Software developers and instructional theorists have claimed for years that playing computer games is more "intrinsically motivating" than more pedantic or didactic learning activities, and that therefore marrying content to form might produce a superior (software) learning product. Speculative treatises like the Jones piece serve the purpose of stirring interest for a few readers; but then the problem of operationalizing a study enters the picture and the whole enterprise seems to collapse. Much of the interest is fueled by the Sesame Street mentality in which some believe that dressing knowledge up as shrill cartoons or portraying intellectuals as goofy (Bill Nye)--"bells and whistles"--are more entertaining--therefore engaging--than serious forms of presentation of subject matter that is assumed to lack intrinsic interest for some audiences.
Despite the apparent applicability of the "flow" construct to a wide spectrum of human activity, the problem of how to establish the presence of flow in any given circumstance is not a simple one. Flow studies rely on statistical analyses of very large sample sizes. In preparation for the data collecting phase of the research for my 1998 University of Washington Ph.D. dissertation I spent four years developing a method of measuring or at least detecting the presence of autotelism (flow) in computer game play--along with an "autonomy supportive" unique apparatus to implement the method. Unfortunately, collecting flow data during computer game play is difficult--and that difficulty is probably why few have gone beyond the speculative stage in considering such questions. But surely before reaching conclusions about the elements that might make a computer-based learning environment "flow-inductive," one would need to establish a way to test the effectiveness of each proposed element. To make the method that I worked out to measure flow during human-computer interaction widely available to researchers I would need a grant to support the conversion of the arcane technology involved to a universally available software platform. If you work in Windows and can allow 30 minutes for the download, you can get a sense of the method from a self-running Powerpoint presentation I've placed on my website at URL http://www.learningtech.com/pplink.html.
The only known method of detecting or measuring flow is through the administration of extensive, detailed questions that reflect the eight elements (more or less) that define the flow condition or state at random intervals during the course of the activity under study (see Jones's tables). Because of its presumed transient nature, flow is measured or detected in situ. Retrospective questionnaires, although I used them in addition to the mid-play variety, do not measure flow directly; they record one's impression whether one has experienced flow, and are therefore less reliable than the mid-stream variety, whereby the question-answer activity and the target activity are temporally juxtaposed. How would YOU solve the problem of requiring introspective answers from subjects to a couple dozen questions while in the throes of Doom, without breaking the "flow" state? I'd love to hear about your creative solutions.
A difficulty inherent in the flow construct is its lack of temporal specificity. Computer games may be played for very short intervals, frequently or infrequently, or for sustained periods lasting months. Flow is broad enough in scope that it can apply to a lifestyle. In my studies, no subject completed the target game in a single setting--nor even in a single week; subjects made several return visits to my "lab" over periods extending up to five weeks. Were they in flow the whole time? I doubt it. Then when were they, if at all? Well, fortunately, I did not rely entirely on questionnaires of the type employed by Csikszentmihalyi and colleagues. I also recorded on video tape everything subjects did onscreen, along with and in synch with their stream of consciousness TAs (think-aloud protocols). The latter proved a much richer type of data for understanding what really went on during the activity than the experience sampling questionnaires, virtually rendering the flow issue of trivial consequence.
In the end, the flow data did correlate with the learning data--but not for the reasons assumed by theorists and claimants pursuing the intrinsic motivation path. Those subjects reporting flow during the activity worked faster and learned more than those who did not report flow in the activity, but it is not clear whether the higher skill and therefore success shown by the higher flow subjects was the necessary condition for the presence of flow, or the presence of flow was the necessary condition for the higher success and skill evidenced by the higher flow subjects. Ultimately flow appeared to be of less interest than two much more important and powerful determinants of learning success in the game environment: (1) understanding what was necessary to succeed, and (2) willingness to do what was necessary to succeed. It turns out that the latter are the two most important factors determining success with ANY learning task, whether set within a "computer-based learning environment" or a traditional classroom.