Home » Blog » ideas » Algorithms, Unbirthdays and Rewiring Facebook

Algorithms, Unbirthdays and Rewiring Facebook

Tuesday was my birthday, and I spent the day largely offline. That meant that Wednesday morning, my email inbox featured hundreds of messages from Facebook, each alerting me to a birthday greeting on my Wall. (I’m an infrequent Facebook user, so I usually find these sorts of alerts useful and haven’t disabled them.) On the one hand, this outpouring of online affection was wonderful – I felt grateful to be remembered by people I’ve not spoken to since high school.

On the other hand, it’s basically impossible to respond to the flood of messages with anything other than “Thanks!” And, of course, there’s usually nothing to the message than the greeting itself – the message is symbolic, not substantive. Which left me thinking

– I should be better about logging onto Facebook and sending my own symbolic, semantically void greetings

and then

– I should write a Facebook ap that partitions my friends into 365 roughly equally sized groups and encourages me to say hi to that specific, small set of people on that day. I’d occasionally reach someone on their birthday (though I could add additional logic to pick only unbirthday folks.) Unbirthday notes would arrive on days when people weren’t overwhelmed, and might actually spark a conversation and a chance to catch up.

Socially transgressive, or a helpful hack for building actual conversations between out of touch friends? Would other people resist such a rewiring of Facebook and the social norms it embodies, or embrace it?

One of the reasons I don’t use Facebook often is that it seems to be wired to persuade me to behave in ways that I don’t find especially productive. It’s great that I can catch up with most of my high school friends via Facebook, and I’m glad to have the opportunity for a glimpse into their lives… but in many cases, these are folks I’d love to check in with once a year or so, not every day. Facebook is utterly brilliant in finding people I used to know, from elementary school classmates to ex-girlfriends. I suspect if I used it better, it would do an excellent job of helping me maintain closer ties with these friends, turning weak ties into stronger ones. What I’ve not found a good way to do is to use Facebook to discover people I don’t know and would like to (something that happens to me all the time through retweets on Twitter).

Are there ways to rewire Facebook to try to create a specific sort of serendipity – discovery of people, places and things outside of your ordinary orbit, but exciting and interesting nevertheless? What would an algorithm look like, and does Facebook expose enough data to make it possible to build such a tool?

In 1994, Pattie Maes started working on RINGO, a music recommendation system, at MIT’s Media Lab. The logic behind it was deceptively simple – rate your fondess for twenty musicians or bands, and RINGO would start to suggest music you might like. Behind the scenes, RINGO used a collaborative filtering (CF) algorithm which determined which other users of the system had liked many of the same artists, concluded that you might have similar tastes, and recommended other bands that user had liked. The RINGO system became Agents Inc. and later Firefly and was purchased by Microsoft, where aspects of the system became the basis for Microsoft Passport.

Collaborative filtering algorithms can produce very impressive results, especially if they’re used in a well-defined realm (music, movies, etc.) with sufficient information to extrapolate from. Amazon and other online merchants use them to offer recommendations, which – though occasionally bizarre – are often quite helpful.

The limitation of this algorithm, for my purposes, is that it’s based on looking for users who are similar and then offering their preferences. It’s easy to imagine this running into limitations over time – if I list only 1980s hair metal as my 20 ur-bands, a CF algorithm will rapidly close the set, finding other folks who like hair metal and listing their favorite bands, including the truly obscure ones. It’s possible that this method will also find me the best Qawwali music if someone happens to like both Whitesnake and Nusrat Fateh Ali Khan, but it’s probably not the fastest way to discover something unexpected and excellent from outside my set of existing interests.

(I may be entirely wrong about that last sentence, by the way. One of the most intriguing conclusions Duncan Watts comes to in Six Degrees is that small world networks – networks in which it is possible to find a connection between two unrelated people through a small number of links – are possible because while two friends have lots of friends in common, they also are likely to have a small number of friends not in common. In other words, if you and I both like the Beatles and the Band, we’re both likely to like the Rolling Stones, but it’s just possible that I might like the Dresden Dolls and you like Fourtet… which might mean we can get from Whitesnake to Nusrat Fateh Ali Khan if we follow the right links in our network…)

My old friend Dave Arnold (now a famous chef and blogger) became a massive Bob Marley fan in high school by deciding he wanted to know something about reggae – a style of music he’d never heard – so walked into the local music store and bought a CD by the artist who seemed to be most popular in the reggae section. (Trust me, if you were a sheltered kid in 1980s Westchester County, NY, it was possible to make it to age 15 without having heard of Bob Marley.)

