Game vs Metagame

Tam’s recent post about solved games really crystallized something for me with regards to why I do and don’t like certain games. In particular why I will sometimes really like the concept of a game and yet end up absolutely loathing it in practice. In a word, metagame.

…game
I’ve come to realize that the existence of a codified metagame for something is a pretty good sign that I won’t want anything to do with it. If there’s a way of doing things that is the way, the truth, and the light, then I get to choose between following the crowd or trying to go my own way knowing that I’m actively not playing optimally. The former tends to lead to me getting bored quickly, particularly since the one right way rarely ever syncs up with the way I want to play, and often relies on degenerate strategies. The latter puts me in a spot where I don’t get to take joy in improving, since I’m aware that I’m actively not playing ‘the best way’.

Beyond that, I’m being forced to spend time figuring out how to play the game ‘properly’ rather than actually playing the game. Part of the problem with metagame for me is that it is, by definition, external to the game. So I end up having to look for FAQs, wikis, or even *shudder* official forums to even begin to figure out what I’m ‘supposed’ to be doing. And then, without fail, I learn that I chose the wrong class, hero, skill set, or whatever and I can either start over completely or bull on with the knowledge that I’m ‘doing it wrong’. At least if it’s a single-player game I can take comfort in playing the character I want to play even if it isn’t optimal. In a multi-player game there’s the added joy of other players more than happy to tell you that you’re stupid and wrong if you dare to step outside of the accepted orthodoxy.

My roots in tabletop role-playing, where metagaming has long been viewed negatively, may also enter into this. For me playing a game is about working within the bounds of the assumptions that are made by the system. Avoiding use of out-of-game knowledge as much as possible is part of this. If a game is well designed and things are messaged properly, I should be able to figure out everything I need to know to play well without having to resort to outside information.

Ultimately, the more time I’m having to spend playing the metagame instead of the actual game, the less I tend to enjoy myself. I want to do my learning as a part of playing, rather than as a prerequisite to even getting started.

Farewell Uncle Pterry

Terry Pratchett has died, and the world is a little darker for his passing.

It would be hard to overstate the impact that Sir Terry had on me over the years; his writing has been a near constant in my life since I really began reading fantasy. It would have been around 1990 or so that a friend of mine loaned me a copy of Guards! Guards! and from there I was hooked. Not that Discworld novels were easy to find in America at the time. For a few years they were out of print here with only used copies and the like available. Most of my copies of the books before Small Gods! are book club editions I picked up used.

There’s one last Discworld novel to be published, the 41st in the series. I have that last bit of Pratchett to look forward to, and then it’s over. I’ll reread the series, have no doubt of that, but for the first time in 20 years I won’t be avidly anticipating the next book. I’ll miss that.

Building Puzzles is the Hardest Puzzle

I picked up Hexcells recently on the basis of Kodra and Rae‘s talking it up, and have enjoyed it quite a bit. I’ve always been a big fan of logic puzzles of any sort, so it was right up my alley. The basic gameplay is much like Minesweeper, where each cell is either part of a pattern (blue) or not (black). Yellow cells are of unknown status and must be marked by you. To solve the puzzle you must successfully mark all of the cells. Unlike Minesweeper, however, each puzzle gives you some information to start with and can be solved with no guessing at all.

As you graduate to more and more complex puzzles, additional methods of providing information to you are introduced. You’ll be told how many blue cells are in a particular line of cells, or that all the blue cells in a particular line or around a cell are contiguous. Eventually you start getting cells that contain a number signifying the number of blue cells within two spaces of that cell. Ultimately you find yourself having to combine information from multiple sources to identify cells properly. It’s very difficult by the end, but extremely gratifying when you successfully solve a puzzle.

There are 3 Hexcells games, each with 36 puzzles of increasing difficulty. The difficult also increases more quickly in each successive game, so you’re quickly solving puzzles harder than the previous games hardest. By the end they’re downright fiendish and I found myself having to take a break and come back with fresh eyes on multiple occasions (thankfully the 3rd game saves your progress on partially completed puzzles). Ultimately though, I made it through and finished the final puzzle.

My wife saw me playing and asked “Are you a bee?”

The third game, Hexcells Infinite, introduces Infinite Mode, which boasts 100,000,000 computer generated puzzles. Put in a numerical seed, get a puzzle. Sadly, but not at all surprisingly, they’re just not as good as the set puzzles you get up to that point. The game’s main puzzles contain a paucity of information, forcing you to determine which of a very few existing clues your latest moves will combine with to identify your next move. The generated puzzles I’ve tried, on the other hand, are overrun with numbers from the very beginning and are mostly a matter of scanning through the puzzle for the next obvious move. Occasionally I’ll have to stop and think a bit, but it’s mostly just click click click done.

The problem is that programming a computer to build interesting puzzles is difficult. Much more difficult than writing a program to solve puzzles; that’s generally trivial if the puzzle is based on pure logic and intended to be human-solvable. Building a puzzle requires you to develop explanations for what make a puzzle good that can be expressed in code. I have no way of knowing how the Hexcells puzzle generator works, but if I had to guess I’d posit that it generates the field of hexes and assigns them each blue or black based on an algorithm that uses the seed as input, creates all the possible clues the puzzle could have, then starts hiding cells and removing clues while checking that the puzzle remains solvable.

With a million possible puzzles, I’d guess that from sheer happenstance some clever ones will show up, but Sturgeon’s Law appears to apply in this case. That said, it’s still a great game and would have been well worth my time if the Infinite feature didn’t exist at all.