In case you missed it, there was a great article in FastCompany this week: Is the Tipping Point Toast? written by Clive Thompson. The article covered research from Duncan Watts, author of Six Degrees: The Science of a Connected Age (amongst others), that challenges the belief that you can use influencers (the well-connected) to seed a new trend.
To grossly over-simplify, the idea behind the tipping point is that people watch people who watch people who watch the influencers. (Classic Pyramid stuff.) Therefore, if you can get the influencers to adopt a new product, it will go viral and grow exponentially = big success. Duncan challenges this claim and argues instead that the likelihood of success has nothing to do with influencers. They are a side effect that can speed up adoption of a trend that would have gone viral anyway. In other words, spending your marketing money on the elite few is unlikely to be significantly more effective than standard mass marketing.
Central to Duncan’s argument is the habit we have of taking an event and then working backwards to identify what happened and spot a pattern that can be reproduced. Anyone who has read Freakonomics will recognise the flaws in this approach – correlation does not guarantee cause and effect, and indicators are easy to spot once you know what you are looking for.
Simple demonstration. Go find somebody, find a table or similar surface, ask them to ‘name that tune’ and tap out the Happy Birthday song with your hand. It will be a miracle if they spot the tune when it is tapped in monotone with no words. Tell them what the tune is and then both tap out the tune. It is easy to ‘hear’ it when you know what is being played.
On a related theme, I am currently reading a book about unpredictable events – The Black Swan: The Impact of the Highly Improbable, by Nassim Nicholas Taleb. The Black Swan is all about unpredictable events and why we never see them coming but think we should have (and therefore think we can predict the next one and get it wrong all over again).
In the Financial Times on Friday was yet another example – Last year’s model: Stricken US homeowners confound predictions:
¨…it seems that mathematical models used to predict future default rates, based on past patterns of losses, have gone wrong because they did not adjust to reflect shifts in household behaviour.¨
In the past, when US households struggled to repay debts, they tended to default in a certain order. Credit cards and car loans were the first to suffer. Failing to pay your mortgage was the absolute last resort. (Losing your house = big social no-no.) This time around, people are defaulting on their mortgages before personal loans or credit card bills. (The current climate has created negative equity and changed behaviour – why repay a mortgage for a property you don’t have any stake in anyway.)
There are two technology trends that need to beware this Achilles heel with using the past to predict the future. One is performance management (and its sibling: busines intelligence) – the use of data visualisation to analyse your information sources and gather new insights that should improve decision making. The classic turkey scenario – you get fed every day and expect to be fed again tomorrow. Instead, you get your head chopped off.
The other trend is social networking applications, in particular any that plan on using the ‘Social Graph’ as a method to track and use relationships. And that leads onto the final link (this post is really a collection of links from the week…) an article in VentureBeat – Google’s Marissa Meyer: Social search is the future. Coincidentally(?) it has come out at the same time as a video clip of Google’s new Social Graph API.
You can find out more about the Social Graph API at Google Code.
The example that Brad gives in the short video clip makes sense but let’s change the players. Instead of Brad finding his friend Bob on Twitter, imagine a spam company creating a loooooooong blog roll of ‘friends’ on LiveJournal and then setting up an account on Twitter to find out more contact details for all those ‘friends’. The potential problem with the Social Graph concept is that it reduces social networks down to a logical drawing, when relationships are anything but. Our concept of who is, or isn’t, a ‘friend’ has changed with the arrival of massive social networks such as MySpace and Facebook, and our behaviour has changed with it. The concept (and associated behaviour) will likely change again in the future, when organisations learn to exploit those friendships in new and unexpected ways. Will the API adapt?