The Authoritative Minesweeper Guestbook

Feel free to chat, make suggestions, or tell your scores! 
You can also chat with us on mIRC - server irc.newnet.net, port 6667, channel #minesweeper



  First Page
  Prev Page
  Post
  Home
Next Page  
Last Page  
Viewing Page 23 of 27 (Total Entries: 2685)
Feb 13th 2008 at 03:50:17 PM
Name:  

Ronny

Best expert:  

FL 91,7 and NF 87,432 -> FL 83,047 (3BV=139, 7 flags)

Comments:  

SWEEEEEET 8 seconds of my flag record (on clone)....only 4s to go get world ranked.


@Joachim: don't even bother posting your fake pictures

    Website Website    
Feb 13th 2008 at 01:58:27 PM
Name:  

Jason K.

Best intermediate:  

3.83-->4.02

Comments:  

Holy crap, finally sup-4! This was one of my recent goals, now I just need to approve on that **** 15...

Email Email     Website Website    
Feb 13th 2008 at 12:47:56 PM
Name:  

Elmar

Comments:  

@Ryan: You were right. I access my hotmail through MSN messenger which uses IE and usually use FF.

   
Feb 13th 2008 at 10:28:58 AM
Name:  

Joachim Askjer

Best expert:  

34

Best intermediate:  

11

Best beginner:  

2

Comments:  

Where do i upload proof-pictures?

Email Email    
Feb 13th 2008 at 09:20:00 AM
Name:  

daniel

Comments:  

@bertie: the first two rows are obvious, the last is fibonacci, ican not (vet) see any sytem behind the thrid one...but they all could be generated randomly i think

   
Feb 13th 2008 at 07:16:49 AM
Name:  

Andrey

Comments:  

@ Bertie: of course randomness of a finite sequence is only confirmed by statistical tests, and this is why we are pretty happy with the pseudorandom numbers (when we do simulations and don't think about security). the principal difference between the random numbers on that site and what you and I have on our computers is that their data comes from some atmospheric noise measurements. There is no seed, there is no algorithm, and thus there is no way to break it ever.
Anyway, this is sort of going off topic

   
Feb 13th 2008 at 06:36:13 AM
Name:  

Bertie

Comments:  

@Andrey: I hate to break it to you but it is mathematically impossible for us mere mortals to generate truely random numbers. The numbers on that website were also generated by a pseudo random number generator, though the might be a bit better than those Ryan creates. Here's a question to illustrate the nature of an infinite series of random numbers: Which of these series are part of an infinite random series?
1 2 3 4 5 6 7
2 4 6 8 10 12
6 3 8 9 3 1 3
1 1 2 3 5 8 13

What do you guys think?

   
Feb 13th 2008 at 03:14:02 AM
Name:  

Ronny

Comments:  

@Gergő: What are your other 2 records on beginner and expert?

   
Feb 13th 2008 at 12:41:05 AM
Name:  

Gergő

Best intermediate:  

22,563

Comments:  

Some new records. Int NF 3BV/s improved from 2,438 to 2,456 an 59 board. And a huge record: 3. in time (23,297) and 3. in 3BV/s (2,422), and first in RQP (9,620) and 1. in IOS: 1,28492. I also made my 2. sub 23, a 22,9. Sooner or later I must have a 21

Email Email    
Feb 13th 2008 at 12:32:02 AM
Name:  

Ronny

Comments:  

@Ryan: Yesterday it didn't work when I tried at work and now it does, so I guess the problem is gone.

@Kamil: Can you please stop playing at hi-games, you make us look like slow.

   
Feb 12th 2008 at 05:20:21 PM
Name:  

Ryan Heise

Comments:  

If it still does not work, it is a cookie problem which must be sorted out first. You need to make sure you use the same web browser to complete steps 1 and 2 of the registration process (if your email program uses its own internal web browser, clicking on the registration url from there will not detect the cookie that was set in your main web browser).

I have changed the registration process now so that the same web browser is definitely used both times.

    Website Website    
Feb 12th 2008 at 04:41:06 PM
Name:  

Elmar

Comments:  

@Ryan: Ok, I have tried again. This time I received the email within 1sec but it still doesn't work. Could you please add me manually?

   
Feb 12th 2008 at 04:32:34 PM
Name:  

Ryan Heise

Comments:  

Elmar: If you tried to register a day or two ago, there was a severe problem with delayed emails which prevented a few people from registering. Note that if you tried to register 4 times, it is only the registration key in the 4th email that will work (unless, perhaps it timed out?).

It looks like email is working better now, so you may like to cross your fingers and try again. I am also logging all registration attempts so that I can bypass the email and register you manually if necessary.

    Website Website    
Feb 12th 2008 at 02:22:45 PM
Name:  

damien

Best expert:  

Website Updated

Comments:  

Sorry for being late. I try to update very Sunday. 'Grats Manu on your #4. Welcome to Ko Yeong-Ju (2-14-50). Fritz got 51, Zhang Hong made 4.33 NF on int...

   
Feb 12th 2008 at 11:46:27 AM
Name:  

Elmar

Comments:  

@Ryan: I seem to be unable to register on your site. I have requested 4 confirmation e-mails now and everytime I got "incorrect key" after clicking on the provided link.
I used the e-mail below.

Email Email    
Feb 12th 2008 at 10:26:46 AM
Name:  

Ronny

Comments:  

@Andrey: I don't think you have to worry about that. Nobody on earth can possibly solve all the divisions and multiplication of the pseudo random generator (EVEN if he knew exact starting seed of the generator that was used to create the board) and see the relation between how the mines relate to the used numbers in time scale anywhere near a split second. If there ever will be a genius that can solve a minesweeper board faster because that can see the relation between a pseudo random generator and the placement of the mines....he deserves it.

