Craft In Practice: a detailed game-design analysis of The Reliques of Tolti-Aph

I realize that tastes differ and people like different stuff in IF and all that, but I think that The Reliques of Tolti-Aph is a very, very bad game. Unlike most bad games, it has a perfectly serviceable premise and setting and so on, and of course Graham Nelson is a fine writer. But the actual game design is ghastly. It's so bad I am almost inclined to think the mistakes are deliberate, although a more realistic answer is that it's the usual combination of sloppy work and total lack of concern for the player's perspective. However, the catalog of game-design crimes in The Reliques of Tolti-Aph does have one major upside: discussing them provides a great applied primer in Things Not To Do In Your IF Game. So this isn't going to be a review in the standard sense — instead, it's a spoiler-heavy analysis for people who've already played the game.

Disabling Undo and Limiting Saves

The most glaring problem in The Reliques of Tolti-Aph (not the worst problem — I'll get to that one later) is that it disables undo. I don't have much new to say here about why this is lame (feel free to look at my review of Negotis), but I do want to point out that this is a good example of how game-design elements interact. Normally, a room like the swamp where going the wrong direction kills you is no big deal — you read the death message, type undo, and then go the correct way. But when undo is disabled, then going the wrong direction in the swamp is suddenly a big production, and you have to jump back to your last save point. Some people's reaction here is "Oh, but that just reminds you that the swamp is dangerous and you have to be careful!" These people need to be kicked in the shins, and laughed at the next time they typo "w" when they meant "e".

Similarly, the game requires you to solve a puzzle (possibly even several puzzles) to earn the ability to save, and you can only create a savepoint a limited number of times (three, according to David Welbourn's detailed walkthrough). This does in some sense "fix" the first problem that comes up with just disabling undo, where people say "well, fine, I'll just save all the time then, if you won't let me undo," but this is like fixing a leak in the sink by burning down the kitchen. By limiting the save command, you force people to replay the beginning of the game over and over again until they solve the initial puzzles necessary to save. Some people will do this — when I was younger and had way more free time, I once played a game that had save disabled entirely — but in general, every time you end the game and the player doesn't have a recent saved game, you're losing some noticeable fraction of the audience.

The save and undo restrictions are presented as necessary in order to fix issues that would otherwise arise due to the random combat/skill-check system. In general, when you find yourself as an author saying "well, I guess I'll have to make things more tedious and less fun for the player so that they can't circumvent this other design decision", that should be a big red flag. Stop and reconsider the other design decision — is it really adding much to the game? Is it adding enough to outweigh the annoyance to the player? If you have to do it, can you minimize the annoyance at all? For instance, an obvious way to minimize the annoyance in The Reliques of Tolti-Aph would have been to restrict save and undo inside combat only.

Yes, they could still restore a saved game after dying and retry combat immediately, but right now they can do almost the same, by running from their last savepoint to the combat. And yes, they can undo failed skill checks that don't result in combat, but by my count there are only two of those: climbing the ropes to the archivist (which is trivial with the spell, and requires you to be level 5 otherwise), and climbing the outcrop in the river (which is a badly-done puzzle anyway — more about this later). Allowing undo to handle these would be just fine for the first skill check, and would clue the author in that the second is poorly designed and should be modified.

Bottlenecks and Setting Goals

One of the most important things to keep in mind as an author is that players have no idea how the game is going to play out. They don't know what their goal is, they don't know where the interesting areas of the game are, and they don't know what puzzles need to be solved first. To help the player out with this, you-the-author have two basic tools. You can set the player a goal and say "hey, go do this", and you can set up a bottleneck and say "you can't do this yet". The Reliques of Tolti-Aph is astonishingly bad at the former. It's one of the few games I've played where the overall goal of the game (to become a level 5 mage) is only explained in an error message (if you try to leave the city). Most players are good sports and will wander around solving puzzles just because they're there, but this doesn't mean having an overall goal is unimportant. For instance, in this case, knowing you need to get to level 5 means you should try to get as many experience points as possible, which means you need to make sure to do all the semi-tedious things like making sure to cast each spell twice, and fighting monsters if possible rather than avoiding them.

The game is slightly better on the bottleneck issue, since there's the wyvern sitting at the halfway point, and it soon becomes obvious that you should work on all the stuff down below first and then you'll be beefy enough to take on the wyvern. But that's pretty much it — once you get past the wyvern the game will cheerfully allow you to charge into the maze despite being too low-level to take it on, or the Old School cellar despite not having the magic missile spell. One of the major problems with Enchanter-style games is that the player can be confronted with a puzzle where it's not clear if the puzzle needs to be solved with clever lateral thinking, or if there is some specific spell that needs to be used that the player doesn't have yet. Good game design in this sort of game means setting up bottlenecks to ensure players have a certain spell before they get to the part of the game where it's vital. One way to do this might be to get rid of the dragon sleep spell, and thus force the player to fight the wyvern with the web spell — this would make sure they have web before they get to the archivist's temple. I'm not sure what sort of puzzle you'd come up with to require magic missile before the dead-end in the Old School cellar, but presumably you could have something flammable you need to destroy, or a monster that is especially and obviously vulnerable to the spell.

