or, The Hitchhiker’s Guide to Fear and Loathing at a Public Library Reference Desk

“Readers Advisory” on Hulu

   September 9th, 2010

I was watching a show called The Book Group on Hulu recently, and got a taste of how they recommend other shows to people.

The bottom of every show page always has a "You Might Also Like" section, recommending similar shows, which I have used that in the past. But because a couple of the episodes of The Book Group were rated TV-MA, and required me to log in, during one of the commercial breaks I got this ad:

Hulu ad

Which I read as,

Brian, not only are we violating your privacy, but we also think you have bad taste.

I'm sure the "27x more fans" thing is just to induce me to watch the other show (Peep Show, which I did watch a few episodes of and didn't really like). However, requiring me to log in and then using that to track me and "personalize" suggestions does feel like a violation. A different ad seemed more reasonable:

Hulu Ad2

This conveys the exact same message, but doesn't also imply a deficiency on my part. So, I guess a word of caution to anyone providing readers advisory or viewing suggestions on your website - careful how you word the message.

Also, this got me thinking about two types of suggestions: item-oriented suggestions and person-oriented suggestions. Item-oriented is like NoveList or LibraryThing for Libraries - basically, providing suggestions based on the characteristics of an item.

Person-oriented suggestion is more like a personal shopper, or saying, "based on our monitoring of your behavior, we think you'd like this" - providing suggestions based on the preferences (or past behavior) of a person (or people). Amazon's "Customers Who Bought This Item Also Bought" or "Frequently Bought Together" sections are like this, as well as their "Recently Viewed Items." Which isn't a bad thing, unless the person being monitored don't know about it, or has no choice about it.

Hulu might be genericizing the data of what other people are doing, but it seems like they're still tracking what individual people do on their website, and I will always feel uncomfortable with that.

Tags: , , , , , , , ,

2 Responses to ““Readers Advisory” on Hulu”

  1. Andromeda Says:

    Huh, I read that totally differently than you did, I think. I’m inferring that you read it as “fans of The Book Group are 27x more likely to watch This Other Show than they are to watch The Book Group”? But I instinctively read it as “fans of The Book Group are 27X more likely than average Hulu watchers to also watch This Other Show”.

    (If it has the latter meaning, also, then the “84% also watch…” idea does *not* mean the same thing. It conveys the same social meaning of “if you liked this, you might also like that”, but it is referencing a completely different, and less useful, mathematical model.)

    I suppose I will be offended as well, but I will be offended by Hulu using mathematical language in imprecise ways. 😉 Now I’m curious what their recommendation algorithm is, though.

  2. Brian Herzog Says:

    @Andromeda: yes, I read it the first way, but your way is equally-plausible. I’m just overly-sensitive in general, so anything that came after the “Brian” I was predisposed to find fault with. But I am curious how they (and anyone) calculate recommendations. I get the tag-based LTfL model, and Amazon’s “people who bought this also bought” thing, but videos seem different. However, it could just all be based on activity.