Ugh. I never liked that Ars Technica article. The stats therein aren't nearly as useful as one might think because of the overly broad or limited criteria with which they used to gather their data.
36% of all owned Steam games go unplayed? Okay, and how many of those games remained unplayed after they were accounted for during their research? How many significant sales did their research timeline coincide with? How many games were played prior to the implementation of the Steam game-time stat tracking system? How many were played offline, leaving their stats untracked? How many were cheap or free games added to smurf/scammer accounts? Why did they look at over 170 million accounts when there are only around 100 million active accounts? (millions less at the time of their research)
It's an interesting statistic, but it leaves more questions than answers. And granted, Ars admits to some of these flaws in the article, but that doesn't really negate their existence within the methodology.
Anyway, this SteamLeft seems like an interesting site. However, given the rash of recent scamming, phishing, hacking, and hijacking attempts on Steam users, I'm hesitant to give out public data on my Steam profile. This is why I still keep it set to private.