The filtered feeds of Facebook (and LinkedIn) are the things I dislike most about them, the unfiltered most recent first approach of Twitter what I love about it, so this possibility that Twitter’s going down the algorithmic-filter route worries me – and not just because of recent concerns voiced over how algorithms can affect net neutrality and news reporting.
I very much hope Twitter at least retains the option of turning on the firehose, though I fully get the need to tame the chaos with some kind of algo or filter to pull in new users. Not everyone can get to grips with lists and Tweetdeck – too confusing for the newcomer.
Now don’t get me wrong: algorithmic filtering has its place. One of my favourite apps is Zite, and I was an early adoptor of StumbleUpon (well over a decade ago) – precisely because of their ability to get to know my interests and serve me up interesting content from sources I’d usually not discover by myself. For Facebook to offer up this kind of service, with its vast databases of its users’ Likes, makes perfect sense (though I’d still prefer a raw feed, or category feeds, so I can split off news about the world from news about my actual friends – a new baby or a wedding is not the same as a terrorist attack).
This is why I love Twitter – it is raw, unfiltered. And at 140 characters a pop, it’s (more or less) manageable. Especially if these old stats are still accurate, suggesting the majority of Twitter users only follow around 50 other accounts. If you end up following a few hundred, you’re already a power user, and likely know order them via lists. If you end up following a few thousand, then frankly you no longer care if you miss a few things.
Could Twitter be improved with a bit of algo? For sure. Why am I only ever shown three related accounts when I follow a new one? Why isn’t MagicRecs built in?
But the fact is we’ve already got this option on Twitter – it’s called the Discover tab. And I never use it, because it somehow manages to feel even more random than the raw feed. The problem isn’t a lack of algorithms, it’s a lack of intelligent algorithms, intelligently integrated.