I should start by explaining how I come at this problem space. By history I am a network guy. I spent most of the 90’s thinking about networks, breaking networks, building networks and alternating between creating the mess and clearing it up as the Internet grew. More recently I’ve buried myself in the human aspects of technology, leading in businesses and studying psychology. My primary interest is in perceptual psychology – how we interact with the world and how that affects cognitive functions like communication.

Social media smashes all of these worlds together in a wonderful way. It can be challenging at times, as most of the people I interact with come from that funny bit in-between the two worlds: the applications. This post draws on a talk I gave at Social Media Camp London, under the tongue-in-cheek title “six-degrees-of-separation-now-3” – It is also a clarification of the Twitter exchange between myself @timoreilly and @monkchips and subsequent RedMonk Post: “Asymetrical Follow: A Core Web 2.0 Pattern“. Just for good measure, it also includes some thoughts from the film US Now, which I had the chance to see at the RSA this week:

I’m not going to touch on the eGov issues raised in Us Now – that’s a whole other post. I do want to share some thoughts on the way that relationships and communication are modeled in social software, and the blending of “conversational” mediums and broadcast ones.

First, let’s clarify some terms about “relationships” in social media / social software. I’m blogging, you are reading. Great. A blog with no comments is something I used to call a narrowcast model – a bit like TV (broadcast), but with less viewers. Information goes to a select bunch of subscribers. Twitter and a number of other social platforms codify this reader-driver subscription model as “following”. You follow people on twitter, or follow a blog via RSS, which means you choose to receive communication from that person.

In other platforms this type of relationship is referred to as being a ‘fan’. Whilst that term has a lot of baggage, it expresses a specific social communication desire nicely: Let’s say someone is a fan of Stephen Fry (in the traditional sense). They probably want to read all about Stephen Fry’s exploits, see photos, read stories, you get the idea. However, I’m guessing that they would be a little weirded out if Stephen Fry started asking for photos of them, etc…, etc… OK, there’s a whole bunch of issues in there, but just hold this one thought: We have a construct of a ‘fan’ relationship in society, built from the prevalence of broadcast media. It is an asymmetric relationship. Broadcast, like narrowcast, means I consume, but I can not (easily) respond. I listen, but I don’t speak. Or framed differently, you can send to me, but not receive from me. For better or for worse, it is asymmetric.

The standard relationship model in Facebook, Instant messaging systems, and pretty much every collaboration tool out there is that of a ‘friend’. I follow you and you follow me. It is a mutual agreement for bi-directional communication, a symmetric relationship. A ‘friendship’, in social media terms at least, is a mutual ‘follow’. Friends can have conversations – two way communication – in a way that fans (and broadcasters) can not. Facebook introduced fan pages to deal with ‘fans’, and create an asymmetric model. In blogs, the fan model is inherent. Unless you choose to comment on this post, I know nothing about you, aside from some aggregated behavioural data.

OK. Fans. Followers. Friends. Symmetric. Asymmetric. Broadcast. Conversation. A useful vocabulary, even if some of the terms are loaded, and you can walk around sounding like a social media ‘expert’. Let me just say something here:

Broadcast is good!

There, I said it. Depending on your background, you’ll have either shrugged your shoulders, nodded in agreement or screamed at me and immediately unfollowed me on Twitter. The wonderful thing about language: Words are more than words. They have complex mappings on to all sorts of meanings and memories in our minds. Some of those meanings are shared, and some are not. Let’s unpick ‘broadcast’.

If you want to get lots of (hopefully important) data to lots of people, then broadcast is the most efficient way of doing it. That’s why networks – from Television to computing – use broadcast. It is good and efficient. It is also one of the reasons marketers have traditionally loved broadcast. However, broadcast carries an association with asymmetric communication. Shouting as some would have it. If you have read the clue train manifesto (and you should), you’ll know that it’s all about the Conversation, not about shouting or broadcast. The difference comes in the listening – communication with symmetry.

Broadcast is bad?

So, in social media, throwing the ‘B’ word around is bad. For me, it is still just a technical term, and a very efficient form of communication. Anyway, with that, now on to that twitter exchange:

monkchipsmonkchips: symmetrical Follow is a core pattern in social networking, so much so it can cause Scaling Problems for networks not designed for it

Tim O'Reillytimoreilly: RT @monkchips: Asymmetrical Follow is a core pattern in social networking; it can cause Scaling Problems for networks not designed

bmje: @timoreilly @monkchips Asymmetric follow is a hack in social software to enable ‘relationships’ to scale. It is broadcast, not conversation”

timoreilly: @bmje Not so. I follow 400; am followed by 16,000. But I respond to lots of people (like you) who I didn’t know before. Not just broadcast.”

