Solving contest speed+%

Anything to do with minesweeper...
Post Reply
scarrabee
Posts: 10
Joined: Sat Sep 01, 2012 12:33 am

Solving contest speed+%

Post by scarrabee » Wed Sep 05, 2012 9:17 pm

Do any of you guys play, not only for the fastest time, but also for the highest number of solved games?

I did some competitions against a classmate: 100 puzzles 1 point for every 20 sec you need and 20 point for every unsolved game.
(This is a lot easier on the windows version of the game because of the starting opening)
I still have 31% solved on 2600 exp games

aradesh
Posts: 90
Joined: Sat Aug 29, 2009 3:37 pm

Re: Solving contest speed+%

Post by aradesh » Thu Sep 06, 2012 5:36 am

31% games solved? To me that feels impossibly high, but perhaps if one really concentrates and has a free opening on click 1, it's possible...

None of us really play for percentage as the clones we use don't keep track of this info (we don't use windows version). I've been considering recently making changes to arbiter so it can keep track of all the blasted games/f2'd games/etc. and using this for statistical analysis of some sort. I would imagine that for most of us, our expert completion rate is <1%.

Sometimes some of us will try to complete as many games as possible in the shortest amount of time. [name removed]recently completed 1000 int games in 8-9 hours, and another time completed 100 expert games in 5-7 hours (can't remember the exact details). But even then he didn't keep track of complete % rate.

PS: scarrabee, you ought to come visit us in the Minesweeper Chat on IRC. See http://www.minesweeper.info/forum/viewt ... f=24&t=177 for details.

EWQMinesweeper
Posts: 410
Joined: Sun Nov 30, 2008 11:50 pm

Re: Solving contest speed+%

Post by EWQMinesweeper » Thu Sep 06, 2012 11:19 am

this contest doesn't seem to make much sense to me. especially not using vista minesweeper. with its 'replay the same board' function. i would blast every second board on purpose and check it for forced guesses. thus 50% win percentage would be no problem. also, this would be considered upk/cheating.

tracking the win percentage also seems irrelevant to me, as many sweepers are able to solve all solvable patterns with enough time.

as aradesh mentioned completing as many games per time as possible is an established challenge here. we use clone, viennasweeper or arbiter (which do not allow any cheating) for these marathons. this challenge make much more sense when comparing players' speed and solving ability. players try to finish games, make as few mistakes as possible and at the end there is an average time taken to complete a board. however, samples for this need to be quite large to be meaningful. from my experience the minimums are 100 games on beginner, 25 on intermediate and 10 on expert, although expert with its higher mine density is far less representative here.
„Das perlt jetzt aber richtig über, ma sagn. Mach ma' noch'n Bier! Wie heißt das? Biddä! Bidddää! Biddddäää! Reiner Weltladen!“

scarrabee
Posts: 10
Joined: Sat Sep 01, 2012 12:33 am

Re: Solving contest speed+%

Post by scarrabee » Thu Sep 06, 2012 12:03 pm

aradesh wrote:31% games solved? To me that feels impossibly high, but perhaps if one really concentrates and has a free opening on click 1, it's possible...
on the windows version you always get an opening
http://www.hortecta.be/site/minesweeper.jpg
aradesh wrote:None of us really play for percentage as the clones we use don't keep track of this info (we don't use windows version). I've been considering recently making changes to arbiter so it can keep track of all the blasted games/f2'd games/etc. and using this for statistical analysis of some sort. I would imagine that for most of us, our expert completion rate is <1%.
my completion rate will also be something like that on the clone version.

EWQMinesweeper wrote:this contest doesn't seem to make much sense to me. especially not using vista minesweeper. with its 'replay the same board' function. i would blast every second board on purpose and check it for forced guesses. thus 50% win percentage would be no problem. also, this would be considered upk/cheating.
I guess vista is the same as 7?
And I don't us replay. thats cheating :) .
EWQMinesweeper wrote: tracking the win percentage also seems irrelevant to me, as many sweepers are able to solve all solvable patterns with enough time.

as aradesh mentioned completing as many games per time as possible is an established challenge here. we use clone, viennasweeper or arbiter (which do not allow any cheating) for these marathons. this challenge make much more sense when comparing players' speed and solving ability. players try to finish games, make as few mistakes as possible and at the end there is an average time taken to complete a board. however, samples for this need to be quite large to be meaningful. from my experience the minimums are 100 games on beginner, 25 on intermediate and 10 on expert, although expert with its higher mine density is far less representative here.
I started playing like that because I had a (for me) crazy record, 118 first sub 150. and I couldn't even get close to it
(thats one of the reasons I like your versions of the game, more statistics and records :) )

computronium
Posts: 23
Joined: Thu Oct 11, 2012 8:32 am
Location: Melbourne, Australia
Contact:

Re: Solving contest speed+%

Post by computronium » Wed Oct 31, 2012 3:48 am

EWQMinesweeper wrote:tracking the win percentage also seems irrelevant to me, as many sweepers are able to solve all solvable patterns with enough time.

