: WPF Puzzle GP 2020

Post by Puzzle_Maestro » Thu 09 Apr, 2020 7:13 am

The Instruction Booklet now has some points. Turns out that if you want to aim for 600+, you are going to have to attempt the monstrous 125-point arrows...

Feadoor
Posts: 177
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Puzzle GP 2020

Post by Feadoor » Thu 09 Apr, 2020 2:46 pm

Puzzle_Maestro wrote:
Thu 09 Apr, 2020 7:13 am
The Instruction Booklet now has some points. Turns out that if you want to aim for 600+, you are going to have to attempt the monstrous 125-point arrows...
One of the positives to not being good enough for that sort of points total is that it means I am free to discount puzzles like that :)

Puzzle_Maestro
Posts: 75
Joined: Mon 11 Mar, 2019 5:23 pm

Re: WPF Puzzle GP 2020

Post by Puzzle_Maestro » Tue 14 Apr, 2020 6:36 am

Official results for round 4:

17. Neil Zussman 662
23. Freddie Hand 638
27. James McGowan 605

This was an easier round so a lot of people finished the test: 16 in total, with Neil coming just behind the finishing cohort. Indeed, Ken Endo would have scored about 1000 had he not made an unfortunate answer key error on puzzle 4 (the four winds). Staggering stuff as usual from Ken.

I was two seconds away from submitting the striped snake, which I was quite disappointed about. Nevermind.
Last edited by Puzzle_Maestro on Thu 30 Apr, 2020 4:16 pm, edited 1 time in total.

Puzzle_Maestro
Posts: 75
Joined: Mon 11 Mar, 2019 5:23 pm

Re: WPF Puzzle GP 2020

Post by Puzzle_Maestro » Tue 14 Apr, 2020 6:51 am

I should add that a couple of the point allocations felt a bit suspect; even if some bifurcation was necessary, the 84-point Yajilin still felt somewhat overvalued. (50-60 would have been more sensible, in my opinion). The 125-point arrows turned out be okay, there were a lot of small numbers after the initial logic so I was able to bifurcate easily (turns out there was a nice logical break-in concerning R23C12, but seeing that would probably have taken longer).

detuned
Posts: 1953
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Puzzle GP 2020

Post by detuned » Thu 14 May, 2020 9:30 pm

Round 5 is happening this weekend

Puzzle_Maestro
Posts: 75
Joined: Mon 11 Mar, 2019 5:23 pm

Re: WPF Puzzle GP 2020

Post by Puzzle_Maestro » Tue 19 May, 2020 5:37 am

Preliminary results for round 5:

17. James McGowan 541
19. Freddie Hand 540
45. Neil Zussman 435

Bad luck with the answer keys Neil - there were more errors across the board, but nevertheless it’s annoying to lose 133 points that way.

As expected, people found that to be a very tough round. Even some of the easy puzzles (esp. Fillomino) turned out to be rather pernicious.

For me, that round had some great solving interspersed with terrible solving. I raced through some of the puzzles (including the hex ones, which I am surprisingly comfortable with) but made silly errors in password path, cipher Fillomino and Tapa (borders), breaking them all near the end.

EDIT: The answer key to puzzle 9 (the tapa borders) was incorrect. The results have now been adjusted accordingly. The statistics show that just under 50% of submissions for this puzzle were correct: there must be some psychology behind this extraordinarily low number.

detuned
Posts: 1953
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Puzzle GP 2020

Post by detuned » Tue 19 May, 2020 9:25 am

I think bad choice of answer keys also comes into play, but it does seem quite low even so! You could have easily have chosen columns from the tapa borders rather than rows. The key to the second galaxies was even more painful.

Puzzle_Maestro
Posts: 75
Joined: Mon 11 Mar, 2019 5:23 pm

Re: WPF Puzzle GP 2020

Post by Puzzle_Maestro » Tue 19 May, 2020 9:34 am

I do agree that columns would be on the whole more desirable (especially for the second galaxies), but at the same time that might have resulted in more people getting marked correct with a small error. In this case, I believe many people erroneously shaded one of R9C17 or R9C18, not realising that the 13 clues had already been satisfied by the shaded cells at the bottom of the last O. This was probably caused by the presence of borders, but I can't say for certain.

I think the grid shape and size of the puzzles naturally gives rise to some nasty answer keys. The Yajilin answer key was horrible, taking me over a minute to input, yet I can't see any potential improvements to it.

detuned
Posts: 1953
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Puzzle GP 2020

Post by detuned » Tue 19 May, 2020 12:10 pm

You could do length of line segments for yajilin like for other loop puzzles. One of these days I’m going to try putting together a database of what keys works best for what puzzles.

Perhaps those puzzles were susceptible to a few final sting in the tail moments. I got caught out on the scrabble like that.

I’m also not a huge fan of the way some of the puzzle GP rules are stated. I mean fair enough, I’d not done any preparation for this round (you know, like even looking at the IB let alone trying examples of puzzles new to me), but I had a moment looking at the hex number placement puzzle where I did not tweak what was going on from the verbose ruleset. I was looking at the adjacency condition not fully realising the full power of the deductions on cells at distance 2 as well as distance 1, and wondering how this thing could ever be unique.

