Do learning analytics have a place in serious games? Learning analytics is “the measurement, collection, analysis and reporting of data about learners and their contexts, for purposes of understanding and optimising learning and the environments in which it occurs” (https://tekri.athabascau.ca/analytics/). Don’t games already have built-in game mechanics that do this? Yes and no. Performance is tested in-game. Game designers think through how a task will be learned and accomplished, and try to optimize this through testing. So if a learner in a serious game gets through the game, they should have learned what it takes to get through the game, meaning that they learned what the game intended them to learn (as long as the game designers were teaching and testing what they intended to teach and test).
However, this data (about learners) is not always measured. What about a task that most learners have trouble with? Game designers (counterintuitively) do not always know the most optimal way to get through their own games (or to learn their own games), even after sitting and observing representative testers. Expert players may, in the end, come up with more optimal solutions than what the original designers thought of in the first place. Without measuring this, collecting data about it, analyzing, and reporting this data about learners and contexts, it is difficult to know or share knowledge about such learner optimizations (or oppositely, sub-optimizations). This does not just happen in-game. Designers can be assisted in creating better games if they are better informed about optimal and sub-optimal performances in games. And game data can help bring this knowledge to designers. Learning analytics seem to have the potential to play an important role in serious game development, getting the designer to the understandings they need to frame the world that the learner needs.
Bob's Noggin Blog 
The designer needs to understand the optimal solution or skill required to play the game and know how to convey that effectively to the player. But how do you quantify the optimal solution though?
I think a quantification of the optimal solution is not always as necessary as the designer understanding qualitatively what the player needs to do/be at the end of the game. But there are several ways to quantify in-game tasks (for instance, setting weights on importance of specific activities, and visualizing/simulating possible optimal solutions to a task or series of tasks based on time, or on skills learned, or on complexity, etc.). A lot of these quantifications can rely on log data, so if a game you are designing logs data on users, you can analyze that. I’m working on a learning analytics quantification project with a colleague of mine presently that uses activity theory and clusters learners by results of the analysis in order to hopefully get at optimization (let me know if you would like to see some of our early exploratory work).
Sure! Would definitely be interested in seeing that.
I had given some thought to developers collecting user data on player habits and I’ve seen some of the stuff that FPSs use around the heat maps in player paths/deaths and things like that, but I’d be interested to see what else can be done with game data.
Actually, here is an even better example than what I am currently working on: http://www.instituteofplay.org/awsm/playtime-online/first-look-digging-into-data-with-simcityedu/
These workshop videos might also be useful: http://create.nyu.edu/?page_id=966
Here are also some excellent resources on getting started with game analytics:
http://blog.gameanalytics.com/blog/collections?collection_name=30
http://andersdrachen.wordpress.com/learn-about-game-analytics/
Really cool stuff thanks!