In general, friction is an artifact that we try to remove. Friction is a literal drag on mechanical devices. Outside the physical world, businesses focus their efforts on creating frictionless environments. In many ways, the power of the Internet is about removing or at least diminishing the economics of scarcity - limited television channels and limited shelf space are all forms of friction. Giants of the Internet economy, like Amazon and eBay, owe their success to their ability side-step the barriers inherent to physical retail stores. Likewise, Brightcove believes that the alternative distribution available through Internet Television will create an explosive new market that is currently limited by the costs and restrictions built into existing broadcast, cable and satellite networks. Friction has a bad rap.
However, as I was reminded while scaling Mt. Liberty yesterday in New Hampshire's still winter-bound White Mountains, frictionless systems (in my case ice) are not always a good thing. Luckily my crampons did a fine job of keeping me secure! In all seriousness, the lack of friction built into many online processes and services can cause tremendous headaches. Take for example spam - some proposals attempt to solve this problem by introducing various forms of friction (micro-payments per message; extra computation, etc) that were not initially designed into the email delivery protocols.
eBay leverages friction by limiting the number of category placements and charging extra listing fees to keep value in their topic tree. A lack of friction in category assignment would enable sellers to place their items in a near infinite set of categories, driving down the value of all classifications - a tragedy of the commons. Recently eBay increased their listing fees for particularly popular features. On the surface this could be seen as an attempt to just boost the bottom line. But, as I recall from an NPR interview, this increased friction is actually an attempt to shape the dynamics of their marketplace. (i.e. The value of highlighting an item will increase, because less sellers will pay the extra money to highlight their items. Or so the theory goes.) In a similar, but rough fashion, the Federal Reserve Board uses interest rates (a form of friction within monetary policy) as a blunt tool in shaping the U.S.'s economy.
In most circumstances the problems one can face with frictionless online environments are not visible until one reaches scale. The lack of friction in defining HTML meta-tags are a key reason why Google and other search engines ignore these potentially useful constructs. Problems with "tag spam" are starting to be encountered with social tagging systems like Flickr. Even if we could miraculously ignore subversion, the open-ended model of current tag based systems could break down on their own weight. David Weinberger highlights some of these issues in his excellent Release 1.0 article titled "Taxonomies to Tags: From Trees to Piles of Leaves."
So when product managers think of new services or developers design new features it is important to think about the usefulness of friction to create a viable and sustainable system. For new ventures, the interesting challenge is to design a robust model that begins with little friction and achieves rapid growth, but can easily evolve and introduce appropriate friction as usage scales and the probability of abuse increases.