Six things we have learnt about developing branching interactions.
This is the last in a series of posts that came from the talk I gave at the Serious Games BarCamp at AFTRS in Sydney.
The others in the series are:
- Why Branching Games (This is a good starting point if you are not sure what a branching interaction is)
- How we have been going about developing branching interactions for STE
This is really my first attempt to distil some of what I've learnt about branching games with STEM over the last year. It could also be applied to ARED decision trees and other decision tree development systems.
The images are links to the examples I talk about.
1) Reality is boring
I talked Janet Fearns from The Tasmania Skills Institute into doing a STEM interaction for the responsible service of alcohol course, because I thought it would make a great application of branching interactions. Walking into the planning workshop with Janet and a couple experienced teachers, I expected to be making an interaction about choosing whether or not to serve someone that might be drunk. Instead, the teachers made it clear "it's not that simple"; it's all about management of the relationship over the whole evening. We ended up planning a scenario where for the first three interactions it was just serving drinks and watching how much someone is consuming. The result was realistic but pretty boring.
What we found was the moment of conflict and critical decision making is most engaging. After the workshop we reworked the interaction to focus just on the critical decision point of whether or not to serve someone.
2) Simple media is often OK
I do a lot video work and Janet was keen with the responsible services of alcohol interaction to see it being done with video, so the learner could pick up nonverbal cues. In reality, shooting branching video and frame matching, while possible (and I did a lot of it during my Ph.D.) is hard. Also, the reality of bandwidth makes it hard to work with video reliably and without preloading. Video does give the rich sensory experience that we associate with games and digital medium, but simple images and text are easy to work with for most clients and keep the time, effort and costs down. In the case of the responsible service of alcohol project, we ended up just using a still image and text. Sound can have great emotional impact and there are bandwidth-friendly possibilities to be explored with sound.
3) Focus on where there is ambiguity between what a novice and expert might choose to do
I've found the greatest possibilities for rich interactions are found by focusing on the moments where an expert makes a different choice from what a novice would make. These are the moments and types of choices where the decision is not straightforward, and there might not be a right and wrong answer. A great example is at the start of the conversation about a marketing scheme in the electrical contracting course.
The expert's responses would be "Tell me more about the ad"; the other responses are OK in some ways, but they are not the expert response. One of the great things we have been able to do with the scoring system in STEM is to provide each response a number of factors, which means that for some answers the learner gets a high score in some areas and a low score in others. With the marketing example, the "Tell me more about the ad" choice gets the highest overall score on all factors.
4) Make feedback part of the conversation not an added extra
With electrical contractor customers there is a simulation where in a conversation the learner has to make a choice that makes the clients happy at this stage but is technically wrong. The electrical distribution company hated the idea that the learner was able make a choice that was wrong. When they saw it their reaction was, "Oh, that has to change." They didn't like the idea a learner could make mistakes. They want a form of feedback, maybe a big red message saying this is wrong. STEM can do that, but in reality when we make those sorts of mistakes in conversation, a big red box doesn't appear. A better way of doing this is like what happened in one of the LifeLine simulations where, when the learner repeatedly makes a bad choice, the client actually says, "I don't think you are listening to me."
5) Loop points are good
What do I mean by loop points? These are spots in the interaction where it could be good to return to, from where a series of themes can be explored and where the branches naturally fit together.
In the business compliance interaction for the electrical contractors course, a "To-Do" list was used a number of times for this. Sometimes in 3D games there might be a room with multiple doors that the player keeps coming back to.
Although working this way makes the interactions easier to author, it worries me that the "conversation never ends" and I'm not sure if the learners think it's boring or find it confusing.
6) Situation it in details
With the business compliance interaction, it was situated in an office and the problems were described in text. This is a classic "decision tree" style of working. It allowed a lot to happen quickly, but it just doesn't quite feel right.
My personal preference about what to consider.
- Where is the experience happening and make the media reflect that.
- What are the events that are going to happen in the space.
In many ways, what has been somewhat easy about many of the STEM interactions that I've done so far are that they are conversations. The locations are really simple and the events are really just the natural flow of a conversation.
For the business compliance interaction to work really well, the images need to be changing all the time, which would have meant a location-based shoot, props and actors, if it was being done photographically or maybe it all could have all been done as 3D renders.
In the near future I hope to put together a full guide and tutorial about how to develop a branching interaction. If you have been making these sorts of branching stories and decision trees, it would be great to hear what your tips are in the comments.