This undirectedness is less of a problem for the game in the smaller scale, partly because the game handles things better there and partly because small-scale undirectedness doesn't stop the action entirely. For instance, there's no clue where to use the infravision spectacles, but, ok, the player can just wear them and wander around, and when they find out they'll find out — not knowing doesn't block off major portions of the game. Similarly, it would be nice if the room description in the ziggurat mentioned that there are exits on four sides for dumb people like me who assume it only has a north and south exit, but this is something people will eventually notice on their own. There are a few decent bottlenecks near the beginning of the game — the combat with the group of goblins is pretty much unwinnable without being able to cast make sanctuary and being of a certain level, so it makes sure those have happened. But overall, The Reliques of Tolti-Aph is a good reminder that authors have a responsibility to give some direction to players if they want to avoid player frustration.

Subsystems in General

This will no doubt startle people who have read this far, but I do like how the game has a number of interconnected subsystems the player can learn about. It is a cool idea to have both a combat system and a magic system, and the player and PC can get more skilled in each, and, best of all, they're tied together by different resources. There's a direct connection in that both magic and combat draw off the hit point pool, which leads to interesting choices to make about using magic in combat — do I spend X hp casting this spell to possibly avoid Y damage in combat?

Similarly, there's an indirect connection created because the magic system requires spell components and the combat system requires new weapons and armor. You can use the silence spell to get a new weapon and armor, and you can fight the goblins to get new spell components. This is cool because the rewards for solving a puzzle enhance the breadth of the player's powers, and not so much the depth. It's also cool because someone who's better at fighting can get further on the fighting puzzles and use those to help themselves along on the magic ones, and vice versa.

Having said all that, you will not be surprised to hear that I have a number of quibbles about the specific implementations of the two subsystems.

Random Combat and Skill Checks

The Reliques of Tolti-Aph gets off to a good start, game-design wise, by saying roughly "things are set up to be random, but in general, if it turns out you can only beat something by repeatedly trying, you're doing it wrong." This is a great principle for people who include randomness in their games, and it is too bad that The Reliques of Tolti-Aph so rarely sticks to it. For instance, take the fight with the four goblins: it's quite difficult (and unlikely) to beat them without running away mid-fight. So is the solution to assume you're doing it wrong, go away, and find an appropriate spell to beat them? No, of course not — it's to run away mid-fight and retry until you win. Or consider the marsh wraith on the pontoon, where it's difficult to fight but there is a spell to beat it easily (satisfying the principle) — but, unfortunately, a good portion of the time it'll kill the PC as soon as the room is entered, making it impossible to use the spell. And since save is so limited, people's natural reactions will be to shy away until they find the nonexistent other ways of beating the wraith or entering the room.

There are similar miscues in the skill system. The outcrop in the river requires a roll of 15 or better, and there's no solution other than retrying until you succeed, taking damage each time. The werespiders, on the other hand, require a roll of 17 or better, and the expected solution is to use the silence spell. This isn't a totally exact comparison, since you have to make multiple checks to get past the werespiders, but the player doesn't know that initially. Another irritating thing about the outcrop is it can turn into a dead end — if you climb up with only a few hit points (not unlikely if you've been repeatedly trying to climb, since the lack of undo makes it irritating to do otherwise), you may find yourself either stuck or dying on the climb back down, since that causes damage also.

Really, this is all a solved problem and has been for some time. The solution is, don't use random combat and skill checks because they're dumb. You can look on google groups for discussion from 1995 about randomness being bad. 1995! Over ten years ago! While it's possible to add randomness to combat-type situations in an interesting way, the trick is to put it in at a higher level, and create a dynamic situation for the player to have to react to. The success or failure of a single command by the player should always be non-random; the challenge comes from picking what the correct action to do is, not from retyping it until it succeeds.