   
Feb 12th 2008 at 09:35:41 AM
Name:  

Andrey

Comments:  

@ Ryan: if you are serious about making your game secure, I thought it might be a good idea to use real random numbers instead of traditional pseudorandom sequences. They are available at random.org and can be retrieved by requesting url like this: http://random.org/cgi-bin/randnum?num=100&min=0&max=10&col=1

   
Feb 12th 2008 at 06:02:25 AM
Name:  

tK (Also inofficial)

Comments:  

OK, my opinion:
All changes to the interface that do not give any more information to the player than (pre-vista winmine without cheats, standard configuration) would have revealed at the same point in the game are to be accepted.
So, counters that are displayed during the game may take anything into account that can already be known by the player (otherwise click counters would have to be hidden until the game is over as well.).

So much for information displayed outside the place where the game itself takes place.
As for the look, I would say that the same applies, with an additional restriction:
Information may only appear in the same place (square) of the game where winmine would display it, because the distribution of the information is also an important factor in a game situation, and because it is critical (click counters are not).
So, the board may not assist in determining mine positions any more than winmine does.

I might write a spec based on the original winmine soon and see what the rest of the community thinks.
As a matter of fact I think I'll do that now

I think that it important to distinguish between information that is available during and after the game.

VC=BVVB

Email Email    
Feb 12th 2008 at 02:27:53 AM
Name:  

Bertie - unofficially

Comments:  

On the clone thing Ryan mentioned. There are many ways you can change the interface, and some have been implemented. The most important thing to consider when looking at these changes is whether or not they give a player an advantage. Luckily for us there doesn't really exist a better control system than the old fashioned mouse, except maybe the Elmar technique. There is also only a few things cheating can do for you like give you extra info, or unwarrented control/speed/.... The other thing is that control might be taken out of your hands, meaning you don't do all the work. I remember one guy who used a mouse click macro to get a 1 on beg. All this considered (in its jumbled incomprehensible incompleteness ) I have no problem with the space bar thing.

On the Stevan vid thing. I think it is doable as well. Although I still have some suspision about a 4.07 game on a 166 with 17 openings. If he did do it though, then I think he has great mouse control . He just needs to learn some efficiency.

BTW. How can I get arbiter to show the top 30 IOE scores in my history on expert? I wanna see how I stack up against those big guns of the efficiency world. Another point: What are you oppinions concerning the difference between efficiency and efficacy (see link for definition )

    Website Website    
Feb 12th 2008 at 12:33:47 AM
Name:  

Ronny

Comments:  

@Ryan: Having (had) hand problems myself, I've used that same spacebar technique for minesweeper to do same training in a period in which I otherwise couldn't play do to overusing the fingers on my right hand by all the clicking. My conclusion was that it is useful to not get worse due to lack of training, practice your general NF pattern recognition skills and to practice on efficienty while having physical problems. However the spacebar is actually a little slower than the good old left-mouse button, so for me it most definitely did not result in faster times or higher 3BV/s. I don't believe I've ever set a record like that except an occasional IOE as result of lower click speed, so I don't see any problems in adding this feature.

This is just my experience, so maybe the IMC disagrees with this completely.

   
Feb 11th 2008 at 07:20:59 PM
Name:  

Ryan Heise

Comments:  

I read in a few places that people take very seriously any minute difference between a clone and the original game as if that makes the clone unacceptable. So I think I should ask what is the view on allowing spacebar as a substitute for mouse clicking? My view is that it is a valid substitute for mouse clicking since someone with a disability could rig their mouse to work in a similar way anyway (e.g. some people with carpal tunnel syndrome use foot pedals to control mouse clicks). As long as it remains that a single click is a single click - i.e. you cannot perform multiple clicks with a single click, then game play is not affected (it is still the same game).

I am not to sure that changing the sound effects makes any difference. I would think sound effects, the size and colour of the squares, and any other graphics can be freely changed as a matter of personal preference without being considered unfair.

    Website Website    
Feb 11th 2008 at 06:12:57 PM
Name:  

Ryan Heise

Comments:  

About cheating (Sorry, long post.)

The central issue is that movies can be faked in such a way that there is no proof that they were faked.

In the Rubik's Speedcubing community, we have a similar issue. We also have some members who actively try to construct fake videos so that we know what methods real cheaters might use to do the same. But in the end, we can never be 100% certain in some cases.

The only proof is a live demonstration.

At least in the speedcubing sport, we have started to see official tournaments gain popularity all around the world, and this gives us the live proof that is needed, and has also meant that less importance is now placed on our old "unofficial" highscores list that was maintained over the Internet (i.e. not live).

On the other hand, there still is a solution to the problem for validating records that are submitted over the Internet, and in speedcubing there still is an active community around electronically submitted records (my online simulator: http://www.ryanheise.com/cube/speed.html)

The idea is to run the game inside a webpage which has continuous contact with the server. The server can therefore witness "live" the record being set by the player.

I could try to set up a similar thing with my minesweeper game. The cheat detection code in this game is not complete yet, but there are already a number of tripwires in there. What happens is:

1. The user signs up and is required to agree not to cheat at that point.
2. As the game is played, the server looks at various indicators, each of which can determine with 100% certainty whether that person cheated.
3. The player is classified as a cheater, locked from posting further records, and is defamed.
4. If the user completely bypasses the user interface and manages to submit a dishonest highscore through a backdoor, this is actually classified as hacking and is punishable by law.

While the cheat detection code is not complete for this game, all it takes is for the player to trip over just one of the tripwires, and then the player is discovered.

