From Introcomp 2011, “Chunky Blues” is a noir mystery set in the town of Bleakton, MA, with some pleasingly innovative puzzle mechanics.
The first part of my feedback is not spoilery, and talks about the puzzle mechanics in general. Then there’s a spoiler space, followed by some more specific discussion of what I think does and doesn’t work.
One of the most challenging aspects of developing any interactive fiction mystery is coping with the knowledge puzzle. You can give the player clues, but how do you know when he’s figured out what the clues mean? Handle this wrong and you have a knowledge conflict: the player’s figured something out but the protagonist isn’t allowed to act on it yet; or, conversely, the protagonist is prompted to do something while the player still doesn’t understand why anyone would want to do that. (More on this here and here.)
An Act of Murder handles this by giving the player a notebook of evidence and then having Duffy interrogate the protagonist in the end-game: the idea is that you have to have the right evidence present in order to offer answers to pertinent questions. The Phoenix Wright series does something mostly similar, except that it’s about deconstructing other people’s statements and alibis rather than building up your own: if you are interrogating someone on the witness stand and believe that they’re lying, you can interrupt them and produce evidence supporting your challenge. Though I haven’t played it yet, I gather L.A. Noire uses a similar mechanism.
When it works right, this system rewards flashes of insight; but it has a couple of drawbacks. First, it’s hard for the player to know whether he is lacking a key piece of evidence and therefore needs to explore more, or whether he’s just not making the right logical leap. When stuck, the player of this type of mechanic tends to have to revisit a lot of territory to re-detect, or else try every piece of evidence he has against every possible lying alibi — a tedious combinatorial exercise that makes for player rage.
The second drawback is that it doesn’t give the player the sensation of constructing something. All protagonist action is negative and reactive. You can’t argue X until someone else has said Not X and given you a chance to disprove it (or in the case of Act of Murder, asked a leading question).
Chunky Blues takes on this problem with a different game mechanic. The player has pieces of information in memory and can “chunk” those pieces together to produce new hypotheses and conclusions. Those hypotheses could be right or not.
There are several things about this I like a lot.
(1) It’s a coherent, consistent puzzle mechanic. We need more of this in IF. Coherent, consistent puzzle mechanics are more likely to be novel, fair, and well-implemented; they can be taught to the player and then used with interesting variations and twists; they work well within classic level design principles; they can support themes that emerge from the procedural experience. Moreover, a tight verb set and consistent behavior mean that the player learns the affordances faster, making the parser less of a stumbling block. (It’s the difference between having three key verbs you use inventively over and over, and having a whole bunch of loose puzzles where every other room you have to suss out a new verb for prying something open or blackmailing someone with something or filling an elephant-shaped balloon with helium.)
(2) It is expressive. It allows the player the pleasure of actively assembling things, and lets the protagonist be a more active participant in the story.
(3) While it doesn’t quite get rid of the “am I missing evidence or do I just not know how to use the evidence I have?” problem mentioned above, it makes that problem less punitively tedious because there isn’t a timing component as well; you don’t always have to sit through someone else’s conversation waiting for exactly the right turn to bring up counter-evidence.
(4) It lets the author control when certain conversation angles become available — because we explicitly know what the protagonist is thinking of and therefore which leads he might be interested in pursuing. That’s really useful, especially for a menu-driven conversation system like this one; otherwise, there’s a strong likelihood that entries on the menu will give things away to the player that he hasn’t figured out yet.
(5) It’s maybe a little off-putting that the system uses terminology like “chunking” that doesn’t feel particularly connected to the way most people think about their daily activities — or the way this particular protagonist would think about his daily activities. But it does make use of a consistent metaphor: the mind as a repository of building blocks with only moderate amounts of storage. The limited memory aspect means that the player gets actively reminded when he’s likely to have enough information to chunk (because his short term memory is getting full). It also means that the player’s actual, real memory is less likely to be stressed by having to retain too many small bits of data, because there are only going to be so many things available at one time. The advantages of the metaphor seem strong enough to justify its inclusion, even if it’s a way of thinking about consciousness that doesn’t particularly suit the period or the protagonist.
