This morning the rumor mill started running again, this time fostering speculation that Apple is in talks to buy Twitter. I won’t bother you with the details of the supposed transaction (*cough* $700M *cough*). The more interesting thing here is that, if the unlikely rumor is true, Apple would basically be buying a massive, seemingly ever-growing bag of information it will need to properly mine in order to effectively market and monetize — if it even wants to creep into the online search industry.
Why Twitter?
Why would Apple buy Twitter? Why would anyone buy Twitter? Well, among other things, it’s a searchable, real-time index of information about what is happening right now. And this information spans from news to corporate feedback (e.g., complaints/questions to companies like Comcast). And that’s powerful, especially as the service grows in popularity.
Mine Your Tweets
But don’t be fooled; Twitter’s not a “just add water” acquisition. Sure, it’s massively growing, and it has seeped into the mainstream seemingly overnight — thanks to the Oprah effect. But, as we’ve seen with the latest hysteria around Swine Flu, an incoming hoard of information tends to feel less like a stream and more like an avalanche. So making sense (or cents) off of this information requires a bit of automated, intuitive prioritizing — the type of prioritizing inherent in Flickr’s Interestingness algorithm, for instance, which bubbles up the site’s most “interesting” content on a timely basis.
When you want to find out what’s really happening right now in relation to, well, anything, you don’t want to waste time sorting through thousands upon thousands of somewhat related musings — or worse, inaccurate information. This is why mainstream news can’t and won’t use Twitter in it’s current state to feed their news tickers. To use Twitter as an independent search, the service needs to surface its most relevant content.
What About Hashtags and Follower Counts?
Sure, Twitter has a band-aid search mechanism (hashtags; “#”) that allows one to search for information that’s properly tagged, but among its other handicaps, a hashtag is blind to credibility — just as a user’s follower count is.
Right now, Twitter recommends a list of users to follow (and in turn the number of followers for that select group of people grows), but that recommendation is editorially based and therefore unscalable and decreasingly relevant as the service grows. When Twitter’s user base was small, a user’s large follower count meant something — that what they Twittered was interesting, even. But this simple measure doesn’t account for gaming the system (read: following people to gain followers) or — more importantly — locating the most relevant information, right now (read: can Jimmy Fallon tell me what I need to know about Swine Flu?)
Cooking Up a Secret (Apple?) Sauce
In creating a Twitter algorithm (“secret sauce,” if you will), the following factors can be considered. This list is by no means exhaustive; I’m including it here for illustrative reasons only:
- Behavioral data: e.g., number of times a tweet is “favorited” (i.e., bookmarked) and/or re-tweeted (i.e., shared)
- Meta data: e.g., a tweet’s (and its source’s current) location. This is especially relevant for breaking news-related twitters.
- Credibility: This one has moving parts. To determine a tweet’s credibility, you need to link information about the source with information about the content. On the source (i.e., Twitterer’s) level, this includes factoring things like a user’s history on the site (for instance, how many of their relevant tweets have been bookmarked/shared?) — and the credibility of those who follow this user (i.e., the followers’ history on the site), as well as that of those who share/bookmark the information.
You can quickly begin to see how complex an effective Twitter content mining tool could be. But the bi-product of building one can help solve some of the service’s core growing pains via organically shifting the focus from people (and hence blind, follower-based popularity) to content, for instance. Sure, things like follower counts can still exist, but they’ll be given some context.
Mining for Google’s Gold? Probably Not.
Given all of this, one can’t help but question the likelihood of this acquisition, especially given that this suggested move into the online search industry would prime Apple to more directly compete with Google — which it’s ironically under current speculation for being in bed with. Whether or not Apple is actually headed in this direction is anyone’s guess, and I suppose acquiring a company can be an easier first step than building a search team from scratch — unless, of course, you’re not trying to build an online search product (which I’m still unclear as to why Apple would choose to do).
In any case, mining data will arguably be a necessary piece to Twitter’s next development stage — whether or not Apple (or anyone with enough cash) decides to buy Twitter, or if Twitter decides to grow up on its own. The service can be anyone’s gold mine — anyone who is diligent enough to mine for that gold, anyway.
You can follow me on Twitter here.
[Disclaimer: In my past career life, I was a Flickr employee.]
[photo courtesy of Flickr/somefool]
