Reading Time: 4 minutes
A colleague of mine asked me this question a while ago when we were discussing the problem of the facebook algorithm. Here’s how it works: facebook shows you the news that it thinks you’re most likely to interact with. After all, if your second cousin posts endless pictures of things you’re not interested in, it makes sense for facebook to dial back on his updates and dial in some more interesting content from your sister, who posts news articles that keep you informed.
The problem is that, although this sounds fine in principle, in practise it creates a very different environment to the one you would expect. Often news is dialed back to make way for easy ‘clickbait’ type content, or videos are prioritised because they’re more engaging.
So, a computer decides what you get to see and what it will hide. And computers – while they can be incredibly smart – are not always going to make the same decisions as humans.
Over on GigaOm this week, Matthew Ingrams discussed the merits of Twitter vs facebook as a source for news. In the wake of the shocking incidents in Ferguson, Missouri, some people were surprised to see facebook almost completely devoid of news. Twitter was filled with live updates, eye-witness reports, photos and videos of events as they unfolded. Facebook: almost nothing. Why so different? While Facebook has a filtering algorithm constantly trying to guess what you’ll respond to, Twitter shows you everything from the people you follow, so you’re going to receive all the updates from people you follow in your timeline, whether you’re likely to retweet them or not. While Facebook is trying to be your personal shopper, hand-picking items it knows you’ll like, Twitter shows you all of the products in the shop.
The Twitter model is great, for a while, and gets around this initial problem of algorithmic filtering. Unfortunately, because you see everything, it can be incredibly difficult to keep track. We humans are, and always have been, fans of filtering and sorting. Even before the internet age, when we were bombarded with data from all sides, we’d rarely seek out everything – choosing instead to curate our sources (by buying a specific newspaper, or watching a particular news channel, for instance). To continue the shopping analogy, Twitter gives you the option of seeing every product, but there are so many on such a fast-moving conveyor belt you barely have time to examine something before twenty other things have gone whizzing past.
Can there be a balance? Well, there are a couple of possible ways to solve this problem. Method one – the one which facebook is trying is to simply make automated filtering better. Facebook tries to improve the algorithms so that they don’t get too one-sided, or churn out too much similar content – their priority is to keep you on the site and get you using it a lot, so ultimately if their algorithm is stopping you from doing that they’ll improve it. Twitter is also tweaking what shows up automatically on the timeline – recent changes to how ‘favourites’ are displayed have met with opposition from users, but it’s one of many experiments to try and make Twitter feel like a more ‘usable’ place. To engage new users, Twitter is trying to introduce a form of content curation that makes it easier for people to find what they love.
Will either of these techniques work? Possibly. But one of the reasons we started Cronycle is that we think there’s a better option. Not better algorithmic filtering – because it will ultimately always run into the ‘machine’ problem – but applying a layer of human curation to the deluge of content.
Human curation is the solution to algorithmic content filtering
Cronycle takes all of your sources (the RSS feeds you subscribe to, the Twitter accounts you follow) and indexes all of the important content (anything that includes a link or image is pulled through). You can then filter and curate those posts into a collection based on criteria you choose – you can add a filter for the latest breaking news story, for example, filtering in only content from the news teams you really trust. You could have a different collection for updates on a particular area of industry, which gathers articles from expert sources that you’ve chosen yourself.
There’s a certain amount of machine help here, for sure – you’re not creating your own newspaper. Cronycle is useful because it helps you cut through the noise, and prevents you having to scroll through reams of irrelevant content just to get updates on the latest news story or blog post. But the key difference between Cronycle and any algorithmic filtering system is that you won’t run into the ‘facebook problem’ – machines pushing you content based on simplistic models of your behaviour. You choose the sources, you set the filters, and Cronycle indexes that content. Unlike facebook, it won’t ever second guess you.
Published on 21.08.2014 by Marina Cheale