[EDITOR NOTE: Visit the 'Intermediate Hall of Fame' archived site to read Matt's 'The Competition' essay.]
Some initial words
The sub-10 barrier has finally been broken. I came to think of Matt McGinleys essay, were our bold hero fights the newbie with the name Sven Hallebrant. Someone complained about another Scandinavian doing well in minesweeper, but I replied that Sven could very well be a German player. And now we have, indeed a new German record-breaker on the intermediate level. Matt also describes the design of a World Championship where the boards are created explicitly for that event. I know that Roland is probably a good overall intermediate player. Without some pre-knowledge of the boards, I doubt he would reach such a final, though. Matt doesn't tell anything about how "the bastard" Sven did in the final, but he (Matt) surely kicked his butt very well. The only reason Sven ever reached the final was probably that some corrupt board designer gave him the layout of all of the boards (I'm sorry that I "rewrite" some of the content of your essay, Matt, but I need it in my text. If you like, you can tell me how much my text sucks afterwards). But even with that knowledge, he was seriously beaten in the final. With how many hundredths of a second, I don't know. Anyway I still think that people like Matt, Damien, Dan, Lasse and Stephan (and a couple more) would beat Roland if all of them got the same, randomly created intermediate board.
Note: Since I haven't totally made up my own opinion quite yet, please don't get offended by these first few sentences.
The conflict, and my first encounter with THE DREAMBOARD
The world record wasn't achieved on a randomly selected board, and the discussion is running high about how "real" this record is. The dreamboard is, indeed a very special board. Even though someone disagrees, I think that the reason the last couple of intermediate world records have been on the dreamboard, is that the dreamboard is THE best board. All the fuzz about it might have drawn people's attention towards that particular board, but I still think it's the best board. And I was convinced that this was the truth from the first moment I saw it. After getting the TSCC codec on my CPU, I rushed to Matt's site to see his amazing board. He surely clicked like a devil, but I soon became aware of some lack of effectiveness in his solving algorithm. I was still in the mark-a-lot-of-mines period of my playing career, and figured out that the board could be solved in 24 clicks. My method involved marking six mines, and some double-clicking. I also found out that I could do the board in 8 or 9 seconds with my present speed, if I recognized it "at once". Now I only needed to get the dreamboard. The bad thing was that I could very well fuck it up if I got it, so were should I click to recognize it at once? I decided to find a big number, since they are pretty rare on intermediate. The 4 besides the 2 in the upper right corner seemed like a good choice. So, I went click-F2-click-F2-click-F2. I've never been more bored in my life. And just when you get maximum frustrated, the 4 pops up: click-F2-click-F2-click-F2... Hmm. Wasn't that last number a 4? Damn, where is the "replay board"-button? Such kind of things happen when you hit F2 too quickly. After a few days of click-F2 and a lot of non-dreamboards with a 4 in that position, I got tired of it. I started accepting some of the nice-looking boards with a big opening in the upper right corner too. I also figured out that the board before the dreamboard has a 2 in that position, so I could be a bit more prepared for what to come. I finally brought my score down to 20, and one day I got another 4. Well, that happens like everyday, so I kept on with a few more clicks. Things were looking nice, and the first opening suggested that I had finally got the dreamboard. Help! The dreamboard. Where shall I click now! The timer is ticking away; I'll never finish underneath 10 seconds. The pattern in the bottom center was the worst. Which of those damn squares was the "key mine"? I used 3 or 4 seconds only there. After a horrible round, I looked up at the timer: 20. Bloody hell. Didn't break my record, didn't even sub-20. I could possibly have gone a lot faster if I had never seen the dreamboard before. (I haven't got any videos of me starting in that corner because I'm not interested in low 3BV-boards above 19 seconds anymore). So that was what I had left from those hours of click-F2. I gave up the stupid "Start in the upper right corner"-idea.
Note: This was my first involvement with the dreamboard. You might think what you like about this; I'm just saying that I did the "mistake" of trying to learn too much about the dreamboard. The reason was probably my enormous greed for fame and a world record. When you look at all the doping in sport events today, there should be no reason to believe that Roland and I are the only ones that have used unreasonably much time on studying (maybe even memorizing) and trying to recognize the dreamboard. The difference is that everyone knows that doping is illegal, but I have seen no rules in the community that suggests that this kind of playing is illegal. But as with my 1-secs on the 9x9-grid, I will consider withdrawing boards I have solved "against the rules" if the community can agree on some common rules. I think that the best way of making such rules is a democratic election in the addicts group. Another way of dealing with this is to split up the community in two "leagues": One league with no or a few rules, and one league with very strict rules.
