Social media have claimed the mindspace of a generation. This is impressive, but it has not proven easy to monetize, and the model may be nearing saturation. Many therefore believe that the name players are out for bigger game, grabbing chunks of search, personalization, and e-commerce generally. Exhibit A in this case is Facebook Like.
The concept is simple and plausible. Users are already linked to their friends. Scatter “Like buttons” around the digital universe to collect preferences from users. If many of a user’s friends like something, the user should like it too. After all, friends recommend things to each other all the time. Facebook is on its way to a virally generated preference engine and maybe the beginning of a challenge to Google, right?
Yeah . . ., not so much. The concept suffers from three fundamental flaws.
First, the strength of the self-selected friends group is its relatively small size—it is restricted to a people with whom the user has some genuine affinity. But this small size is also a weakness. Small groups of people experience very little of the universe, digital or otherwise. Think of anything from books to travel destinations to Boston restaurants to information about global warming; no one has personal knowledge of more than a tiny fraction of one percent of the possibilities. Any system based on a selected affinity group will therefore have a very restricted reach. What the user would really like may never have been sampled by any of his or her friends. Even for items with more “data density”, like recent movies, the size of the friends group is a liability. “N” will necessarily be a small number, and small n means low reliability.
Which leads to the second shortcoming. Friends do tend to like the same things, but it is just that, a tendency, and not an especially strong one. Take Chinese food. Because I like it, how likely is it that my friends do also? A little, but not much. This problem only gets worse as the item becomes more specific. Hunan cuisine? A Hunan restaurant with family-style service, lots of Chinese patrons, and Chinese beer? The affinities of taste among friends at this level become vanishingly small. The same holds true of informational items. If I care about global warming, do my friends? Maybe; people of similar outlooks are more likely to be friendly. But if I am especially interested in stratospheric particle shields, are they? Of course, recommendations and marketing decisions get made all the time based on age, gender, education level, geography, ethnic background, and so on, where the affinities may be equally weak. The difference is that those weak affinities are coupled with massive groups. A large enough n may compensate for the predictive weakness of any one affinity. Facebook Like will, in many areas, be stuck with a weak affinity and small n—not a happy combination.
The third flaw is so basic as to be easily overlooked. Where are the “Dislike” buttons? Dislikes provide as much information as likes, and often more. I tend to like films with quality acting and direction. Here are a few that fit that description: Saving Private Ryan, Schindler’s List, Goodfellas, and Casino. Unfortunately, I haven’t the stomach for graphic violence. My friends rave about all of those films, but they never recommend them to me, because they know of my peculiar distaste for screen violence. Facebook Like has no way to capture their knowledge.
Perhaps “Like” will somehow enrich the social medium of Facebook and help prevent erosion of the user base when the new, new thing comes along. If Facebook was expecting much more than that, it will be disappointed. Unfortunately for them, a simple thumbs up or even the addition of a thumbs down, will be terribly inaccurate as a predictor and will not drastically change the way people use the web.
About the Author
Owen Paepke is an attorney in Phoenix. He has practiced extensively in the field of intellectual property and as a General Counsel. He has also served on advisory boards of several companies in the Internet and software fields and on the board of directors for a pharmaceutical R&D venture. Mr. Paepke is the author of The Evolution of Progress (Random House 1993), which discusses the changing nature of technological and economic progress. Owen has authored several articles on intellectual property matters, and is a frequent speaker and writer on technology and its impact on social and public policy issues. Mr. Paepke is the inventor of the patented recommendation technology know as Affinity Analysis.