So, if it is ever necessary to have evidence of someone's minesweeper abilities, it could be possible to ask them to do a live demonstration on my game.

On a side point: at some point in the future, I plan to make a multi-player version of minesweeper in which you race another player to sweep some particular board (both players get the same board each time). This is another example of a real-time game.

For an example of what I mean, you may have a look at what I have done for the Rubik's Cube head-to-head racing game at: http://www.ryanheise.com/cube/speed/ It is not very easy for someone to cheat at this because both participants are playing in the same time frame.

    Website Website    
Feb 11th 2008 at 05:24:55 PM
Name:  

Michael

Comments:  

I don't mean to seem harsh on the IMC, but that summary contains virtually no info. There aren't even conclusions given for all topics. For the ones that have conclusions, there's no account of *why* those conclusions were reached, what kind of points were put forward, etc. Why not just put a transcript online? Is there any sensitive information that is just too dangerous for public consumption? If so, you can hold those particular bits back and post the rest. And was anyone outside of the IMC notified that the meeting was going to take place? Because I didn't hear about it. I think it's important that people outside of the IMC can raise issues that they have for upcoming meetings, or give their opinions on the issues that are going to be discussed anyway. IMC members are not the only ones whose opinions matter; the IMC is there to serve the community, after all.

Perhaps these issued could be discussed at the next meeting.

   
Feb 11th 2008 at 02:49:57 PM
Name:  

tK

Comments:  

Oh and BTW, seeing as I'm IMC, that does not represent the official opinion of the IMC, its just something I noticed.

Email Email    
Feb 11th 2008 at 02:09:11 PM
Name:  

Lawrence Schwartz

Best expert:  

64.10

Best intermediate:  

17.13

Best beginner:  

3.06 ---> 2.93 ---> 2.84

Comments:  

Hmm, i guess that doesnt really make any difference....im trying to get to 1.**

   
Feb 11th 2008 at 02:02:38 PM
Name:  

Lawrence Schwartz

Best expert:  

64.10

Best intermediate:  

17.13

Best beginner:  

3.06 ---> 2.93

Comments:  

slowly moving up the list.....

   
Feb 11th 2008 at 01:58:55 PM
Name:  

tK

Comments:  

@Curtis:
I intentionally didn't include MS X explicitly regarding skins ;-)
In that case, OK. It still would be nice though!

Email Email    
Feb 11th 2008 at 01:17:57 PM
Name:  

daniel

Comments:  

just saw manus upk exp 21 - what a game! shows what could be possible

   
Feb 11th 2008 at 09:45:51 AM
Name:  

Curtis

Comments:  

MSX does have the black and white skin, you just have to download an extension. As for sound, MSX was based on the Win98 version, which doesn't show that as an option.

   
Feb 11th 2008 at 08:54:02 AM
Name:  

tK

Comments:  

I personally don't think the flags are suspicious. The first one is fast, yes, but in 0,34 secs, its doable. and the second one is definetly unconventional and as far as I could see neither safe nor very useful (at least he doesn't use it), why should a misflag like that be suspicious? He could have misread the board and thought that there was a mine there, seen that he could be wrong but wasn't necessarily so, and so didn't unflag it to save time.

Whatever, rilian and curtis (and I don't know about rodrigo, but I believe this applies to you too):
Your clones are not perect minesweeper clones - they don't have winmine sound

Seriously, I think that a clone that claims to be a clone of winmine should really have every feature of the original (not that its hard), even if it's annoying and most likely not to be used.
Don't take it personally, I just think that it should be included and that there is no problem in doing so.
Also, I would like to note that the same applies to the black and white skin.

Email Email    
Feb 11th 2008 at 08:27:12 AM
Name:  

Leo (Ryo Nishihara)

Comments:  

But stevan's 41s on arbiter is too suspicious to let go by (especially the first flag and the one at 26s.)
And if once someone trys to cheat, he should not be accepted with his any scores.
I know it's kinda sensitive matter ,but still... Let me know how you guys take.

    Website Website    
Feb 11th 2008 at 06:20:17 AM
Name:  

damien

Comments:  

In the past, some players did not believe Stevan about some of his scores (especially his 41 on Winmine), so when I transferred scores into the new database I marked certain scores and players as needing investigation. Now finished, I talked with Stevan last month and re-instated his scores. (He has made many 48's on Clone, a 46 on MSX, he beat my scores on minesweeper.org before that site died, and he has 41 on Arbiter). As Joni said, you can't delete someone just because you have a 'feeling'.

   
Feb 11th 2008 at 05:55:01 AM
Name:  

Name is not required

Comments:  

Damien, why did Stevan Gvozdnovic disappear from your ranking and why did he appear again?

   
Feb 11th 2008 at 04:43:05 AM
Name:  

Gergely Nagy

Comments:  

Hello Everyone!
There was an IMC meeting yesterday of which you can find the summary here:
http://www.minesweeper.cc/index.php?page=archive
Unfortunately I wasn't able to attend this meeting but I'd like to express a minority report here regarding the topic of registering the IMC.
In my opinion the reasons that were listed at the meeting weren't sufficient to reject this idea. If any of you have insight of how it could be done (any lawyer around? ) or what are the problems with it, I'd be pleased to hear about

Also, any ideas are welcome about
a) how the live tournaments could be made available to more sweepers
b) how the online tournaments would look like
send your ideas via e-mail to i.m.c@hotmail.com

Email Email     Website Website    
Feb 11th 2008 at 03:58:47 AM
Name:  

Gergő

Best expert:  

NF 86,928 -> 86,502

Comments:  

@Tibor: Good work!
@Ronny: I have the same problem as you. But soon we both will breast the tape!
@James: Practice makes the expert. You will be a sub 100 player in few months if you practice a lot. Everybody did it so .
@all: In Sunday there was an int NF 3BV/s record shower! On Sunday morning my "old" record was 2,354@70 3BV. Then it developed to 2,355@76 (yes, only withy one thosandth -> 2,404@67 -> 2,411@61 -> 2,438@76. Best one in my NF folder. Besides My NF exp record fell down from 86,928 to 86,502 (3BV:144).