Some words about how randomly chosen the boards really are
Because of the evidence that the board before the dreamboard was always the same, and that there was more than one shift of the dreamboard AND the board before the dreamboard, I started a new project called the "determine the entire cycle of intermediate boards". I first recorded almost 3000 inter-games, trying to hit a mine as soon as possible (took me a few hours). After some careful examining, I found two "almost identical boards". The difference was that the second board was shifted two units down and six units to the left. After some counting, I found out that there were a little more than 1500 boards in between. So the length of the cycle was 1500 boards, and each board was shifted two units down and six to the left after one cycle (these numbers are 3 and 1 for beginner boards). I don't know about expert. The only thing I know is that I've had a few deja vu on a couple of expert boards. After some more examining of the recorded inter-boards, compared to some of my best inter-videos, I concluded that some boards didn't seem to fit in, and I couldn't find the dreamboard at all. About half of the boards were not in the cycle, and I guessed that there was a different cycle. After getting a dreamboard-shift, I repeated the click-and-lose strategy until getting another dreamboard-shift while recording. So I finally had the two cycles (the only two I believe). I've used a lot of time on developing this ever since, but I've never used the sequence of boards actively to predict what boards I should expect. I still can't foresee the dreamboard, but I sometimes recognize which cycle I'm in when I get a nice board, and I must admit that I probably play a few minutes extra when I've gotten a board from the dreamboard cycle. (There are about 1500 boards * 8 shifts = 12000 boards in that cycle). I just can't help it that I remember it if I get two nice boards in a short interval, but as Matt has said, there aren't many sub-20-ish boards before the dreamboard.
Note: I'm of the opinion that the use of this strategy to exactly plan how the next board will look like might ruin the entire game of Minesweeper. Maybe I shouldn't tell this, but I guess that programs like Sorins 3BV recorder and player and maybe Yoni's reader program (haven't tried that one yet) can be used to determine how the cycles looks like. The 3BV-values (or manual board-recognition) could be a way to figure out where you are in the cycle. I'm telling you this because I don't want to be the only one to know the truth, and someone else could well be using this information to their advantage if they get it first.
The disadvantage of possessing knowledge
The third time I got the dreamboard (after missing it once), I finished in 12 seconds: A nice improvement. I didn't recognize it at once, so I lost a few seconds on an unneeded marking and some lack of determination. I solved it as if I knew where the mines where, and I don't think I would have played it optimal if I were to think: Is this really a logical click, or should I clear those squares first to show that I haven't memorized the board? What is really a pity is that I memorized the dreamboard the first time I saw it, and I never thought that I could come close to 10 seconds without knowing how to solve it effectively. I was absolutely sure that Matt had planned the dreamboard down to the last detail before getting his 10. He had just not been able to figure out the most effective way of doing it. I sometimes regret that I ever saw the dreamboard, but back then I never thought it was possible to go that fast without memorising the board. What Matt wrote about watching his 14-vid thousands of times, didn't suggest that memorising was not accepted either.
It can be a bad thing to know too much. If you have played a lot of inter-boards, you soon learn something about the sequence of boards. What the hell are you supposed to do, then? Act like you don't know what's coming? If you play like this, you'll probably play worse the more knowledge you have. Suggest you know how the next board looks like. Say it's the dreamboard. Shall you just skip it? You know too much about it, so it's not fair that you should use your knowledge in solving it. You could try to pretend that you don't know it. Where should you start? Do you usually start in the middle, near the corners, on the edges? You wouldn't like to start with the largest opening. Maybe the smallest opening, or click a couple of numbers before you click an opening, or should you try to hit a mine. That's one probable opening. Maybe you should generate a random number to tell you where to open, and where to make your second and third click. But using time to click on some specific squares does not give you good seconds. And what shall you do when you get the opening? How long time is it reasonable to believe it takes you before you realize there is an opening. You could hit a mine in the meantime. What next? You know what's the most effective way of solving the board, but you would never have done that the first time around. I think the dreamboard is about 30 3BV. It could be solved in 23 clicks, in 30 clicks or in 40 clicks. How should you decide how ineffective you should solve it? Someone gives the damn and solves it the most effective way, like Roland, and a lot of other people (including myself) probably would do. I'm not sure if I trust all the people that tell me they play every board normally. I remember that I once got 21 seconds on a board I had seen on Matt's site. The last click was a 50-50 guess. Something in my sub consciousness told me that I had seen the board before, and that the leftmost square was a mine. I clicked the other square. I'm not even 100 % sure that it was recognition or luck that led me to the right square, and I don't think anyone can control their sub consciousness so that it doesn't hint you which of the squares are the right one.