Spells and Spellcasting

In contrast to the combat system, which is fairly clearly explained in the about text and the game feelie, the spell system is awful to work out for new players. There's your journal, but you have to hunt through the feelie to find out the cryptic keywords to use to look up entries, and the entries are presented out of order, and half of them are unhelpful anyway. If someone is used to Enchanter-type games they'll presumably be able to work out >MEMORIZE and >SPELLS on their own, but then they'll be confused when it turns out you can't cast spells from scrolls, that casting spells drains strength, and that memorizing a spell means something totally different in this game. It's absolutely vital that you explain major subsystems right off the bat to the player. If they're important enough, have a tutorial section or something, with a command summary in the about text. This is especially true when it works kind of like another system, but with important differences — the about text should make the differences clear (though it also shouldn't assume people are familiar with the original, unless that's really necessary).

Spell components are another good idea poorly implemented. There seems to be some design confusion over whether the purpose of a component is to provide a pre-puzzle you need to solve before being able to cast the spell (like the sanctuary spell), whether it's intended to limit the number of times you can cast a powerful spell (like summon elemental), or whether it's just to be irritating (like magelight). I call out the last as a special category, since a large part of the time, it seems like the main purpose of components in this game is to have you accidentally run out of the thing you need, making the game unwinnable. It's like the old-time games that had a lantern with a battery, and if you wasted too many turns, the battery ran out, and ha ha, you're screwed. This isn't like Journey, where figuring out which components to conserve was part of the game meta-puzzle. It's purely a device to punish the player for casting fashion staff too many times.

The auto-selection of components to use up can be a problem too — sometimes I was annoyed that the game destroyed something I wanted to keep, and there was no clear rule for how it picked (and, again, because undo is disabled, it's not easy to deal with surprises like this from the game). Games shouldn't include auto-selection unless it's going to be completely smooth for the player, either because the player can override the game's selection, or because the things it's auto-selecting between really are indistinguishable.

The >SPELLS listing is kind of a mixed bag. I appreciate that there's a separate command to just list the spells, but it's just irritating at the end of the game to have inventory list all your items and then 20 spells so the actual item list scrolls off-screen. The design point here is that it's important to remember how things will look both at the beginning and end of the game — if you have lots of useless items for the player to pick up, are they going to hate you at the end of the game when their inventory is three pages? It's also good that the spells list gives spell type, cost, and material component required, but where's the spell description? The PC knows how fashion staff works, but I don't, and figuring out the mechanics of that is not a fair puzzle to give the player (especially when, as mentioned above, casting it too many times while experimenting can make the game unwinnable). When the spell uses up a limited component, like force labyrinth, the experience is even worse.

Finally, I feel like the particular spells that were included need some editing. There are a number of spells that you learn so late in the game that they're rarely useful (corruption, summon elemental, ironbones). This is especially true with summon elemental, which consumes its material component, so how many times are you really going to be able to use it? Other spells feel clunky because they're only there for a single puzzle, like silence and dragon sleep. A second use, like there is for aerial shield and web, goes a long way towards making a feature not seem like a deus ex machina. Not all the spells are bad about this, I should add — mend, in particular, probably has a dozen uses and is extremely fun to experiment with (which encourages players to try spells on different items in the game, which is very important in this style of game). A few more spells with as many applications would have been very nice.

Miscellaneous Issues

There are a handful of things I haven't touched on yet in The Reliques of Tolti-Aph that seem like they're worth some discussion from a game-design perspective. To start off with, here is the promised worst thing in the game:

You prepare yourself with due solemnity, and yet - and yet - you sense somehow that the proper words must be used, if the prayer is to be answered.
I only understood you as far as wanting to pray.
I only understood you as far as wanting to pray you loose-ro clouds be with me.
Communing in this special place with the swirling clouds - but was this not, a moment before, a clear autumnal sky - you feel seized with a sense of purpose and ability.
I am awestruck by how bad this is. What possible clue is there that you should keep trying other text when the initial pray attempts all end in failure? And it's not encouraging failure either, but cold, indifferent sir-I-do-not-know-you parser failure. If the player tries anything close to right, they should never get a parser failure — that signals that not only are you doing the wrong thing, but you're on the wrong track entirely.

This passage also points out another irritating thing, where the game teaches you a spell without telling you that's what it's doing. Since a new spell represents a substantial increase in the player's capabilities and probably allows access to new areas, it's very important to make it clear when it's happened (and given that The Reliques of Tolti-Aph happily tells you "You progress to level 3!", you'd think it could also let you know that you learned a new spell). Speaking of spells, this is a personal preference, but I don't approve of getting experience for the first two times you cast a spell. In general, people expect things to do something only the first time, or every time (or, rarely, three times); if you make it happen twice, they will assume it'll happen every time.