It’s my guess that the Arnold/Marley algorithm isn’t a bad first pass at an algorithm to make serendipitous discoveries. Find a set of unfamiliar people, places or things and look for the most popular in the set. Often, there’s a good reason something is popular. Try that and see if you’re intrigued. If not, you could dig deeper into the set, though you’re probably best off trying the most popular from another set.

How does this help us with Facebook and serendipity?

Facebook is all about popularity. It’s high school with less acne. (That may also not be true – I have no working online acne-detection algorithms.) You can track who’s got the most friends, what applications have the most users, what groups have the most members. All throughout the web, you have the opportunity to improve someone’s score, clicking the insidious “Like!” button.

However, Facebook doesn’t like revealing the scoreboard, except when there’s a business case to do so. In the past year, Facebook has evidently done a great job of encouraging certain types of advertisers – “Bands, Businesses, Restaurants, Brands and Celebrities” – to drive people to their Fan pages. Nick Burcher tracks these sorts of things, and he’s got lists from May 2008 to October 2010. There’s a pretty fascinating shift from mid-2009 to mid-2010 – in 2009, people are fans of abstract things (sleep, vacations, the beach, pizza) and by the next year, brands have won out (YouTube, Lady Gaga, Family Guy, Coca Cola.)

Before I found Burcher’s lists, I started making my own from Facebook’s directory of pages. Each letter includes a list of top twenty pages with the corresponding number of fans. So A is for AKON, AC/DC, Alicia Keys and Aston Kutcher, B is for Bob Marley, Beyoncé and so on. I found only only 20 pages listed in the directory with more than 10 million fans (a totally arbitrary cut-off) – my guess is that the directory’s top 20 isn’t comprehensive as Barack Obama (with 17 million fans) doesn’t make the B or O directory pages, while Michelle Obama, with 3.3 million fans, is #12 on the M page.

My top 20 list, or Burcher’s top 30, isn’t going to surprise you – the most unfamiliar figure on the list to US audiences would be Real Madrid star Cristiano Ronaldo, otherwise lost in a sea of pop stars and cartoons. But the folks listed lower down in the top 20 lists on each directory page get pretty interesting.

A.R. Rahman, India’s famous film composer, sneaks onto the A page with 2.6 million fans, and the music streamable from his page (including international hit “Jai Ho”, the theme song from Slumdog Millionaire) isn’t a bad introduction to Indian film music, including the current, unfortunate embrace of autotune. Reggaeton star Daddy Yankee, Turkish balladeer Emre Aydin and Colombian rocker Juanes all boast between one and five million fans, suggesting that while they’re not international megastars, they’ve got huge followings in their regions, and might serve as excellent introductions to what’s going on in that corner of the musical world.

Of course, there’s other places where the Arnold/Marley algorithm might find you an interesting cultural figure, but not offer much help in understanding his importance. Take Mario Teguh, who has 3.3 million fans on his Bahasa Indonesia language Facebook page. Not speaking a world of the language, I assumed he was an Indonesian politician. Wikipedia’s not much help either – while his Indonesian Facebook page is well-developed, he doesn’t merit a page in the English-language Wikipedia. Teguh, from what I’m able to gather online, reading his Facebook page in translation, is a motivational speaker, philosopher and business coach with a popular show on Jakarta’s MetroTV.

Discovering that Teguh is popular on Facebook tells me a few things. One, there are a lot of Indonesians on Facebook. (According to Burcher, there are now 32 million Indonesians on Facebook, vaulting the country past the UK for the #2 spot in terms of Facebook users. Internet World Stats projects only 30 million Indonesians online, which suggests either that someone’s got their numbers wrong, or that Facebook penetration in Indonesia is amazingly high. The probable explanation is that many Indonesians access Facebook via mobile phone and may not show up in IWS’s stats.) A second is that I’m probably not going to have much luck penetrating Indonesian culture online without a guide and translator. Music show Dahysat and comedy show Opera Van Java both boast a few million fans. The latter looks particularly fascinating:

From what I can tell, Opera Van Java is a comedy show that combines aspects of traditional Javanese puppet theatre, including accompaniment by gamelan musicians, with sketch comedy. If I were to venture a guess, I’d suggest that the actors might be playing the parts that traditionally would have been played by puppets, acting out famous stories… in this case, the story of Drunken Master, the legendary 1978 Jackie Chan film. What I desperately want is someone who’s knowledgeable about contemporary Indonesian culture, traditional Javanese puppetry and fluent in English to walk me through this video and help me understand the popularity of this show. Here, my glimpse of Indonesia through Facebook shows me little more than there’s something potentially fascinating and thoroughly inaccessible without translation and bridging.