Email Email    
Feb 11th 2008 at 03:19:50 AM
Name:  

Daniel Lynch

Comments:  

I actually got a second NF IOE 1.000 last night but it was a 12 3BV and solved in 5 seconds. That means I must have done it a few times previously.

Nice to see some familiar names. Perhaps I should make a come-back this summer and try to get that sub-60! My chess has been going pretty well in the last year and I've been concentrating on that. Maybe it will have helped my sweeping :p

Email included below. My best times are 61, 15 and 2. Yeah, I was the other efficient one - unfortunately I was also the slow one!

Email Email     Website Website    
Feb 11th 2008 at 01:49:25 AM
Name:  

damien

Comments:  

@daniel lynch: nice to see you again! i sent emails to all your old addresses, and searched your eclipse-chasing website, but i couldn't find you...can you please send me an email? i don't have your current scores on my site (that's why i was trying to find you). Glad you still come around from time to time.


@tam: thanks! after that 4.175 i got a 46 two games later, but i didn't think anyone would wnt to watch such a slow game.

Email Email    
Feb 11th 2008 at 01:35:26 AM
Name:  

AreOut

Comments:  

Grats Arj

And hi Daniel!

   
Feb 11th 2008 at 12:03:23 AM
Name:  

Ryan Heise

Comments:  

Ronny and Elmar: The email delay is apparently not due to hotmail, but rather due to some recent outages where my website is hosted. They have tried to tell me that it should be back to normal, soon ;-)

    Website Website    
Feb 10th 2008 at 06:52:26 PM
Name:  

Tam Minh Bui

Comments:  

@ Daniel Lynch: Whoa! I know you who you are because I remember watching your videos, and thinking that you were the other efficient player aside from Zhang. It's nice to see you post here :D

@ Damien: Nice games, I always enjoy watching your style.

   
Feb 10th 2008 at 05:56:09 PM
Name:  

Ronny

Comments:  

@Ryan: Yeah I noticed it the problem with the int and expert records....they seem to work fine now. The only down side: everyone can see it took me 36 sec to solve an int board.

   
Feb 10th 2008 at 05:36:17 PM
Name:  

Ryan Heise

Comments:  

I just woke up and noticed I made a big mistake with the highscores. All highscores are saved to the beginner table so that the highscores for the other levels are empty I corrected that now.

For clicking on press vs release, I suppose I should check how this works on windows. One problem will be simulating this on the spacebar (working around auto key repeat in Java is a big problem).

Thank you for the positive feedback and I'm glad to see the first release sort of works as intended

    Website Website    
Feb 10th 2008 at 05:27:35 PM
Name:  

Daniel Lynch

Best expert:  

61 never did get that sub 60 :(

Best intermediate:  

15

Best beginner:  

2

Comments:  

It's been a LONG, LONG time since I posted. Hope some of the old school remember me. I've been playing a tiny bit over the holidays since I have some spare time. I got 2 interesting results on beginner boards.

Index Of Efficiency of 1.316 on a 25 3BV board. Finished in 6.5s. Not sure how that racks with when I was playing regularly.

After that I did something I don't think I've ever done! Non-Flagging Index of Efficiency of 1.000 played on a 19 3BV board which I finished in 6.885s.

That last one is kinda cool :)

I'm now a maths and economics (as well as chess) teacher in a secondary school. Hope you're all keeping well.

   
Feb 10th 2008 at 04:50:48 PM
Name:  

Arjádre (Aryeh D.)

Best expert:  

60---->59.697 NF

Comments:  

After a VERY long time, sub60 NF!

2 Good NF games today:
67@185 and 59@152

Other records:
2.753--->2.785 3BV/s, 25.035--->24.213--->23.053 RQP

Vid linked

VC: HYKU... haiku?

    Website Website    
Feb 10th 2008 at 03:56:05 PM
Name:  

damien

Comments:  

I made 2 new records this month, both of them records I haven't broken since 2002!! 14nf (previous was 15x4), and 4.174 3bv/s on expert (previous 4.06, 4.05, 4.02, 4.01). I also got my 5th sup5 on Int. Vids in my folder.

    Website Website    
Feb 10th 2008 at 03:00:07 PM
Name:  

Ronny

Comments:  

@James: Don't worry. There are dozens of tips and tricks to get faster. Try reading the attached link, probably there are even things you don't know yet, that could improve your time.

@Elmar: Yep, did get the email after about 4 hours....currently I'm first on beginner with 5,08s. Let's hope Kamil doesn't register anytime soon, cause he would probably solve a 17 3BV board in 3s

    Website Website    
Feb 10th 2008 at 02:10:37 PM
Name:  

james canoot

Best expert:  

192

Comments:  

I am new to this game ...but there seems to be more to it than meets the eye...my time looks like snail speed compared to these sub 100 scores...

Email Email    
Feb 10th 2008 at 11:13:42 AM
Name:  

Elmar

Comments:  

I just received mine. Took about half an hour. I am also with hotmail, so hang in there.

   
Feb 10th 2008 at 10:29:40 AM
Name:  

Ronny

Comments:  

@Ryan: Works pretty smooth . Only I did blast a couple boards because a square is opened on a press and not a release.

BTW, I've registered but didn't receive an email on my hotmail account...maybe it got auto delted as spam?