The song on the loom is another game-design decision which seems to be there purely to irritate the player. The first time, >SING SONG pops up a quotebox with the lyrics of the song, giving a clue. After that, it doesn't print the quotebox. This is such a bad idea — players have the right to see clues as many times as they want (in some situations they could undo to see the clue again, but of course that's disabled). I'm not fond of this clue anyway, since it doesn't seem to me that there's much reason to think it applies both to the painting and the loom itself, given that neither is obviously hiding something. People are likely to find one application of >TURN, assume that's all the clue is there for, and miss the the other.

With the large number of traps in the game, it's frustrating how seldom the player can do anything about them. The most grievous error is in the worst possible place, the first room of the game. People are likely to cast detect trap as they're trying out their spells, see that there's a trap on the doorway, and then beat their head against the puzzle of trying to disable that trap. But the actual best course of action for a new player is to just ignore the detection message, walk through the doorway, ignore the "I bet there was a way to avoid that trap" message, and go on with the game — because while you can avoid the trap, it's a totally complicated puzzle, and almost no new player is going to find it. And forget trying to poke a stick in the doorway or something to set off the trap — there's only the one pain in the ass way of doing it, which doesn't even require detecting the trap in the first place.

The description of the starting room in The Reliques of Tolti-Aph is a good reminder about how important it is to ease the player into things. I've seen a number of people start up the game, collide with the room description, and come to a confused halt. I think the main problem is that it starts out badly. The first sentence is colorful but has no information useful to the player for navigation or getting their bearings, and indeed it mentions "east" and "west" but not as part of laying out what's in those directions. Obviously this is mostly a style thing, and there is a place for fancy room descriptions — it's as a reward for getting past obstacles, like the top of the ziggurat. Early in the game, clarity absolutely has to win over style, or the player will be overwhelmed.

The maze is a big enough deal in the game that it seems like it needs its own gripe section. Trouble starts almost immediately once you're in. The natural reaction by this point in the game is to use a spell to try to get one of the leaves, since physical actions are so often ineffective or unimplemented (see the earlier point about trying to disarm traps with a stick). Once you get the quest, then you know what you have to do. Of course, knowing and doing it are two different things — some are trivial, like slaying a monster on level 4, and some are almost impossible without extensive research on the maze in previous games, like finding Eurydice. The random assignment of quests is a serious pain, and the player has no idea that they vary so wildly in difficulty, so they're likely to get discouraged after a failed playthrough with one.

And they're likely to fail maze playthroughs, a lot. I assume the "magic user has run out of maze!" bit is some limit imposed by the Z-Machine's dynamic object creation stuff, but it feels totally inexplicable, and it actually penalizes people for sensible behavior (you're better off charging ahead into the deeper parts of the maze rather than carefully exploring each level, because you only have a limited number of rooms you can enter). Plus there's the whole thing about some exits being blocked for no good reason. You can force that with the force labyrinth spell (which, again, you learn without the game telling you that you learned it), but that's expensive in terms of components and is unlikely to help a huge amount. Then if you survive for a while and don't hit the room limit, the maze is going to get boring fast. You might not win your quest, but the monsters in the maze become just experience-point prizes once you've found a few of the magic items in the maze. This just isn't balanced, and it's not particularly fun. It'd be one thing if it was an optional puzzle that was clearly marked "for experts only" or something, but no, going in is pretty much required if you want to get to 5th level.

It seems appropriate to close this essay by talking about the game's ending. Man, what a lousy way to finish. First off, there's no "hey, you hit level 5, you win!" message when you hit level 5, or even a "you can leave the city now!". You just have to remember that's the point of the game, go back to the entrance, and leave. At which point you get an anticlimactic message about how you won, hooray, PS you're still kind of a loser. Gee, thanks. This is, again, the sort of thing that could have been fixed relatively easily. The maze is already the hardest thing in the game, and the game already says that the maze was the old testing-ground for mages, so how about making the goal of the game to solve a quest in the maze? Put the you-have-won bit when you emerge from the maze, and it'll come at a point in the narrative when the player feels like they've really accomplished something. Given that this is a try-stuff-randomly kind of game you don't have to end the game there absolutely, but you at least have to make it clear that this was the climax, you've won, you can keep playing or leave the city as you like.


So, yeah, I dunno. There's some decent stuff in The Reliques of Tolti-Aph I haven't touched on — I was particularly fond of the puzzle for getting past the sand-filled passage, and the writing is definitely charming in spots. But overall, man, it is honestly hard for me to understand why someone would release something like this, with design decisions that seem only there to be irritating. If this was "Annoyotron, Special I7 Edition" that'd be one thing, but as it is, I don't get it. Still, I'm definitely glad I've played it. Zarf's games, for instance, are not particularly useful from a design-analysis perspective because they work so smoothly — there aren't any seams to pick out how the individual parts work. This game, on the other hand, is all seams and misfit parts, but it's just that quality that makes it interesting to talk about.