"click" - self-portraitbmje: @timoreilly the wonderful power off twitter and good people – its asymmetry is only partial, due to the power of @’s :)

A side note, James cites my quote saying:

There are those that would would say their is something “wrong” with Asymmetrical Follow, which I would argue is just a function of the power laws you see in any community. For example, yesterday Benjamin Ellis

I definitely don’t think I said it was wrong – quite the opposite. It’s a very useful hack for enabling conversations to scale. I’m guessing that Tim also missed what I meant, since his tweet reads like he thought I was accusing him of the ultimate social media sin – “broadcasting” – see above – and having briefly met Tim, it isn’t the answer I’d have expected from him. Of course I could be wrong. The joys of theory of mind and the limitation of 140 characters. On the contrary, Tim is a very active listener. Asymmetric follow is a way of allowing a form of broadcast, and thus allowing scaling, but all is not what it seems.

The conversation demonstrates something quite different. It shows an unusual property of the Twitter platform:- its ‘follow’ function is not really asymmetric. Tim responded to my message, but Tim isn’t ‘following’ me on twitter – which is fine by the way – although I’m sure I’d be happy if he did, I doubt he’d find me very interesting though!

So, how did Tim get my message if he wasn’t following me? Here’s the magic: If you are on Twitter, anyone can ‘@’ you – essentially directing a message towards you, even if you aren’t following them. It is actually quite a complicated hack and in the great traditions of a good hack, you can fiddle with the settings – see this post on the twitter blog – what people see depends on how they have set up Twitter and what client they use to read messages. This partial symmetry is one of the things that causes Twitter to work so well, and it gets around one of the issues that stops conversations from scaling. Twitter has cracked the broadcast problem with a clever filter.

The Broadcast Problem.

That broadcast stuff. I said it was efficient, but that isn’t the whole picture though. Back in the early 90’s I was responsible for looking after a particularly large computer network. Over the course of a month or two, something strange started to happen. The computers got slower. And slower. And slower. It was a mystery. We hadn’t changed the applications on the computers, or done anything else we thought might slow them down.

After digging around, we found the problem. Some of the applications on the network had started to use broadcast messages rather than the usual unicast (directed) messages. This reduced the traffic on the network, since each message was only sent once, rather than individually to each machine. Very efficient. However, because it was broadcast, EVERY machine on the network had to listen to all of those messages to work out if they were relevant or not. That took a reasonable chunk of their processing power. One machine sent a message, several hundred had to receive it. A little bit of processing power consumed a lot of everyone else’s. Now that is asymmetric.

Conversations Don’t Scale (well).

Imagine if everyone you knew sent you every communication they wrote during the course of their day. Your inbox might feel like that sometimes, but it is nothing compared to what it would actually be like. Now imagine that you had to reply to every single one of those messages. I don’t know about you, but I’d break out into a sweat just thinking about it. Imagine if every viewer of a TV program wrote in with a question. Hang on a minute. You don’t have to wait, it already happened “Strictly Message Board: What Happened”. The result: communication melt down. And that wasn’t even with everyone writing in.

Conversations are tricky things. Huge chunks of our brains are dedicated to making conversations work. I’m not even talking about understanding the conversation, just the sequencing of it. Conversations involve “turn-taking“. Turn-taking is one of the basic mechanisms that enables conversations. Try talking to someone and listening to them at the same time. Oh, you know someone like that already? OK. More seriously, I hope you see the issue, our brains aren’t wired that way.

Now, think about a group conversation. That turn-taking is still going on, just like an old mainframe computer switching between multiple tasks, the listening is divided between the group members. Now think about that group getting bigger. What happens to the amount of listening time? Well, the available listening time stays the same, but the number of people who want to talk grows. Everyone has to make do with a smaller slice of the pie. Conversations don’t scale. Really. Unless some of those new members are just listeners, but then we are back to broadcast.

You’re still reading? Deepest respect! Let me stitch some of these threads back together then. Do you remember “Us Now?” – scroll back to the beginning and watch at least the first few seconds if you missed it.

“More people can say more things to more people than ever in history,” Clay Shirky.

I’d not dispute that, but we must remember that it doesn’t extrapolate to everyone can say everything to everyone. Let’s not kill ourselves trying. Technology has cracked the problem of enabling someone to say something to (almost) everyone. However, we are tired of TV and of broadcast marketing.

Now, technology must find a way for everyone to say something to someone, without breaking that ‘someone’ in the process – be they a politician in government, a genius CEO, or an ever so slightly eclectic techno-psychologist. That requires some very clever filtering.

Has social media cracked the problem? I’d say not yet, but I will exit stage left with this thought from Erica Grigg, of Carbon Outreach, who said this to me (via twitter of course):

carbonoutreach@bmje maybe social media does goodness to scale!?



That it does.