Infocom, puzzle design, game length

I’ve been interested in learning about puzzle design for longterm games vs short games, and I’d like to learn more about it from the community, because it seems like this is something people have thought about a lot.

Many of the short IFComp winners have much more text than the early Infocom games, but those early games could take a month or more to win; many people did not get frustrated during while working on the puzzles, or got frustrated but enjoyed it. I played Adventure after having tried many more ‘new school’ games, and I played it over a month, and I didn’t get frustrated; I loved the challenge it provided.

So where do games with a month’s worth of gameplay get it from?

-Is it the puzzle ‘flowchart’ (Solving A unlocks B, C, and G?)? I considered this, that perhaps the easier games just had a linear flowchart with few puzzles while long games had a broad flowchart. However, Spellbreaker is pretty linear, and is one of the longest Infocom games. It has a structure a lot like Violet, with a tight opening and then a gentle broadening before a finale.

-Is it the difficulty of coming up with a solution to a particular puzzle? This has to be part of it; if all the puzzles are easy, the game will be over soon. Although A Mind Forever Voyaging had no real puzzles (except for the endgame), but seemed to provide a LOT of gameplay time with, again, less text than many modern classics.

-Is it the red herrings? This is one thing I’ve considered: that early adventure games had a lot more red herrings, and that Infocom (and other publishers of similar quality) made games fun by providing interesting responses to interactions with the red herrings.

-Is it the cruelty rating on the Zarfian scale? Many old games locked you out of victory unknowingly (like Curses! and Spellbreaker), requiring you to play as far as you can until you get stuck, then replay, changing your actions to make that part winnable.

My guess is that the last two are some of the best ways to make a long-lasting, fun game, but I don’t know.

By the way, I’m talking about a specific class of game here. I feel like there are several classes of Big Games (like Gijsber’s first and second ideals):

Class 0 (the one I’m talking about): 10+ hours of gameplay, much of it retrying the same portions over and over

Examples: Adventure, Zork, Hadean Lands, Curses!

Class 1: 4+ hours of gameplay divided into manageable ‘chunks’, with each chunk completable in an hour or so, and with little to no need to revisit puzzles.

Examples: Anchorhead, Counterfeit Monkey, Worlds Apart, Blue Lacuna, King of Shreds and Patches, Andy Phillips games

Class 2: 2+ hours of gameplay, designed for maximum player accessibility and offering branching or procedural gameplay for replay value

Examples: Superluminal Vagrant Twin, 80 Days, Choice of Games

I feel like I could confidently work on a Class 1 game that I felt good about (basically stringing IFComp-size games together) or a Class 2 game (where writing content is the main obstacle), but I have no idea how to write a Class 0 game. Very few have been written in the last decade (Make it Good and Hadean Lands, maybe Endless, Nameless and One Eye Open, although the last one is fairly short for class 0). I really like class 0 games; does anyone have any suggestions on how to write one?

1 Like

The two strategies for your class zero are clear:

(1) Put in lots and lots of support for shortcuts, so that even though the player is notionally “retrying the same portions over and over”, they spend very little time or effort repeating familiar tasks.

(2) Release your game in 1996 or earlier. Several years earlier, if at all possible.

I don’t have answers, but would like to add some thoughts;

I would say, class 0 games have a lot more logic in them then the others. In other words, the puzzles are more complex, possibly with partial or multiple solutions. Also, to be entertaining, those puzzles also need amusements and side effects.

Another way to analyse it is to suggest the puzzles (or main gameplay elements) are extensions of the world model. older games, i think, had a richer world model in general;

typically a particular game’s puzzles would involve doing something slightly outside of the scope of normal game world modelling; mixing liquids, combining things, tying things together, riding around on vehicles, for example.

When you extend the world to accommodate this, say for a specific game puzzle, next thing you know there are all sorts of crazy things that the player can now theoretically do with other things in the game and with the correct things but in bizarre combinations. Most of these cases needs to be catered for, resulting in a mixture of constructive and comedy outcomes.

So the player is entertained even when not actually solving the problem.

IMO, class 0 games are characterized primarily by big maps, and by big, I mean roughly 80-100 rooms.

Naturally a large map like that benefits from a large number of puzzles, ideally organized into a big puzzle dependency chart, but as you point out, that’s not required. AMFV had a huge map. I think there are roughly 100 rooms in the simulation’s map, and each simulated time period had substantially different implementations of those rooms.

(Didn’t somebody say recently that one of the proposed approaches to limiting the scope of IFComp entries was to limit the number of rooms? It’s obviously the wrong idea for IFComp, but I think it reflects on the ingrained assumption that “big games” means “many rooms.”)

Interesting, @dfabulich. I found this list of rooms in Infocom games at this website:

[code] VI Game Statistics

Game Version Rooms Words Objects Opcodes
(int) (takeable) (total)

Zork I 88.840726 110 697 60 6798
Zork II 48.840904 86 684 50 6804
Deadline 27.831005 51 656 37 6977
Zork III 17.840727 89 564 23 5952
Starcross 17.821021 86 557 25 6566
Suspended 8.840521 63 676 33 6902
The Witness 22.840924 30 715 22 8945
Planetfall 37.851003 105 669 45 7879
Enchanter 29.860820 74 723 33 8070
Infidel 22.830916 77 613 57 7386
Sorcerer 18.860904 84 1013 36 8963
Seastalker 16.850603 30 911 15 14460
Cutthroats 23.840809 68 790 21 12600
HHGG 59.851108 31 971 45 10723
Suspect 14.841005 57 674 43 10737
Wishbringer 69.850920 52 1043 35 16223
AMFV 79.851122 178 1812 30 18696
Spellbreaker 87.860904 79 850 60 12472
Ballyhoo 97.851218 36 962 42 15132
Trinity 12.860926 134 2120 49 31389
LGOP 59.861114 75 978 41 13763
Moonmist 9.861022 69 955 26 15900
Hollywood Hijinx 37.861215 67 854 58 10355
Bureaucracy 116.870602 50 1416 44 24116
Stationfall 107.870430 105 789 53 10662
Lurking Horror 221.870918 71 773 44 12398
Nord and Bert 19.870722 41 1230 69 13831
Plundered Hearts 26.870730 57 816 28 13859
Beyond Zork 57.871221 128 1569 77 32778
Border Zone 9.871008 111 803 42 11273
Mini-Zork I 34.871124 69 536 46 5204
Sherlock 26.880127 92 1194 67 19702
Zork Zero 393.890714 215 1624 106 23587
Shogun 322.890706 75 1389 63 28346
Journey 83.890706 - 27 - 16187
Arthur 74.890714 90 1059 32 28242[/code]

An interesting chart. To my eye, not all of the games on the list are “class 0” puzzle fests, but all of the ones that ARE class 0 have a bunch of rooms. Perhaps the only glaring exception is HHGG. IMO HHGG has a large puzzle list instead. (I can think of at least four puzzles that have to be solved in front of Arthur’s house.) But perhaps HHGG is really class 1?

Side note: Back when Infocom games took months to solve, there was no internet to quickly obtain hints from. You had to buy a physical hint book or work it out. I remember as a teenager actually waking up one morning due to a flash of insight how to get past a puzzle in Trinity.

They were considered full-sized games and often it was assumed you’d be playing it over weeks.