So. Neat concept.
Execution is a little rockier. Spoilers ahoy, so I can explain what I thought didn’t work so well:
There were a few points where the conclusion was given away in advance — for instance, even before I’d figured out that the note was from Velvita, the description strongly suggested it was. At other points, because the system has such expressive potential, it’s all the more disappointing when attempts to chunk two pieces of information together don’t form any interesting result. There were several times where I had a genuine idea about how two bits of information might be related, but the game didn’t allow for them — a sharp reminder that the expressive potential of the system is very limited in practice.
Another problem is that it’s not always clear how the player’s discoveries relate to the triggers that allow the plot to move forward. I wasn’t able to move out of the pile of garbage until I’d made a certain realization, and this was loosely hinted at by some of the game text, but in ways that only made sense to me after the fact. Similarly, there was a point at which I was suddenly inexplicably able to get past the bouncer, but I have no idea why, other than perhaps because I was carrying a hamburger. But maybe it was because I’d made some other discovery? I don’t know. The game just wasn’t clear about this.
The narrative triggers problem could be resolved with more rigorous plotting and better signaling — e.g., by having a lot of the game’s blocking puzzles depend on your ability to talk your way past someone contingent on what you’ve realized, rather than more abstract blocks like “you don’t feel like moving yet [and won’t until you’ve realized something, but I can’t tell you what it is]”. That would need some design changes to fix, but needn’t be a killer.
It’s much more problematic that this system invites the player to combine ideas inventively and imaginatively but has no way of coping when he invents something the author hasn’t specifically anticipated. I’m not sure exactly how I’d handle that. If the number of possible combinations over the whole game were small enough, it would be possible to write some entertaining easter egg text for what happens when you, e.g., think about elephant tusks and alphabet soup — something that made it clear to the player that this wasn’t an important step to have made, but that you had some funny, juicy comeback to the attempt.
If there are too many inputs to make that work (and for a large mystery there almost certainly will be many many pieces of evidence), then perhaps the next fallback would be to have some general categories of feedback; e.g., a standard message for any time you tried to relate the elephant tusk fact with something that mentioned Mr. Elephante. These sorts of feedback wouldn’t really advance the player’s progress, but they might implicitly hint a bit about which leads and connections are likely to be meaningful and which aren’t — thus contributing to the learning process, and still being more juicy than a default message that just says you can’t see how to link those two ideas together. The challenge then would be to figure out how to categorize types of information in such a way that you could generate good default feedback about the player’s combination choices.
There’s one other point that affected my enjoyment, not related to the puzzle mechanic. The game doesn’t hold itself to a very consistent standard for writing and atmosphere. The intro notes provide a disclaimer for this, in effect saying not to expect something literary. But I’ve enjoyed lots of games that weren’t literary. What’s missing here is rather a clear sense of what kind of writing it wants to be, a confidence about the narrative voice. Because it veers uncertainly between modern and period tone, serious issues and pop cultural jokes, the writing constantly makes the reader readjust his own mental relationship to what’s going on. This is apparent throughout, though especially strong in the hobo scene. Are we supposed to feel bad for the hobo, or are we supposed to want to toy with him in a cruel fashion? Is the problem of poverty and homelessness being invoked here seriously or is it a joke? Is his bindle full of girlie magazines meant as a serious stroke of characterization about what this man prizes? Or is it intended as a mocking gotcha to the player who has gone to all this effort for a bag of useless junk? It’s not clear. And because of this I’m constantly being forced out of engagement with the story.
All told, then, I was more interested in the gameplay mechanic than I was in the story and setting. That said, I’m really happy to see people doing this kind of work with puzzle mechanics. With some revisions and a stronger narrative voice, the full game could be very good.