It made sense after time, but I think it’s a reasonable goal for instructions to convey the gist of the puzzle so that someone really experienced can skim the rules. Maybe in this case it was just me and my lack of preparation, but it does seem to me that things could be clearer.

Feadoor
Posts: 177
Joined: Thu 17 May, 2018 7:08 am

Re: WPF Puzzle GP 2020

Post by Feadoor » Tue 19 May, 2020 3:42 pm

For a dissenting opinion, I find the puzzle GP rules to be written very clearly, and I think it's been particularly noticeable this year that there has been a push for consistent terminology between puzzle types and consistent phrasing of common rules - for example the definition that is used for a "region" in shading puzzles.

For the particular case of Hexa 7, I thought the first two sentences of the rules perfectly adequately described the puzzle. I'm not sure what you are referring to when you say "adjacency condition", Tom, because there isn't one, apart from the implicit one that arises conincidentally because of the fact that, in a hex geometry, adjacent cells also happen to be at distance 2 from each other via a different path.

detuned
Posts: 1953
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Puzzle GP 2020

Post by detuned » Tue 19 May, 2020 9:43 pm

The sentence started "if two cells touch...", the subsequent reference to the "...same (numbered) cell" was split over two lines (and also by parentheses), and in general these puzzles tend to have adjacency conditions. I freely admit I wasn't prepared, all I'm saying is I think I might have had a better shout if the rules had been stated more clearly.

It seems to me the phrasing of the rules is more geared to that weird example where they have two 4's touching each other [which feels like its had a disproportionate effect of bending the rules to fit what looks to me as being a poorly designed puzzle].

If it were me, I'd state the rules as:
Place a number from 1 to 7 in each cell in the grid. Numbers placed in cells a distance of 2 (or less) apart must not repeat.
I feel like that gives someone a fairer chance of quickly getting the gist of the puzzle.

detuned
Posts: 1953
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Puzzle GP 2020

Post by detuned » Tue 19 May, 2020 9:59 pm

Another suggestion I've seen elsewhere is: rather than have huge long paragraphs where you have many sentences strung together, a clearer format is something like:

Describe the solving mechanic [E.g. draw a loop/shade some cells/place some a number in ever cell etc] first.

Then every constraint is listed as a bullet point:

1. (the most important constraint)
2. (then the next)
3. (then the next)

etc

Under this format, and maintaining that weird exception:
Place a number from 1-7 in every (unshaded) cell in the grid

[Remove the reference to integers only, that's a decidedly unhelpful distinction. If you are seriously worried about that level of pedantry, then simply state "Place an integer"...]

1. Two (unshaded) cells that both touch a third (unshaded) cell must contain different numbers
2. Two cells that share an edge, but do not both touch a third (unshaded) cell may contain the same number

[I'd be against including the word unshaded as I think its unnecessary, but I suppose for highlighting this example it makes more sense]

Nilz
Posts: 276
Joined: Fri 02 Sep, 2011 5:21 pm

Re: WPF Puzzle GP 2020

Post by Nilz » Wed 20 May, 2020 10:03 am

Congrats on a really good result Freddie!

For my part, I made a small error on the yajilin that I didn't catch- it was an actual solving error rather than a answer string error, which is perhaps slightly less annoying (but more worrying), but still a big chunk of points to throw away. I also solved a decent chunk of the Hex Islands before making a mistake and grinding to a halt, so another bunch of time and points wasted there. Luckily, scores seem down overall, so I think a lot of people will be dropping this one. Which is a shame actually, I enjoyed the theme, Hungarian rounds are almost always good fun.

For the Hexa 7, my position would be between Sam and Tom's. It took me longer than it should to understand the rules and example- it feels like there's a clearer way to say the same thing (though I can't be bothered to think of such a way!)- but luckily I did this ahead of starting the clock, unlike Tom.

Puzzle_Maestro
Posts: 75
Joined: Mon 11 Mar, 2019 5:23 pm

Re: WPF Puzzle GP 2020

Post by Puzzle_Maestro » Thu 21 May, 2020 6:54 am

I partially agree with both sides here; I don't think the rules are as badly explained as Tom makes them out to be, but the bullet point format is definitely something I would like to see (rather than huge paragraphs where it is easy to miss a rule/exception). I didn't have any trouble myself as I remembered that this puzzle had occurred in WPC 2011 (as Honey Seven) and used that for practice.

detuned
Posts: 1953
Joined: Mon 21 Jun, 2010 2:25 pm
Location: London, UK
Contact:

Re: WPF Puzzle GP 2020

Post by detuned » Thu 21 May, 2020 1:14 pm

You what?!? Exactly how badly have I made things out to be - I’m somewhat bemused by some of the reactions here.

I’m not sure how I’ve managed to start a (somewhat petty) argument... I’ve said that I didn’t understand the rules during time, I’ve said that’s largely due to lack of preparation on my side, but I’ve also said the rules could be clearer - providing a couple of examples which might do a better job. I don’t think anyone here is against clearer rules, so I’m struggle to see where all this controversy has come from :(

david mcneill
Posts: 430
Joined: Mon 21 Jun, 2010 2:02 pm

Re: WPF Puzzle GP 2020

Post by david mcneill » Thu 21 May, 2020 3:21 pm

No comment. :|

Post Reply