The essence is that the lack of randomness in Beginner and Intermediate (and maybe even Expert), compared with the "small" number of boards (1500 boards * 8 shifts * 2 cycles =24000) can never make this a fair competition. People can claim not to use this to their advantage, but they can't prove it. If a person with knowledge tries not to use it, he might be worse of than a stupid person that has not got the ability to remember an entire board. In that case the ideal minesweeper would be a player with good pattern-recognition and bad board-recognition. A lot of people can learn 1-2-1 - and 1-2-2-1 - situations, but only a few of them can recognize a board they have gotten a few times before. I remember to have read that it was something like "big news" when somebody finally realized that a lot of people had got the same board (the dreamboard) a couple of years ago. I think that's strange. There should be some smart people around in the community that would recognize a good board if they got it twice. I remember that I played beginner one day, and finished a board in two seconds. A little later I finished a board in three seconds. A few days later I recognized the two-sec-board, and I also remembered the 3-sec I got on a very fast board. This will probably make my second 1-sec less than impressive, but I played every single board after the two-sec with the fast 3-sec-board in mind, and I'm really not very proud of my second 1-sec being on that board. I could have tried to play normally, but then I would have gotten nervous when I got the fast board, and could have screwed it up.
Note: The clue here is that if a person does not use his knowledge to his advantage, the knowledge would probably become a disadvantage. This is due to the fact that the optimal way of solving a board could be achieved by pure luck, but the optimal way of solving it should not be used (?) if you know about it in advance (someone will probably agree). Other factors like nerves etc. also gives the player with knowledge a disadvantage right from the start, so a positive ability (the ability to gain knowledge) becomes a handicap for the players with that ability. Screenshots and videos are not foolproof (look in my specials section for an example), but what is worse is that it doesn't tell anything about if the player did prepare himself for a particular board. A click that might look stupid if a board was totally random might be a planned click or just a misclick, and I guess that players shouldn't be punished for misclicks. To judge if someone makes a planned click on a 50-50 guess would not make sense. So there are three ways to deal with this: Allow "planned" clicks, trust (or not trust) a player that claims he had never seen the board before, or make a perfectly randomised Minesweeper-game (I don't know to what degree that is possible).
My very own conclusion
I'll try to list my constructive suggestions here. I think the notes are the key part of the text, but I'll make a summary of some suggestions right here:
Ø Make a democratic election on different rules in the addicts group, and let everyone vote. The polls should mainly contain "neutrally written" suggestion with alternatives yes, no and don't know/care.
Ø If no agreement is possible, split the community into leagues according to what rules they prefer. There could be different world records and rankings for each group. I guess that the owner of the site is free to choose the rules, but some democracy inside the league should be preferred.
Ø Someone will always try to cheat. Rules should include what kind of evidence is needed (screenshot/video), and eventually how to prove that a player did not prepare himself for a board (If he for instance clicks the correct square on a lot of 50-50s or claims to have made a lucky misclick. Should there be reason not to believe him, and should he be punished? ).
Ø Rules should be made on how to deal with any kind of pre-knowledge of a board
Ø I know that Eduardo Cros has suggested that we make our own Minesweeper game. If it's possible to make "more random boards, boards that doesn't repeat cyclic or just improve the randomisation in some way, I support his project totally. (I'm not an experienced programmer though. [I guess this counts as a suggestion in the project]). Since there are a big lot of Minesweeper-clones all over the net, I guess Microsoft doesn't mind that we make one (?). Maybe a multiplayer version for the Internet could be used to make tournaments, like the world championships Matt describes in his essay.
- Jon Simonsen