Minesweeper is a classic! Most people would agree with this statement... most! As a Minesweeper player, I had assumed that, like its Windows neighbour Solitaire, everyone knew how to play it and, even if they didn't love it, they would happily occupy a boring afternoon at work or school in its company. To my surprise there is, however, a contingent of people for whom Minesweeper sucks - youtube search ‘Minesweeper sucks’ for some comic evidence.
Our soon to be released game, Grave Matters, is based on Minesweeper. The squares replaced with graves, and mines replaced with killer Steamborgs (steampunk cyborgs) but the puzzle logic is more or less the same. As a game developer, when you build upon an established puzzle game, you inherit the proven gameplay and the good will of the fan base but you also inherit all of the problems inherent in the design. Our extremely difficult job as designers is to then improve on the original and attempt to overcome any problems in its design. Our attempt at solving the problems of Minesweeper follows.
Problem 1. Lucky Start
The most likely complaint you will hear from a Minesweeper player is that for a game of logic, the beginning of the game is too heavily subject to luck. When you start a new game your first move is always safe (you won't hit a mine) but there is no guarantee that your first move will reveal enough information to stop your second, third and potentially further moves becoming complete guesswork.
One solution we considered to overcome this was to seed the board with information. Effectively, designing each level to start with a few bits of information already revealed, as you would in a chess puzzle - we use this approach in the Tutorial but not in the main game. Beyond the obviously laborious task of designing every level, we felt that this approach took away from the replayability that you get in a game where the levels are procedurally generated. An additional weakness of this approach is that it removes some of the player's personal strategy in favour of a 'correct’ solution.
The solution we finally opted for in Grave Matters was the introduction of items that in different ways help the player to overcome the unknown. The first item, Dog-Sense, overlays a heat map on the grid that shows the player the danger areas in the level but without giving them the precision to pick out the location of an individual Steamborg (mine). The second item, Lightning, randomly vanquishes one Steamborg in the level, reducing the overall danger and giving the player the definite location of one Steamborg. The final, and perhaps most useful item, is the player’s loyal companion, Pup the Dog, who acts as a lifeline, sacrificing itself to save the player from one mistake.
The second most likely complaint you'll hear from a Minesweeper player is that for a game of logic the end of the game is often too heavily subject to luck. Nearing completion of a game, it is very common to end up in a position where you are again forced to make an uninformed guess, made extra frustrating by the sheer amount of time you have invested so far. Whilst this does add a great deal of tension, it occurs far too often and means very few players ever win at Minesweeper. Instead they will spend ages clearing a board using logic to then lose the game on luck.
In Grave Matters, we differentiate between a level being Passed and a level being Perfected. When a player has dug up a certain amount of the available graves (e.g. 50%), the level is Passed, the gates unlock and they are allowed to escape to the next but they can also choose to continue and attempt to perfect the stage and be heavily rewarded if they do. This two tier system, combined with the inclusion of items, allows players in most cases, to escape a level at the point when they feel it is becoming too much about luck.
For me, this is one of the biggest problems inherent in Minesweeper (it probably should have been Problem 1) and the reason why some people just do not ‘get the game’ or at least find it harder to pick up than most puzzle games. I believe the problem is that when you dig up a cell in Minesweeper, you find a number but this number doesn't tell you what you’ve unearthed, it instead tells you about the cells adjacent to it. The first thing we do mentally when we dig up a cell is to relocate this information onto the surround cells. I think the information is in the wrong place...
Not everyone agrees with me on this, in fact we have regular debates in the office as to whether this is even a real problem. Regardless, it has occupied more of my design thinking than perhaps anything else during development. I believed that to solve this problem, we needed to move the information from the dug up cell onto the adjacent cells.
Our first attempt was to implement a system of highlighting, where we showed the danger level of a grave (cell), after a neighbouring grave was dug up. One precaution we had to take with this approach, was to not reveal more than the player would know from the equivalent number system. We accordingly tracked the ‘potential’ danger of a grave based upon all of its dug up adjacent cells (as we do in our mind). This then gives us a danger percentage for each grave, which increases or decreases as we unearth more information. At this point, you may have spotted the problem with this approach.
