I’ve been reading on posts regarding puzzle designs, and I don’t think I’ve found any reference regarding size. It seems like either they do it really quickly, as in game jams, or they take really long time, as in months or years.
In my view, it really boils down to Verb+Noun. The number of locations is actually rather irrelevant to me, even as a player. If the game features large featureless maze, then that’s undesirable. However, I have no problem with large, expansive landscapes for atmosphere, assuming that it is consistent with the tone of the game.
What it comes down to, both as a player and author, is the interactivity of the game. That means how natural are the objects implemented? Some games advertise as having so many words, but how many are distinct and how many are synonyms?
Assuming distinct verbs and nouns, a 2 word parser game will take V×N actions to implement, assuming that we’re disregarding Time element. Hence, a 3 word parser will take V×N×N. If you consider Inform with its extremely capable parser with large Verbs, then you can see why a significant game will take years to implement!
Then there is the type of puzzles:
- Pattern recognition: moon logic, puns, and the likes. Solving these kind of puzzles elicits “Aha!” moments.
- Combinations: logical sequence. 15 puzzle, sudoku, rubic cubes. Solving these puzzles requires “look ahead” minmaxing, type of thinking.
- Personality Matching: different people have different reactions to the same things. Dating sims, wargaming, RPG.
- Systems: These feature some algorithms and/or mathematical equations that you need to figure out and solve in order to win the game. Card games, simulations, RTS, action strategy games.
I find that the majority of IF puzzles lies in pattern recognition type. Some has personality matching types. As for the rest, very few. About as rare as those walking sims type. I don’t have anything against walking sims type, if they are of the “iyashikei” or “healing” category.
So, back to size. In order to properly implement a game you would either have to do V×N or V×N×N, multiplied by modifiers (adjectives, positional, possessive), multiplied by combinatorial puzzle factor. Once I realized this, I decided that number of locations is not a significant factor in game size. N^2 dwarfs just about everything else.
So, then what should be a good number for games? Counting the number of objects (or locations, assuming their numbers are about the same):
Less than 10: tiny / experimental games.
10-25: What I call Alphabet Design games, listing objects by letters. Small games, like short story (10) or half hour episode (25). Old Scott Adams games.
Up to 64: standard. Most 1980s parser games are of this size. Modern Introductory games are of these size.
Up to 256: large. Large Infocom games. Think z5 games. Level9 games. Professional games.
More than 256: Mega games. Think Sierra On-Line Time Zone, with 1200 locations.
I like small games, that I can play bite size sessions. As far as making them, I can’t imagine doing larger than 64 locations, unless I’m doing it as a professional. The combinatorial explosion does factor a lot in this case, especially if you’re not doing it using 2 word parser.
2 word parser, especially those with limited verbs, can be done relatively quickly, even with large number of nouns. The question is: do you prefer simple parser and rich worlds, or rich parser and simple worlds? Seems to me people here prefer the latter, while general populace prefer the former. There’s nothing right or wrong either way, but if you’re looking for commercial success, maybe you should consider building games with rich worlds, while sacrificing actions. That looks to me as the secret recipe for success both for choice games and visual novel. IMHO.
With systems puzzles, I find myself spending a lot of time balancing the game. If I find myself getting bored, it means the game needs more balancing, and the process continues for eternity. Obviously, I don’t release too many of those kind of games.