trendpreneur

Managing online identities can get confusing, especially for algorithms trying to understand who you really are…

I’ve had two articles kicking around in saved tabs for a while now: Home Bases and Outposts over at ProBlogger, and How to Build Your Online Brand from Mashable. Both posts are aimed at an audience wishing to extend their online ’self’, mostly for branding or business reasons; the overall message to anyone establishing a web presence is ‘don’t limit it to just a website’. Instead, reach out to different related social sites, build up profiles and relationships, find people to connect with — this extended network, spanning multiple social media outlets, will enable you to reach far further than a standalone website can.

Okay, but we knew all that already, right? That’s why we all have Facebook profiles, Twitter accounts and so on and so forth. Whenever the latest hip Web 2.0 site goes into beta, we all scramble to sign up and become ‘friends’ with the great and the good. What’s interesting, to me, is what happens next…

To start with, let’s look at the overall shape of this identity we’ve all built. The picture above sort of gives it away — it’s a wheel with ourselves (and our blogs) at the centre, and spokes reaching out to a multitude of outlying micro-selves, little windows on to the big picture that interact in increasingly complicated ways. Because, after all, it’s not a simple graph. We post original content to some of these secondary sites, and some or all of that might feed back to our blogs; we maintain a group of entirely separate social networks, though each person in that network has a ‘wheel’ of their own that we mesh with in one or many places. (I’ve got the same people added on Facebook, Twitter, flickr, FriendFeed, my RSS reader, etc.; but only I know it’s the same person!)

We also further complicate matters by duplicating content. Sure, not everyone is guilty of this, but many people (yours truly included) do things like announce blog posts on Twitter, repost a flickr photo on our blogs, and so on. Some of the points on the wheel are connected this way, and others are entirely disjoint. A human with multiple subscriptions to the same ‘friend’ might run across a link being posted several times across their information feeds, especially if people re-blog the item or add it to their favourites.

Why does this matter? Firstly, it wastes our time. With the current amount of content being published, it’s fairly easy to manually skip stuff we know we’ve read, especially if we read everything from one browser that helpfully makes visited links a different colour. Once you get into the realm of today’s higher-volume information junkies, or look forward into the future a little, this could become a real problem — removing redundancy and streamlining subscriptions is something that really needs to be made accessible.

Secondly, let’s imagine we’re an algorithm trying to determine which links are going to be the most interesting out of a set of data. If we scan across multiple social feeds, which we should, we can’t simply use “number of appearances” as a baseline metric, because those who are more aggressive at self-promotion will win simply due to volume. Of course, simple frequency isn’t usually a great metric, but it’s a particularly bad one in this case. We have a few different ways of approaching this problem — we can backtrack to find authorship and filter out anything posted by its own author, we can recreate the wheels to identify a ‘person’ who exists across networks and only give them credit for posting a link once, or we can take each link on a network-by-network basis and come up with an overall decision about it.

Why would we do the latter? Link and idea propagation within networks and within the Internet as a whole is an utterly fascinating topic. How does one link become popular? How does an idea become popular? How do memes and other viral content spread, and who are the main carriers? Something spreading like wildfire within the Twitterverse might be insignificant in the grand scheme of things, or it might be tomorrow’s Times headline.

What I find most interesting is what happens to links on an individual basis. You can see from my FriendFeed that I might bookmark links posted by people I subscribe to. I might then eventually blog about that topic, linking back to the original article (and forgetting where I found the link in the first place). The ‘friend’ has influenced me, there’s an extra notch on the “my definition of ‘interesting’ overlaps with this person’s” meter, but the only way of actually tracking that is to watch my micro-presences and look at my long-term behaviour.

Is this Big Brother? Yes, but it’s slightly different; instead of an omnipresent disjoint entity monitoring from afar and plotting a multitude of nefarious schemes (like sending me Tesco vouchers for money off things I regularly buy), it’s more like a benevolent mini-deity peering over your shoulder and whispering to you. “Psst, this person bookmarked or blogged about some of the stuff you’ve bookmarked or blogged about. Maybe their other stuff is going to interest you too.” This meta-information and resulting recommendation engine needs to be kept personal and, to a large extent, private; while there’s definitely trends and user profiles that can be pulled out for the greater good, rummaging around in people’s identities without any guarantee that the resulting information won’t be used for evil purposes is a surefire way to lose users (or not get any in the first place).

So, there’s two separate things here. Firstly, identifying today’s most interesting (or must-read) content by looking at what ‘people like me’ are reading and finding interesting, and also identifying those people as new potential subscriptions. That can then be used to find the overall must-read stuff, either filtered to match specific topics or just over the entire Internet; as a side-effect, finding this stuff before it gets old is hugely important to people like professional bloggers. Secondly, we’re looking at the spread of influence from wheel to wheel — who’s the ultimate authority on a subject and repeatedly reblogged, who’s a carrier, who has fingers in every pie, etc. By knowing where the wheels intersect and mapping out the sub-networks and sub-sub-networks we can learn a lot about data flow, but we need to be ethical about it.

The web being what it is, I’m 100% sure someone’s working on this already (several someones, no doubt). There are services I know of that could add in some really cool “interestingness” or profile-matching algorithms because they’ve already got most of the tools they need: Facebook and FriendFeed are both profile-centralisation hubs that know a lot about your other doings on the Web, for example. Whether it’s a valid business model for them to do so is another matter entirely. But wouldn’t it be cool if your wheel could be used to make your networks, and your subscriptions within those networks, more useful to you — without you lifting a finger?