I generally like recommendations. They require some existing data and an understanding of the data to make the recommendation. My friends often suggest articles and direct me to links that I usually like and enjoy. The reason their recommendations are “successful” is because they have some preexisting knowledge about my interests and can infer what I’d like based on that data. Fortunately for me, my friends are really good at this. So, you could imagine my excitement when the NYTimes introduced their Recommendations page for readers who have an account on the site, which combines a list of recommended articles and aggregated data of previously read stories. It’s pretty good, though it doesn’t include articles read on other devices (iPhone/iPad). The problem with data collection and the recommendations borne out of them is that once we become aware of the collecting, we alter our behavior to accomodate a preferred recommendation. That’s not to say that that’s a bad thing. We use data to gauge and manipulate our behavior accordingly all the time.
Also, I say “generally like” because, well, look at those topics. Embarrassed.
If you’re curious, Derek Gottfrid talks a bit more about this project on his site.