OPEN

I want to play with a neural network!

The more I immerse myself in the world of the web, the higher my aspirations for using technology become.

Right now I’m lucky enough to be working alongside some very knowledgeable experts on search engines, have an array of analytics tools and am working with a genius developer to create more things to help make sense of social media and the web generally.

Which explains why now I can read articles about neural networks and start musing about what I might be able to do with that kind of computing power.

(I might start mwah-ha-ha-ing myself if I go any further with that train of thought. Put the internet connection down, Mr Mayfield, and back slowly away…)

Anyhow, Malcom Gladwell‘s written a scorching piece about neural networks and using them to understand highly complex and Dave Pollard‘s thoughts on the subject are worth reading too. 

What I really want to do is understand as much as possible about how networks (social and otherwise) form, act and evolve online. 

Other people are looking to predict the success of movies, how markets will move, and presumably the ultimate complexity system obsession of the English, the weather… 

Antony Mayfield

2 responses to “I want to play with a neural network!”

  1. Heather Yaxley Avatar
    Heather Yaxley

    I think the predictive investigation of networks has application for Public Relations in relation to issues and crisis management. Traditional models tend to present a linear chronological approach to how an issue or crisis will develop, peak and recede. But recent crises, and the emergence of online social media and networks, must make these models redundant. Are there patterns and key dimensions that can be mapped and modelled in order to predict which issues will peak, when and how? The old way of responding by relying on the pre-determined plan in the cupboard is no longer relevant as crisis development today seems more akin to chaos theory. Any thoughts?

  2. Antony Mayfield Avatar
    Antony Mayfield

    The first step is to understand how online networks work – and to start thinking in terms of planning for these rather than for the news-cycles we were used to when dealing simply with channel / broadcast media.

    As I’ve mentioned before, the main differences are:

    1. Scale: there’s so much more content and players: do we know what our communities of interest look like online? How do they behave and communicate when “crisis” stories break?
    2. Speed: a day seems like an age in networks. Silence is incriminating and damaging – how do we deal with that? What’s best way we can start communicating soonest without
    3. Evolving nature: you’re right to say that networks look different at different points in a news cycle. It’s reflected in the kinds of results that get to the top of search engines in a crisis. The kinds of results for “Dell laptop fire” or “Dell battery recall” were very different on day one of the crisis to two months later – we need to plan to communicate so that we meet the short and longer term needs of networks of interest around a story or subject and ensure we are producing communications and content that will be successful for both.

    In summary, networks are highly complex systems. Like the weather and economies we can study how they behave and have behaved in the past and develop theories that can be tested and help us to plan – but we need to acknowledge that we can’t predict how they will work in any given scenario with absolute certainty.

    There are many tools out there for looking at behaviours of networks and online communities. A few firms – including mine, Spannerworks – are developing ways of seeing and expressing how networks develop.

Leave a Reply