Now that we have covered basic techniques for recommender systems.
Choose one commercial recommender and describe how you think it works (content-based, collaborative filtering, etc).
Does the technique deliver a good experience or are the recommendations off-target?

I’ve chosen the Steam recommender system; a hybrid recommender system.
The statistics steam collects on you are :

Which users are your friends

Your owned items

Your item ratings

The time you spend with each item

The amount of money you spend

While I can not pin point what exactly steam does to recommend new items to its users; I can imagine how they use each metric.

The friends list you build probably weighs in on their collaborative filtering process; where all similarity measures regarding friends have a flat boost as opposed to users in your same neighborhood.

Owned items definitely is used for a item based recommendation where term frequency comes into play with the attributed item tags.

Time spent with an item carries a weight to how much you really “liked” an item; could possibly replace item ratings because this data is dense; unlike ratings.

Amount of money you spend on items definitely helps the recommender system select items in your price range.

Item ratings is probably used in a user-item matrix to help finalize whatever results the above metrics gave us.

I found a neat article on someone who tried to improve the steam recommender system : http://dataconomy.com/2014/08/improving-steam-recommendations/

He used a main metric of time spent with an item; then filtered his results with Pearsons R correlation; and logarithmic likelihood ratios.