I was recently part of a PhD thesis project that asked some folks to do a Delphi process about the educational use of MMORPGs. It was interesting, and of course thought-provoking, and now it‘s done I can talk about it.
Beyond the obvious benefits of a potentially motivating context for learners, and commitment by the learner to the extent they‘ve customized the experience, there were some deeper issues. However, there appeared to be assumptions that it had to be massively multiplayer, and that existing such games would be used, as opposed to designing ones with specific characteristics to work with a selected cohort of learners. So we can first talk about those assumptions, and then move beyond.
One obvious concern is that in an existing environment, there are no specific learning affordances other than the game mechanics (which may not have much social benefit: there are little benefits to beating up kobolds outside the game environment). Now, some of the game mechanics may have transfer, particularly social ones, c.f. the leadership skills purportedly developed in World of Warcraft, so there are reasons. And, of course, you can always talk about learning in such environments.
The flip side of the social environment is the possibility for inappropriate social activities that can happen in real life, e.g. bullying, but this is not unique to the online learning environment and merely needs the same approaches of education and monitoring that you‘d want in real life.
Now, if you can design characteristics of the environment, such as the ability to build things (e.g. as in Second Life, where you can do 3D modeling, but it‘s not a game), and you can create the context and task for learners, you can embed specific learning outcomes into the environment (you know that designed learning environments is what I‘m about).
Of course, I‘ve also mostly been about individual learning experiences, and have argued that unless you‘ve social learning objectives, there‘s not a principled reason to build social games. However, that is neglecting the benefits of collaboratively problem-solving (though it can be done by post=game reflection), which often has great learning benefits (e.g. social learning theory: Bandura, Vygotsky, etc).
One of the big themes that emerged that I hadn‘t really tweaked to but now embrace is that such environments may foster 21st century skills. Such environments naturally include communication and collaboration, and could easily be augmented.
And, of course, one of the challenges even if we could develop and deploy these is ensuring that mentors or teachers are capable of scaffolding the learning from these environments. That, I think, is a 21st century skill needed now amongst educators, and it still needs to be developed (and motivated and rewarded!).
It‘s pleasing to see these explorations, and here‘s hoping there‘s more.