A Clockwork Mines
The timer is perhaps the most important thing in minesweeper. Without it, minesweeper would be merely a skills excercise, instead of a speed competition. The thin line between frustration and exhiliration lies on the accuracy of the timer - 19 or 20, 49 or 50. One who decides to retire after hitting 49 in expert is at the mercy of an accurate timer to determine their sweeping career. This all-to-important instrument has been taken for granted as efficient and synchronous with real time, but lately, the timer has been revealed as a faulty doohickey with little relevance to real seconds.
Georgi Kermekchiev was the first to notice the discrepancy between the minesweeper timer and the real time given by the system clock or other means. The first, and most reasonably plausble explanation was that there was a slow-down in playback of games. Either Camtasia had an adverse effect on the timer, or the videos recorded were played at slower than real time. The former turned out to be partly true.
A couple people did some experimentation with time intervals. First are Roelof Smit's results. The even number on the right represents the stopwatch, or real time. The second number is what the minesweeper clock showed at that time:
Twenty seconds lost every 500 is a lot! In fact that's about a 4% reduction! You may also notice that the reduction isn't exponential each time, but this could be partly blamed on human error.
David Barry took things one step further and used varying system resources and various programs running and used the same 500 second mark to gauge the lag. The results follow.
(Started at 2 to aviod any timer jump...Seconds accounted for)
It appears that having programs running and a slower computer slows down the minesweeper timer. Marc Schouten elaborates.
Marc's very fast system yields no difference between minesweeper timer and system clock. The evident difference between the timer being slowed as much as 3 seconds in a minute, and the minesweeper clock running sunchronously with real time is a product of the available resources and memory of a certain system. To take a further look into this, I constructed a chart modeled after one of David Barry's that gives different sweepers' times in their respective recorded videos and the lag time of their game.
Expert videos:Lag time
With the previous evidence, we can determine who has fast or slow computers here. Marc's is obviously very fast, Lanyje's is moderate, probably average, and Dan's system seems to be overloaded. In theory, we can add those approximate lag times to the total and get this chart:
Lasse 52 = 52-53
Does this mean anything? Do all the people with 59's suddenly now have 60's or 61's? Does Lasse now have only maybe one or two sub-50's? I, including the majority of the minesweeper community say absolutely not. And anyways, if you have a game that takes an immediate jump from 1 to 2, you can say, theoretically, that those two "wasted" seconds are now accounted for!
Intermediate is a different story. The timer jump is still very prevalent and is sometimes the difference bewteen tying your record or beating it. Dividing down the lag time to 19 seconds for an intermediate game gives about a one seconds lag to 18.05 seconds, assuming a three-second lag at 60 seconds.
About different versions,
Lance Votroubek on system clocks:
Another interesting point is that there are differences between Matt's timing and my timing of the videos of Lasse Nyholm, Dan Cerveny and Marc Schouten - I get lags of about 3s, 4s and 0s, whereas Matt found them to be about 0.7s, 1.5s and 0s repsectively. So, it appears that the video slows down the timer while you are playing, and also the playing of the video is slower than real time. (Either that, Matt, or you forgot to take into account 1-2 jumps, esp. in Lasse's 50).
When I first came across this bug, I did not like the idea of some people's scores being "artificially" aided, particularly as I run Winamp whilst I play, which I now know to slow the timer. This is important for me as an Expert player, as a 2sec lag is not unreasonable over 50 seconds. Ideally, all Minesweepers would switch to the League version, but obviously, this will not happen, as the vast majority of sweepers are not active. (And also, I think that the League is a slower version to play, because the squares don't open up as fast. So, even allowing for a difference with the timer, you best times wouldn't be as good.) So, the Minesweeper community will have to accept inaccurate timers, along with the 1-2 bug.
Using what Ben Van Calster said, it's fun to think about things like this, but it's just minesweeper!
-Matt McGinley, David Barry