   
Feb 10th 2008 at 09:43:30 AM
Name:  

Ryan Heise

Comments:  

Finally it is finished

http://hi-games.net/minesweeper-beginner/
http://hi-games.net/minesweeper-intermediate/
http://hi-games.net/minesweeper-expert/

Now for some sleep!

Oh, one feature I want to mention. For people who suffer from Carpal Tunnel Syndrome due to excessive mouse clicking, this game has a special feature. It is possible to use one hand to move the mouse, and the other hand to perform clicks using the spacebar.

...since I've started, I may as well mention the other features

- as mentioned, the program tries to generate all boards to have similar difficulty so that highscores can be fairly compared timewise (this may improve in the future)
- records are automatically submitted
- replays are automatically recorded
- (if you login) you can personalise your colour scheme and also how mines are to be placed around the first click.
- you can embed replays of your highscores into webpages, a bit like youtube.
- oh yes, and it's a no-flagging version only.

    Website Website    
Feb 10th 2008 at 08:17:01 AM
Name:  

Andrey

Comments:  

Here's what I think on this...
3bv/s = IOE * clicks/s. Thus your speed is how quickly (clicks/s) you are able to find an efficient path (IOE). I noticed that at very low clicking rate I consistently finish boards with IOE > 0.9 REGARDLESS of 3bv. This means that low 3bv boards require more thinking and pattern solving and good moves are less obvious. So I agree with Ryan, we should try to measure pattern complexity instead of pure count of clicks needed (whether NF or F).

   
Feb 10th 2008 at 07:58:14 AM
Name:  

Ryan Heise

Comments:  

You might have misread me if you think I'm proposing an NP-complete problem. To begin with, I am proposing to approximate thought time per click. Of all the different ways that you can approximate this, you can expect some would involve testing every single path, and some would not. It is definitely not such a ridiculous thought... and probably a useful one.

    Website Website    
Feb 10th 2008 at 06:32:03 AM
Name:  

Elmar

Comments:  

@Ryan: Good idea, and besides the endless glory in teh minesweeper community you could also make a couple of easy bucks.
http://www.minesweeper.info/articles/MillionDollarMinesweeper.pdf


   
Feb 10th 2008 at 05:52:43 AM
Name:  

Ryan Heise

Comments:  

However, like 3BV we could possibly address path dependency by just considering the optimal path, which then leaves only(? ;-) the issue of finding an optimal algorithm.

    Website Website    
Feb 10th 2008 at 05:17:49 AM
Name:  

Elmar

Comments:  

Well, the problem with such a measure is, that is would always be path dependent...

I think it would be a big step forward if we had a 3bv for flagging style, i.e. the minimum number of clicks using double clicks, lets call it FBV.
Although it may be difficult or impossible to find an algorithm that determines the exact number reasonable time, to find a good estimate is not that difficult.
Christoph and I once sat down and came up with a simple algorithm that we then tested manually . For a board with a 3bv of 120ish we found a FBV somwhere around 90 in the first iteration...

   
Feb 10th 2008 at 02:22:35 AM
Name:  

Ryan Heise

Comments:  

I was thinking more about 3BV. As we know, not all clicks are equal. Are there any board difficulty metrics that attempt to measure the relative amount of thought that different clicks require? This would be interesting.

An idea I have considered so far: There are often chains of clicks that quickly flow from a single deduction. So in a chain of 3 clicks, there was only thought before the first click, and the next 2 clicks would not need more thought. I think it is possible for a computer to algorithmically determine these cases. It is also necessary to make this metric independent of the user's click order.

    Website Website    
Feb 10th 2008 at 01:36:19 AM
Name:  

AreOut

Comments:  

@Andrey: older versions of Clone and Arbiter dont have limits. also WinMine doesnt have.

   
Feb 10th 2008 at 01:29:25 AM
Name:  

Schu (Andrew McCauley)

Comments:  

@ Ronny: what you pasted there is BAT, a related index. OBV is this:

(3-2.5*Level+0.5*Level^2)*(.07*numbers+.43*3BV+2.27*openings) + (-3+4*Level-Level^2)*(.2*numbers+.32*3BV+1.38*openings) + (1-1.5*Level+0.5*Level^2)*(.38*numbers+.23*3BV+.99*openings)

   
Feb 10th 2008 at 01:24:18 AM
Name:  

Schu (Andrew McCauley)

Comments:  

@ Kamil: I made OBV. It is an attempt at improving 3BV. You see, I think in an ideal world, 3BV should show you proportionally how much time a board would take to finish - in other words, for low or high or medium 3BV boards, 3BV/s for a player would be pretty similar. But that isn't so, low 3BV boards take longer per 3BV to finish, and high 3BV boards take less per 3BV, making low boards get low 3BV/s and high 3BV get high 3BV/s.

OBV I made with some statistical regression, and basically it fixes that. The reason the formula looks so crazy is because OBV is different for each level, and because it takes in to account openings, non-opening 3BV, non 3BV numbers all seperately with seperate values.

Let me know if you have questions. I don't think I've explained it very well.

   
Feb 9th 2008 at 08:03:23 PM
Name:  

Ronny

Comments:  

According to local folklore it's supposed to predict the time and date of the next WR based on the current WR.

   
Feb 9th 2008 at 07:05:30 PM
Name:  

Andrey

Comments:  

Jesus!... What is that formula supposed to tell us?