And that’s okay – my goal isn’t to solve the whole problem of encountering another culture through a single algorithm. I’ve been making the argument that using the internet to discover a wider view of the world involves some combination of translation, cultural bridging, and structured stumbling, and this Facebook directory trick suggests at least one way we could stumble in the direction of understanding what’s popular and compelling in another corner of the world.

The tricky part of using this algorithm is figuring out what’s popular where. Some years back, Amazon published lists of what books were most popular in particular geographic locations, based on an aggregation of people’s purchasing information. Choose a city like Barcelona or Cape Town and you might get a very different view on what literature might be interesting to explore. Unfortunately, they’ve disabled the feature (probably due to privacy concerns) and replaced it with “customer communities“, opt-in affiliations which may help you find serendipity within a topic, but reflect what’s popular in a community of interest, rather than a physical community elsewhere in the world. It’s worth noting that Amazon has this data – if someone could make the case that knowing what’s hot in Tokyo or Turin was exciting for a US or UK audience, and if there’s a way around privacy concerns, you could sort the Amazon catalog to enable this sort of serendipity. (Indeed, my friends working on ShelfLife at Harvard are asking questions like “Can we reveal what books are popular with philosophy grad students, without compromising their individual privacy? And is this useful information to help other people discover new books?”)

One of the few sites I know that offers good, regularly updated charts of what’s popular in different countries is Alexa, which tracks the top 100 most popular websites per country. Again, the top sites (which inevitably include Facebook and YouTube) aren’t as interesting as what’s in a second tier, like Taringa, a community bulletin board site popular in Latin America. If we were rewiring Alexa to help find what’s exciting in other parts of the world, we might do a diff between the list of what’s popular in our home region and in another country.

Of course, what I really want is to know what’s popular within massive sites like YouTube and Facebook which are visited by users around the world. What videos are most popular in Malaysia? Pete Warden offers a great visualization that shows us who (some) people in different nations are connected to on Facebook, what their first names are likely to be, and who they like – Nigeria, for instance, is particularly fascinating, in that the two most “liked” are charismatic preachers, not athletes or rap stars. Would we use tools like Facebook and YouTube differently, if the platforms themselves tried rewiring themselves for discovery and serendipity, rather than relying on outsiders guessing at popularity and connections from outside the walls?

10 thoughts on “Algorithms, Unbirthdays and Rewiring Facebook”

  1. Happy belated Birthday wishes, Ethan! I saw your bday anouncement on FB but didn’t react on it because of the reasons you mentioned. Because it would force you to reply with a “thanks”. Do we really want that?

    Otoh, your FB profile also has 1024 “friends” which is a bit different from the average user, eh?

    Serendipity, yes. It would be nice to have a useful algorithm that connects ppl based on their offers and needs, like what you mentioned “someone who explains these videos to me”.

    Are we ready for all these conversations? And if yes, when?

  2. You know, I don’t usually engage in FB interaction unless there’s a point. And, it has to be said, I don’t say happy birthday to everyone.

    And I probably wouldn’t bother at all were it not for one thing. That is, when I see an alert informing me its someone’s birthday I feel kinda guilty.

    Simply put, in these days of using Facebook, not saying Happy Birthday to someone I’d like to wish a happy birthday seems a bit… well, rude.

    So, I don’t do it for everyone, but there are just come people who can’t be ignored, and you’re one of them. Now, quick disclaimer time…

    Lest others I haven’t wished happy birthday to on Facebook, it could be that I missed the alert for one reason or another so please don’t get mad at me. :)

    Now, back to my own pet gripe. Loads of Happy Birthday wishes and welcomes on the GV mailing list which generate far too much traffic.

    As a result, I often miss stuff that I shouldn’t.

    And not least because we’re encouraged to be part of the GV community through the mailing list. Yet, those types of messages are more time consuming than FB.