The danger percentage was converted into a colour system; white, yellow, orange and red, indicating the perceived danger of the grave at the current point in time. This system fell apart when we tested it, as it was misleading to the player. Digging up an orange cell from amongst a group of reds (the logical choice given the available information) and being rewarded with death, unsurprisingly, caused anger amongst players. It turns out that if a computer is providing you with information you expect the information to be 100% accurate, you do not want probable outcomes, you want facts.
We kept this system in the game as a means of assisting new players but simplified it considerably, opting instead for two states, safe and potentially dangerous. If you dig up an empty grave, all of the surrounding graves are guaranteed to be safe and so we hint this through highlighting of the safe cells. Likewise, if you dig up a cell with a skull icon, one of the surrounds might be dangerous, so we err on the side of caution and indicate the surrounding cells as potentially dangerous. Digging up a cell you think might be dangerous and finding it is safe is much less frustrating than the opposite.
If I am honest, I don't think this is a perfect solution yet.I think there is an alternative way to do it, where the numbers could be completely removed from the dug up graves. I tend to find myself scribbling new solutions to this problem on a near weekly basis. If I ever solve it, I think I will be able to massively multiple the intuitiveness of Minesweeper. Then again, it won't really be Minesweeper anymore; it'll be some other new game of deduction, likely to have its own inherent problems, a challenge for another time.
Playing expert on Minesweeper is more akin to solving algebraic equations than playing a game, as difficulty is controlled solely by the proportion of mines in a stage. There seems to be a natural tipping point, beyond which the game just becomes too complicated and simply a arduous task of addition.
We wanted to have variable difficulty in Grave Matters but we did not want to achieve it purely by saturating a level with Steamborgs. Obviously, different players will have different tolerance to difficulty, so the first thing we did was divide the game into levels, each getting progressively more difficult and thus accommodating varying skill. Beyond this, we also vary the arrangement of the grid on different levels and introduce new mechanics to some of the later ones, adding variety and a necessity for different modes of thought.
One of the key features in Grave Matters that influences difficulty, which was actually added quite late on in development, is the time pressure. The game is turn based but the player’s health is constantly decreasing and is only boosted when they dig up a grave – akin to playing chess on clock. What this enables us to do is have a less complicated puzzle that is just as taxing on the player but taxing because it forces them to think on their feet rather than spend an age carefully considering every move.
Ok, so this is not really a gameplay problem but it is so serious a barrier to mass adoption that we felt we had to address it. Minesweeper is hideous; a grid full of numbers is visually appealing to only a minority of people (I say minority but likely mean no one). No matter how pretty you make the graphic of the mine, it still looks like Sudoku or worse still a school maths problem and for many people, maths was not a favoured subject at school.
In Grave Matters, we went OTT on the graphics; 3D characters, sepia styled colour palettes, gothic steampunk environments, cut scenes etc. Another major difference between Grave Matters and traditional Minesweeper is our use of a honeycomb/hexagonal grid structure. The grid creates a nice twist to the gameplay but also gives us a number of visual benefits. For one, it breaks up the uniformity of a square grid, making it much more visually appealing - bees seem to know what they're doing. In addition, each grave having less surrounding graves means we can replace the ugly numbers with icons (clusters of skulls) without affecting comprehension - it could also probably be argued that six surrounding cells is significantly better for short-term memory retention than eight but I'll leave that point to the psychologists.
Whilst I’m sure we haven’t cracked all the problems in Minesweeper (for now), we're happy with what we have developed in Grave Matters. It is our first attempted solution to the problems and we are confident that we’ve improved on the original but the players will be the real judges of that. Grave Matters will be released in a few months times, till then, sneak updates will be posted to our Grave Matters Facebook page (http://www.facebook.com/gravemattersgame).