   
Feb 9th 2008 at 06:38:31 PM
Name:  

Ronny

Comments:  

@Kamil: It's just a ratio like IOS and RQP to indicate how well you've played your board and mean Optimized Board Value. To only formula that I've found (you can add it as customized index in clone) is the following:

1+((time-1)*((3-2.5*Level+0.5*Level^2)*(15/(.07*numbers+.43*3BV+2.27*openings)) + (-3+4*Level-Level^2)*(62.5/(.2*numbers+.32*3BV+1.38*openings)) + (1-1.5*Level+0.5*Level^2)*(170/(.38*numbers+.23*3BV+.99*openings)))/(Solved3BV/3BV))

   
Feb 9th 2008 at 12:40:44 PM
Name:  

KAmil

Comments:  

I have 2 questions:
1. What is OBV?
2. Clone has only sup29 3bv for int and sup98 for exp. And what about beg? Is it possible to get 1 3bv on clone 2007?

   
Feb 9th 2008 at 12:05:42 PM
Name:  

KAmil

Best intermediate:  

10 NF

Comments:  

Just got 11,835 NF on 41 3bv as my 2nd best RQP In fact i could play faster, but I blasted 11NF 2 times today in last click, so I decided to play it slower this time
@Damien: Manu has bad date of 10 in news

   
Feb 9th 2008 at 05:40:58 AM
Name:  

Arsen

Best expert:  

sub60x100

Comments:  

Finally )

   
Feb 9th 2008 at 03:03:52 AM
Name:  

KAmil

Best intermediate:  

NF 4,606

Comments:  

Just made my 2nd best NF 3BVs: 73 3bv (7 openings !!!) in 16,985 @ 4,567 Vid in my folder
Also blasted 39/40 @ 11,113 @ sub3 RQP

   
Feb 8th 2008 at 07:27:25 PM
Name:  

Ronny

Best expert:  

FL 93,352 -> 91,700 (3BV=181!!!)

Comments:  

My 2nd 'hop' in just 2 days on Clone.

Time = 93,352 -> 91,700 (clone record only)
3BVs = 1,954 -> 1,996
IOS = 1,1469 -> 1,15328
RQP = 49,491 -> 45,951

Finally I seem to be able to use NF & Flag on the same board without making mistakes, losing accuracy or slowing down.

Still need to get 12 sec of that time on Clone to get world ranked, but looking at that 3BV of 181 I still have plenty of room for improvement.

    Website Website    
Feb 8th 2008 at 05:45:14 PM
Name:  

Andrey

Comments:  

@ AreOut: how do you get to see sub30 3bv boards? Aren't they automatically sifted out by Clone/Arbiter? (when I get to play winmine I don't bother counting 3bv )

   
Feb 8th 2008 at 10:03:37 AM
Name:  

KAmil

Comments:  

@Tam: I watched 1 vid sup7 on 12 in 2, but I dont remember whos was it

   
Feb 8th 2008 at 06:19:01 AM
Name:  

Bertie

Comments:  

Man this math thing is running rampent! I stopped reading the post halveway up.

@The large numbers thing: The two methods I mentioned can be classified as one being impirrically based and the other analytically based. Meaning that the one using a computer to generate a set of boards randomly and calculating the average gives an approximate answer (accurate to however far we want to go into the decimals) while the one calculating the multiplicities gives the exact answer. When I say exact, I mean exact! I might write it some day.

In this case the impirical method is, for practical reasons, the better one to use.

Anyway, congrats to all the record breakers.

   
Feb 8th 2008 at 06:00:52 AM
Name:  

Tibor

Best expert:  

75

Best intermediate:  

18,86--->16,98(NF!!!)

Best beginner:  

1

Comments:  

third sub-20
easy 34 3bv board...
i'm surprised of style... i'm not a NoFlagger!!!
expert time remain too high...

i update my folder whit 3bvs records

    Website Website    
Feb 8th 2008 at 05:46:08 AM
Name:  

Tam Minh Bui

Best beginner:  

0.34 -> 0.33 RQP

Comments:  

I don't know if this is the first sub-3 sup-7 score, but here is the game that I just got early this morning. The board is incredibly easy and set up nicely. This is my seventh sup7 score, and my second best 3bv/s score overall.

    Website Website    
Feb 8th 2008 at 05:46:06 AM
Name:  

tK

Comments:  

Well, not unpopular, but I still continue playing a game until I blast/win regardless of the way it started.

Email Email    
Feb 8th 2008 at 04:57:47 AM
Name:  

Curtis

Comments:  

@Andrey: Storing/manipulating the numbers for 3bv distribution is straightforward; time is the limiting factor. You could easily do Beginner by brute-force but you could never do Intermediate, not even with every computer in the world. Any solution will require mathematical analysis (actually this was on my Minesweeper to-do list).

    Website Website    
Feb 8th 2008 at 04:39:32 AM
Name:  

Ryan Heise

Comments:  

I think I was not aware of the standard way minesweeper works on windows. Is having a guaranteed opening on the first click considered an unpopular style of game?

I should finish the programming of the game in the next 2-3 hours, hopefully release it, and go to bed ;-) But before that, I hope I can make the right decisions about the game rules.

Definitely, it will be based on the following rules:

1. non-flagging only
2. boards will always have a constant 3BV (i.e. consistent difficulty level).

The last decision is whether mines should be cleared from the area where the user makes the first click. This will affect the average board 3BV that U will use for (2).

    Website Website    
Feb 8th 2008 at 03:32:40 AM
Name:  

tK

Comments:  

Oh yeah, and there is more that we need to take into account:
distribution is also important. How many of the generated boards have a low enough 3bv to make a highscore probable?


Oh yeah, and why do you leave the 8 adjacent squares free as well? I finish many boards that I start without getting an opening immediately, as a matter of fact I like starting in a dense 3bv area.

   
Feb 8th 2008 at 03:29:06 AM
Name:  

tK

Comments:  

Yes, but what about people who start in the corner (I do), that skews the average in a different way.