I disagree with this statement, actually; even if everyone were to play perfectly in deterministic situations, there is no consensus on optimal strategy at all for when one is forced to guess. Also, I use a different guess strategy (basically, I just envision some reasonable-looking mine pattern for the area I'm looking at and proceed as if it's true) when playing for speed versus when I play for percentage. I suspect that completion percentages differ fairly significantly even among the top players.

I hope to back up these statements some time soon with some actual evaluation of guessing strategies!

aradesh
Posts: 90
Joined: Sat Aug 29, 2009 3:37 pm

Re: Solving contest speed+%

Post by aradesh » Wed Oct 31, 2012 6:14 pm

computronium wrote:
EWQMinesweeper wrote:tracking the win percentage also seems irrelevant to me, as many sweepers are able to solve all solvable patterns with enough time.

I disagree with this statement, actually; even if everyone were to play perfectly in deterministic situations, there is no consensus on optimal strategy at all for when one is forced to guess. Also, I use a different guess strategy (basically, I just envision some reasonable-looking mine pattern for the area I'm looking at and proceed as if it's true) when playing for speed versus when I play for percentage. I suspect that completion percentages differ fairly significantly even among the top players.

I hope to back up these statements some time soon with some actual evaluation of guessing strategies!

I agree. I think it's a shame that no clone keeps track of information about blasted and f2'd games. It would be interesting to know things such as when you've completed 50% of a board, what is the likelihood you'll go on to finish it. Also simply interesting to know how much time we invest in the game, or completed games per average hour, etc.

EWQMinesweeper
Posts: 410
Joined: Sun Nov 30, 2008 11:50 pm

Re: Solving contest speed+%

Post by EWQMinesweeper » Fri Nov 09, 2012 6:30 pm

computronium wrote:
EWQMinesweeper wrote:tracking the win percentage also seems irrelevant to me, as many sweepers are able to solve all solvable patterns with enough time.

I disagree with this statement, actually; even if everyone were to play perfectly in deterministic situations, there is no consensus on optimal strategy at all for when one is forced to guess. Also, I use a different guess strategy (basically, I just envision some reasonable-looking mine pattern for the area I'm looking at and proceed as if it's true) when playing for speed versus when I play for percentage. I suspect that completion percentages differ fairly significantly even among the top players.

I hope to back up these statements some time soon with some actual evaluation of guessing strategies!
there is consensus on most difficult patterns and all other situations come down to pure luck. thus there is no use in comparing top players' win percentages. (top players include less than 10 sweepers in this context)

aradesh: use sorin's recorder for these stats.
„Das perlt jetzt aber richtig über, ma sagn. Mach ma' noch'n Bier! Wie heißt das? Biddä! Bidddää! Biddddäää! Reiner Weltladen!“

computronium
Posts: 23
Joined: Thu Oct 11, 2012 8:32 am
Location: Melbourne, Australia
Contact:

Re: Solving contest speed+%

Post by computronium » Mon Nov 12, 2012 4:56 am

Admittedly, you top scorers are operating on a completely different plane from me, so maybe there's some subtlety here that I'm missing. It seems to me, though, that we are maybe talking about different things. By "there is consensus on most difficult patterns", I assume you are talking about the situations where the mine pattern can be determined unambiguously, and with "all other situations come down to pure luck" you are talking about when there are no unambiguous situations left to figure out, and thus you have to guess. I disagree that all guessing strategies in the latter situation are created equal, especially aggregated over a large number of games.

And your conclusion, "thus there is no use in comparing top players' win percentages.", suggests that you are envisioning a second set of criteria which we would maybe start including along with the best players' rankings. I don't think that's what OP was going for at all. This would be an entirely different sort of competition, with different leader boards. I suspect that the fastest players in the current rankings might not want to come anywhere near it, as it would likely mess up their speed game, having to worry about higher completion percentages.

EWQMinesweeper
Posts: 410
Joined: Sun Nov 30, 2008 11:50 pm

Re: Solving contest speed+%

Post by EWQMinesweeper » Mon Nov 12, 2012 4:15 pm

decided to try the original suggestion.

completed 27 out of 100 games
took 1h 12min with many breaks
completed <10%: 38 games
completed <5%: 35 games
completed >90%: 12 unfinished
made 2 careless mistakes that lead to lost games

so that's a win percentage of:
27% - if you count all games
43% - if you ignore boards that were lost very early due to a forced guess
42% - if you ignore boards that didn't start with an opening.

i used sorin's recorder for winmine and the xp version of winmine. the zipped txt file is attached. i assume i don't need to explain how to use sorin's recorder and player. ;)

feel free to have a look at these games and share your thoughts on all guesses.
Attachments
12-11-2012_15-07.zip
(520.61 KiB) Downloaded 197 times
„Das perlt jetzt aber richtig über, ma sagn. Mach ma' noch'n Bier! Wie heißt das? Biddä! Bidddää! Biddddäää! Reiner Weltladen!“

Post Reply