minesweeper - place your bets | The Coffeehouse

evildictaitor
if( !succeed( try() ) ) { while(true) try(); }

Minh wrote:

evildictaitor wrote:
Minh wrote:I'm doing a histogram to see which cell is more likely to be assigned to.

Then you are no doubt aware that it is using the random() function which is pseudo-random up to eight digits and thus you won't be able to work it out by brute-force statistical measurement?
It's OK... I have a feeling that MineSweep.exe also uses a pseudorandom method

It uses the Windows random() function, as I just said. Consequently you won't be able to brute force it statistically because your machine won't have enough digits of precision to notice the difference between the pseudo-random windows call and a uniform random function.

Minh
WOOH! WOOH!

evildictaitor wrote:

It uses the Windows random() function, as I just said. Consequently you won't be able to brute force it statistically because your machine won't have enough digits of precision to notice the difference between the pseudo-random windows call and a uniform random function.

What if I max out the 8 digits, but do it twice Wink

That should cover it?

Coming on a million iterations..

evildictaitor
if( !succeed( try() ) ) { while(true) try(); }

Minh wrote:

evildictaitor wrote:
It uses the Windows random() function, as I just said. Consequently you won't be able to brute force it statistically because your machine won't have enough digits of precision to notice the difference between the pseudo-random windows call and a uniform random function.

What if I max out the 8 digits, but do it twice That should cover it?

Coming on a million iterations..

That would require 2^2^16 iterations to get a result with a 98% uncertainty.

Minh
WOOH! WOOH!

evildictaitor wrote:

That would require 2^2^16 iterations to get a result with a 98% uncertainty.

Oh? Using my new analagization technique, I'm able to only do it in 1,000,000.

And we have a winner!!!

evildictaitor
if( !succeed( try() ) ) { while(true) try(); }

Minh wrote:

evildictaitor wrote:
That would require 2^2^16 iterations to get a result with a 98% uncertainty.

Oh? Using my new analagization technique, I'm able to only do it in 1,000,000.

And we have a winner!!!

It disturbs me that there are people who know so little about statistics and statistical error.

Just to be sure though - run the program another five times and see.

Minh
WOOH! WOOH!

evildictaitor wrote:

It disturbs me that there are people who know so little about statistics and statistical error.

Just to be sure though - run the program another five times and see.

Are you nuts! If it came up w/ a different answer, it would mean I have wasted all this time. HellsNo!

1,000,000 iterations are solid enough!

SeanChilds

oO

Minh, you're evil! [6]

evildictaitor
if( !succeed( try() ) ) { while(true) try(); }

Minh wrote:

evildictaitor wrote:

It disturbs me that there are people who know so little about statistics and statistical error.

Just to be sure though - run the program another five times and see.
Are you nuts! If it came up w/ a different answer, it would mean I have wasted all this time. HellsNo!

1,000,000 iterations are solid enough!

You should run for president. You clearly know more about stats and their application than the current administration.

SeanChilds

evildictaitor wrote:

Minh wrote:
evildictaitor wrote:

It disturbs me that there are people who know so little about statistics and statistical error.

Just to be sure though - run the program another five times and see.
Are you nuts! If it came up w/ a different answer, it would mean I have wasted all this time. HellsNo!

1,000,000 iterations are solid enough!

You should run for president. You clearly know more about stats and their application than the current administration.

You do know that Minh is evil, don't you?

evildictaitor
if( !succeed( try() ) ) { while(true) try(); }

SeanChilds wrote:

evildictaitor wrote:
Minh wrote:
evildictaitor wrote:

It disturbs me that there are people who know so little about statistics and statistical error.

Just to be sure though - run the program another five times and see.
Are you nuts! If it came up w/ a different answer, it would mean I have wasted all this time. HellsNo!

1,000,000 iterations are solid enough!

You should run for president. You clearly know more about stats and their application than the current administration.

You do know that Minh is evil, don't you?

Oh yes. Ironically this makes him more suited to the job of Grand Vissier of the United States of Minhmerica.

Minh
WOOH! WOOH!

evildictaitor wrote:

You should run for president. You clearly know more about stats and their application than the current administration.

Mark Twain once said, "There are three kinds of lies: lies, damned lies, and statistics." I think Mr. Twain has a good chance of being right.

For the record, this this is the final 2 mines as predicted by my model.

So, if Maurits is giving odds, say... 7:1 would do?, I'm placing my bets.

SeanChilds wrote:

You do know that Minh is evil, don't you?

What if I promise to only use my evil to do good?

evildictaitor
if( !succeed( try() ) ) { while(true) try(); }

Minh wrote:

Mark Twain once said, "There are three kinds of lies: lies, damned lies, and statistics." I think Mr. Twain has a good chance of being right.

Statistics can be abused, but they don't lie.

Dan

Unfortunately Gotdotnet disappeared as Peter Hallam (ex-MSFT C# compiler guru) had a Minesweeper solver named Sweep.net, but Peter left Microsoft and I can't seem to find him on the internets...

The other alternative of course is to use your developer powers and...cheat.

Use WinDbg on minesweeper.exe and dig around to find the in-memory representation of the unknown squares. Yes, with great (developer) power comes great responsibility, but it would be interesting to see how Minesweeper's internals work and perhaps justified by the scientific inquiry as to how it works and if this could even be done.

Sven Groot
My name has 9 letters. Coincidence? I think not...

If memory serves, in the pre-Vista minesweeper there's a way to cheat and reveal the mines, something with the upper-left corner of the screen but I don't remember the exact procedure.

VB Man
Year of the Linux MCE.

This is quite an interesting thread. I have enjoyed reading it.

Another neat find is this post on using Visual Studio to pause the execution and see what values MineSweeper is hiding in the memory.

http://latkinson.spaces.live.com/blog/cns!A9D90665EB0960DD!156.entry

Dan

VB Man wrote:
This is quite an interesting thread. I have enjoyed reading it.

Another neat find is this post on using Visual Studio to pause the execution and see what values MineSweeper is hiding in the memory.

http://latkinson.spaces.live.com/blog/cns!A9D90665EB0960DD!156.entry

Nice article find, perfect for those folks who need to look at how the game works Smiley

Matthew van Eerde
AKA Maurits

Rowan
Look, no errors.

I lost.

CannotResolveSymbol
{insert caption here}

Why is it that I haven't won Expert in Windows Minesweeper in a long time, but I just won Gnome Minesweeper (on the same level) on my first try?

Maybe it's using Debian's random number generator Tongue Out

(I'm on Ubuntu)

evildictaitor
if( !succeed( try() ) ) { while(true) try(); }

CannotResolveSymbol wrote:
Why is it that I haven't won Expert in Windows Minesweeper in a long time, but I just won Gnome Minesweeper (on the same level) on my first try?

Maybe it's using Debian's random number generator (I'm on Ubuntu)

Leave the rng out of it! Debian's minesweeper may well have an algorithm that prevents unsolvable puzzles (like the one above) from happening.