I'd say take the average, because it is also the average when taking all possible first clicks into account (every square in the game will always be free in the same number of cases, making the first click irrelevant again.)

What would interest me is what square generates boards with the least 3bv on average ;-)

On the other hand:
If the difference is only one 3bv, I guess the difference can be neglected when taking things like path into account.
This is the reason I like starting in the corner - I don't have to go back. And moving the cursor from one end of the board to the other can make a difference of many solved 3bvs.

Email Email    
Feb 8th 2008 at 12:03:47 AM
Name:  

Ryan Heise

Comments:  

Aha. I discovered the subtle issue with my first calculation. It is quite interesting and raises some more questions.

The 10,000 boards that my first program generater were *not* completely random boards. Why? There are two reasons.

1. My generator will always avoid placing a mine on the first square I click. To run this test, I always clicked in the centre. As you might expect, a board is likely to have a LOWER 3BV if your first click is in the centre of the board rather than the edge of the board.

2. My generator will also not place a bomb on any of the 8 neighbouring squares of the first click. This by itself will automatically produce boards that have a lower 3BV than the "pure" average 3BV for a truly random board.

But, the interesting thing is that in practice (well, at least in versions of minesweeper that will not place a bomb on any of the nine squares near the first click), the boards that we see are not purely random but are skewed according to the nature of the game.

So what are my revised numbers? Using a sample size of 1,000,000 they are as follows.

For a purely random arrangement of 40 mines on an Intermediate board, the average is 64.931694.

For a board that does not allow mines to be placed within one square of the first click, and if the player clicks in the optimal centre position to achieve the lowest 3BV, then effective average is 63.128661 3BV.
So the question this raises is, which figure should we take if the desire is to produce an average board? I am leaning to take skewed figure since the purely random figure would not occur in practice.

    Website Website    
Feb 7th 2008 at 11:38:47 PM
Name:  

Roman Gammel

Comments:  

@Ryan: At first why do you think there is the difference? I used your algorithm and got absolutely the same result after 10'000 boards. I also think there is no chance that random number generator can affect on results, 3BV calculation is too complicated to depend on any difference. Even most simple random generator in most applications can generate more than 100'000 different boards! Using famous Chebyshev's inequality you can easily get that the difference between your average on 100'000 boards and real one is less than 0,1 with probability 0,99. So it can't be 63 anyway!!! Otherwise its just lack of luck

@AreOut: sub30 is like 1 in 5000 games.

   
Feb 7th 2008 at 11:23:12 PM
Name:  

AreOut

Comments:  

I agree Andrey, btw sub30 isnt that rare, you can see it every day if you play int often :)



WR3J : new WR on exp will be ... 3J?!

   
Feb 7th 2008 at 09:11:43 PM
Name:  

Ryan Heise

Comments:  

I highly suspect that the different results we are getting could be due to dlfferences in random number generators.

I ran my calculations again using a different language and a dlfferent computer and got different results.

I have made my algorithm available for testing at http://www.ryanheise.com/minesweeper/3BV.html

    Website Website    
Feb 7th 2008 at 05:51:34 PM
Name:  

Andrey

Comments:  

Is there really "thorough mathematical analysis" behind the 3bv limits? If so, it would be interesting to see it. To me, they make no sense (except maybe for beginner). If their purpose was to prevent easy super-results then I think the idea has failed (we have recently seen several extremely easy int boards with 3bv>30). The REAL difficulty of a board largely depends on the number of openings and length of their borders. Furthermore, simulations show that sub30 and sub100 3bv are extremely rare (in my several thousand simulated boards I got none), so the limit rule looks more like a psychological barrier and dreamboard panic than a useful measure.

   
Feb 7th 2008 at 05:00:41 PM
Name:  

joni

Comments:  

i think 4 is accepted too Ronny, if i remember correctly.

   
Feb 7th 2008 at 04:51:11 PM
Name:  

Ronny

Comments:  

@Joni: I addition (If i remember correctly) the rule for 'beginner 3BV/s records' is that records on Sub5 boards are excluded to avoid records based on random clicking.

   
Feb 7th 2008 at 04:31:19 PM
Name:  

joni

Comments:  

math guys are back with a vengeance . btw, I like Andrey's approach , there's rarely a real need for crazy numbers.

on an slightly unrelated note, wanted to remind you the lower "limit" for expert is 99 (i.e. a record on a 99 3BV game would be accepted). No idea what the various clones internal limits actually are though...
How it became like that is quite an interesting story . It was one of the first (if not THE first) decisions of the IMC. [since it was 2.5 years ago, and everyone's been talking about transparency... i thought i might tell what i know ]. They surely must have put very thorough mathematical analysis into that decision (...) before starting a poll for 3 options 26,28,30 for int and 98, 99, 100 for expert (not 100% sure on the 98..).
Surprisingly all 7 members had voted two days before the deadline (a rare feat!). 30 was a relatively clear winner on int, but it was much tighter on exp. I suspect the main argument being which was the rounder number between 100 and 99 . 99 looked like winning that battle with four votes to three, when one of the members changed his mind and his vote in favour of 100. When the voting closed, another IMC member announced 2-30-99 were chosen as the limits. He then noticed that it was 100 who had actually won (4-3), but at the end of the day they decided to stick to 99 anyways. A very promising start, the IMC had just taken their first arbitrary decision... with no less than a minority vote

   
Feb 7th 2008 at 04:14:33 PM
Name:  

Ronny

Best expert:  

FL 1,885 -> 1,954 (3BV=187)

Comments:  

Now that my sub100-Clone curse was broken...I felt like doing a little hopsing today.

