Tuesday, January 3, 2012

DEV DIARY: Students at last!

I've made substantial progress on the quiz game, and I am now finally at the point where I've worked out many of the most crippling kinks in a variety of the game's basic functions: I've managed to create and load the game's answer database, track data and options selected by the player, compare data entered by the players with the answer database, I've implemented a countdown timer, and I've created a crude visual display so that players can see the answers they've selected. Now, I am at last to the point where I have started working on what I envisioned as being the central hook of the game: building the AI students and structuring the player's interactions with these students. From a coding perspective, this is going reasonably well at this point. But from the designer's perspective, I'm beginning to worry that these students are not currently adding fun and value to the game in quite the way that I'd hoped.  So it seems like now is the time to pause in the game-building process and do a little bit of a prototype postmortem.... 

Types of Interactions

As I mentioned previously, the goal of this game was to take the basic quiz-game and try to make the process of rote learning by question-and-answer more engaging by building in some element of social interaction and reward. In this particular game, "social" still meant interacting with an AI rather than with other real-life players, but I wanted to create some interactive students that would respond to player inputs, and would ultimately be the ones graded for the player's overall performance. While I remain convinced that there's something to this idea, and that on some level this kind of interaction between a player and an artificial AI has been used successfully in many very enjoyable games, right now I'm having difficulty implementing this in a fun and valuable way in the context of this quiz game. Below I've listed the different types of interactions I envisioned, how I've currently implemented them, and how well/poorly it seems to be working.

Students would provide immediate feedback regarding the accuracy of player choices.

Right now, the "quiz" is set up such that players are given a list of roughly 4 to 6 possible answers to a question, only one of which is correct.  Players must then pick the answer they think is correct, and then they have a brief window of time in which they may change their answer before given the final confirmation for their choice. When they initially select their answer, students respond by changing facial expressions.  If the answer is correct, some number of students will smile or show some other kind of positive expression.  If they answer is incorrect, some students will grimace or otherwise look confused.  This lets the player know that he/she may want to change that answer before giving it their final confirmation.

I see this interaction as one of the most important in the game.  The player needs some immediate feedback about his/her answers, at least on the beginning levels where the player is likely still learning the material.  Without any system of feedback, I have a hard time imagining how the game would be anything but frustrating.  However, I'm finding that because I've also built a time-limit into the game, the incentive to correct incorrect answers is less than I had initially expected.  The students' reactions to an answer is almost unwanted in this context.  It creates a sense of panic and frustration in the player, and paying attention to student feedback seems too cumbersome to be worthwhile.

Possible solutions to this problem?  I don't think that it's possible to adjust the student behavior other than to eliminate it entirely.  Right now, their feedback is as instantaneous as it can be without appearing overly artificial (I found that when students all reacted with the same face at the same instant that I made a selection, it didn't just feel strange and artificial, it was downright creepy!).  But rather than eliminate student reactions, I wonder if maybe the solution need to be the removal of the countdown timer.  I thought that the game needed a timer to provide a useful index of the player's knowledgeableness about the material.  The players who knew the material best I assumed would not only know the correct answers, but they would be able to identify them quickly. However, the timer adds an unwanted sense of freneticness to the game that seems to detract from actually interacting with the students.  So I wonder now if the timer needs to go, or at least be hidden from the player such that quicker average response times result in a better overall score up to a point, but players don't feel like they are slaves to the timer, rather than to the students.

Student's level of engagement with the player would change based on how well the player was doing overall.

Even though the visual feedback above seems an important way to provide players with feedback as they play, I did not want the students to be little more than graphical representations of the player's "score."  Ideally I was hoping to create some mimicry of what actually occurs in a classroom, and thereby make the students seem like they actually have some individual personality.  So I wanted a feedback mechanism based on the assumption that instructors are more effective if they can distribute their attention to all their students, and the classroom will have a sense of morale that depends on the instructor's ability to maintain trust and control.  I was envisioning a feedback mechanism similar to what one finds in many timed-button-press music games: the more notes the player misses, the worse the game's audio track sounds.  In this case, the more wrong answers a player submits, the more restless the students become.  Not only would this hopefully make the students seem less two-dimensional, but it would also provide some means of on-the-fly difficulty adjustment.

However, this has been more challenging to implement than I expected.  From a technical point of view, I've found it difficult to find a good way to assess a player's overall performance, and also to decide how frequently it should be assessed. Gauging how often to assess player performance is also directly related to the feedback mechanism itself.  I wanted this mechanic to force players to directly interact with the students by clicking on them. Players who were struggling might have students start falling asleep in the class need to wake them, whereas players who were doing well might have to deal with keener students who wanted to ask a question.  But as with the difficulty with the using the students' facial expressions as a feedback, it currently feels like too much to track.  A player trying to answer questions and keep ahead of the clock doesn't have much time to watch students to see if they've tuned out or have questions.  Moreover, for players who are already struggling, having another variable to track, like waking sleeping students, just makes poor performance spiral into absolute disaster.  So instead of making the player feel accountable to students, the students come off as an obstacle that drags the player down. 

Student grades(scores) at the end of a level would depend on more than just whether or not players selected all the correct answers.  It would also depend on how attentive the instructor was to the students.

I wanted the student's grades in the game to reflect more than just a player's accuracy in answering questions, so I originally conceived of this rubric for measuring player performance as being based on the above mechanic of having students either ask questions or fall asleep in class.  If players chose to ignore these students, they would be penalized.  The longer a student was asleep, the poorer his/her performance would be, even if the player answered all questions correctly.  Similarly, if the student had a question, the time that student spent with his/her hand in the air was time that student was thinking about his/her question and not thinking about the answers being provided.  So again, the longer that players ignored these students, the worse the student would do in the final evaluation.

However, the problem here was much the same as the problems mentioned above: there was simply too much activity.  When players are focused on "beating the clock," the frequent interruptions of the students become overwhelming.  Tapping on students to wake them from their slumber is shockingly distracting from the task of answering questions correctly.  If the player really gets into a groove answering questions (and ideally the player wants to get into a groove in order to achieve the best time), stopping to attend to a distracted student instantly kills the player's flow.  This is decidedly a bad thing.

The question now is whether to scrap this mechanic altogether, which I am loathe to do because I feel like this is the most important mechanic of all in terms of making the player feel a connection and responsibility to each student.  Can this mechanic be saved by eliminating the visible timer?  Would changing the frequency with which students become distracted either by questions or with sleep help diminish the sense of panic that this particular call for interaction currently creates?

Saving the Theory

The point of getting a prototype working as soon as possible is to see if the game's basic mechanics make for a good game.  Even without polish, the prototype should seem fun, addictive, worth exploring.  And if it's not, a good designer needs to make adjustments or start over.  I admit I am worried about what I'm seeing in the current prototype.  The game feels overly complicated, to the point of being more frustrating than fun to play.  My first step in trying to remedy this will be to try eliminating the countdown timer. If the player feels less rushed in every answer, maybe he/she will have more meaningful interactions with the students, which was the original goal of the game. Will eliminating the clock make the game too easy for a player who already knows the material?  Possibly.  But since the emphasis here should be on learning new material anyway, maybe this isn't something I should be worried about.  And if eliminating the play-clock doesn't help... well, maybe it will be time to rethink some of my assumptions about the basic gameplay.  More updates to follow...

No comments:

Post a Comment