    So, my opinion is simply that Facebook is a better medium to do this than, say, the GV mailing list. Of course, then I get guilty if I don’t do it there.

    Yes, yes, I know. I should seek counseling… :)

    P.S. Re. your comment about partitioning Facebook into groups. Basically, I think you’re going to love Diaspora and its Aspects.

    Oh, and can’t resist…

    Happy Birthday once more, Ethan. ;)

  3. And, of course, a simple solution is there on Facebook too. Remove your birthday from your info so the reminder doesn’t come up on our FB pages. Sorted.

    Meanwhile, think I might try to set up a filter for Gmail to filter out the birthday greetings into a separate folder on Gmail.

    This for me is the killer and really suffocates what’s going out. I’ll still look at them and respond, of course, if only to avoid any sense of guilt… ;)

    But my point is simply this. What happened is considered the socially right thing to do when people are alerted to someone’s birthday.

    It happens even with email and the GV mailing list, for example. And again, it can be avoided by simply choosing not to reveal your birth date.

    In fact, some security conscious people advise this anyway to avoid contributing to the risk of identity theft.

  4. Hi Ethan. My question is: supposing music to be as you suggest, a well-defined field but also one with deeply secretive subcultures — (and setting aside the issue of what Google or Facebook or Amazon actually use) — don’t we simply approach a point where the outlier or anomaly becomes something to be clicked through and explored?

    To whatever degree people type their preferences into Facebook, doesn’t Facebook have a certain advantage in this aggregation process? Suppose I steadfastly refuse to tell FB about my love for the Stiff Little Fingers or whatever, but blurty fellow fan-page fans of a big record store like Amoeba or what have you, which I also was willing to friend on FB, are willing to confide in FB, and see enough social value in confiding in FB, to type in the name of my super-secret favorite band.

    Doesn’t the issue then shift slightly, away from the algorithm that’s available to users, and more to the potential for patient data mining that FB can make available to customers?

    If so, I propose that this introduces a distinct class stratification that is likely to remain.

    Al Matthews
    Atlanta, GA

  5. Al, I think you’re right both in identifying the secretive subcultures in music, and in noting that one possibility in this sort of exploration is clicking on the anomaly or outlier. And you’re certainly right that Facebook has a huge advantage in having access to all this data – they’re very well positioned to identify the hot new bands being identified by a subgroup of particular fans, which is something they could then sell to that band, a record label, a rival band, etc. as advertising placement.

    The challenge I was trying to put forth in this piece was less about privacy and more about what it would mean to restructure Facebook to encourage serendipity. Within that framework, I think I’d likely be endorsing Facebook trying some creative – and potentially invasive – new ways of mashing data in order to try to build broader connections.

  6. Points taken, Onnik – I certainly could opt out of the whole birthday greeting nonsense entirely. Where I was trying to go with that example was to introduce the idea of rewiring. We’re wired to behave certain ways socially – wishing friends a happy birthday – and technology can both reinforce or fight those norms. Facebook is absurdly good and reinforcing some community norms (send birthday greetings!) and amazingly sinister in pushing on other norms (display your personal information to everyone!)

    My goal was to introduce this idea of rewiring through a (mostly) theoretical ap, then think about how I’d rewire more broadly on the serendipity front. For now, I think I’m going to keep my birthdate visible. Even if I scrub it from Facebook, I think it’s already on Wikipedia… :-)

  7. Facebook was very deliberately designed, at the start, to be the online complement to your real-world social life. Although it has strayed somewhat from that, especially in the direction of enabling interaction with organizations, companies, and products, it’s still primarily about providing online tools to go together with the connections you already have elsewhere. It’s never been about finding and making new connections, and this seemed at least at the beginning to be a very deliberate design decision. It probably still is.

  8. Pingback: Joho the Blog » Berkman Buzz

  9. I think that’s right, Cos – Facebook certainly began as a networking service for tight, real-world social networks. It’s moved a bit from that, but not very far. The trick is, it’s now such a powerful tool, it invites thought about how it might be used differently. My hope here was to suggest how we might rewire even within the existing structure, either as outside tinkerers or as interested insiders. But I agree with your characterization of the assumptions behind the network.

  10. Pingback: Kubatana.net speaks out from Zimbabwe » Blog Archive » Facebook musings

Comments are closed.