Time = 101,257 -> 93,352 (clone record)
3BVs = 1,885 -> 1,954
IOS = 1,1346 -> 1,1469
RQP = 55,649 -> 49,491


    Website Website    
Feb 7th 2008 at 02:49:07 PM
Name:  

Andrey

Comments:  

@ Roma: I get the borders easily by shifting the board in all 8 directions, but it doesn't get me the counts of openings. I thought there might be a smart algorithm to get it quickly...

   
Feb 7th 2008 at 12:51:54 PM
Name:  

daniel

Best expert:  

64

Comments:  

@ronny: not that long ago i made my 65-score with one single flag - and didnt even use it nevertheless it was count as a flagged highscore - my unflagged one is a 69 which i made lateron

   
Feb 7th 2008 at 11:04:21 AM
Name:  

Roman Gammel

Comments:  

@Ryan: few minutes ago I generated 1'000'000 int boards two times, I got 64,93 average 3BV for each of generations. So that's probably what we're searching for.

@Andrey: I use simple recursion for neighbour cell to spread from one cell and thus to find borders of the opening.

   
Feb 7th 2008 at 10:33:23 AM
Name:  

KAmil

Best intermediate:  

Dreamboard: 9

Comments:  

Ive just decide to make an experriment: spend 5 minutes/ day trying to get Dreamboard again I got one in 2 minutes playing and blasted so now I probably have to wait 2 weeks to next one

   
Feb 7th 2008 at 10:31:03 AM
Name:  

Ronny

Best expert:  

FL 101 (5x) -> 93,352 (3BV=157, 4 flags)

Comments:  

Hehe, finally broke the curse of my Clone of never finishing a Sub100 flagging. Still a far way from my 81s on Winmine, but at least that cursed 100+ time is no longer bugging me on clone.

(Luckily there is no minimum on flags to qualify a game as flagging, because I only used 4. )

    Website Website    
Feb 7th 2008 at 10:04:15 AM
Name:  

Ronny

Comments:  

Why doesn't it surprise that the minesweeper community loves math?

   
Feb 7th 2008 at 09:15:46 AM
Name:  

KAmil

Best intermediate:  

NF 10 (9)

Comments:  

Just made 11,595 NF on 38 and few games later 74 3bv in 17,514 @ 4,481 3bv/s NF as my 5th sup 4,4 NF Vids r in my folder

   
Feb 7th 2008 at 08:33:12 AM
Name:  

Andrey

Comments:  

I'm wondering what algorithm you guys use to count openings ? This seems to be the trickiest job in calculating 3bv of a board...

@ Bertie: Your approach will produce numbers so huge that you may have problems to store them let alone manipulate them. I prefer the good old Monte Carlo. Btw, if you want the averages, there is no need in such a crazy number of simulations as Roman made. 1000 already provides standard deviation small enough.

   
Feb 7th 2008 at 07:59:07 AM
Name:  

daniel

Best expert:  

64

Comments:  

hmm.../me blasted exp game --> Est Time: 59,82 «» 3BV: 150 (168) «» 3BV/s: 2,8

   
Feb 7th 2008 at 04:55:31 AM
Name:  

Ryan Heise

Comments:  

It looks like 63 3bv could be the answer for Intermediate, but on the other hand, I find it suspicious that our calculations would agree for Intermediate but not for Expert.

Anyway, I will build my game for intermediate first and make it always generate 63 3bv boards. Hopefully the number is correct ;-)

(Btw, the game will be strictly NF... What do you guys like to call that? "No flags" mode or "Non-flagging" mode?)

    Website Website    
Feb 7th 2008 at 03:52:35 AM
Name:  

Roman Gammel

Comments:  

@Ryan: I also tried to find average 3BV, two years ago I generated 50 000 000 expert boards and this gave 173,57 3BV in average. I think this is no more far than 0.1 from real result on expert. I even made complicated verification system to be sure that I generate different boards.

I also generated 250 000 000 intermediate boards, but I don't remember where I saved the result :( I think average was 63.

   
Feb 7th 2008 at 02:46:36 AM
Name:  

Bertie

Comments:  

Ooh! Some real math talk! Let me, let me!

I don't know how many of you have ever studied thermodynamics before, but it has something to say about this particular discussion. They key thing is a thing called multiplicity. Basically, in minesweeper terms, the multiplicity of a particular 3bv is the number of ways a board can be made having that 3bv. This is emensely important because the spread of the multiplicity has a uge effect on the average which we wish to obtain. This problem can thus (from a programmer's point of view) be solved in one of two ways. 1)Generate a large number of boards using the algorithm in question (Btw @Ryan: the 3bv limit of 100 is a minimum for exp, not a maximum for int.) and basically just get the average (which implisitly deals with the multiplicity thing) or 2)Write a program to determine the multiplicity of each allowed 3bv and work out the weighted average. No. 1 is fine, but no.2 (the one I'd like to see done) would need a supercomputer (or at least a good amount of time) because the multiplicity of a single 3bv count go up to the trillions and up. I think I might just take a peak at this problem.

Okay bye

   
Feb 7th 2008 at 01:51:41 AM
Name:  

Gergő

Best expert:  

F 77, NF 87---> 86

Comments:  

Record! NF exp went down from 87,905 to 86,928 on a 141 board. Video in my folder, and my sup 2 NF exp from yesterday, as well. 80 is approaching slowly, but surely

Email Email    
Feb 6th 2008 at 10:28:09 PM
Name:  

@kyle

Comments:  

(x*y)/(x/y)=y^2

   


  First Page
  Prev Page
  Post
  Home
Next Page  
Last Page  
Viewing Page 23 of 27 (Total Entries: 2685)


powered by Powered by Bravenet bravenet.com