Predicting popularity (revised)

Posted by on Apr 23, 2007 in Uncategorized | 3 Comments

Justin

A number of folk have picked up on the Duncan Watts piece (“Is Justin T the product of cumulative advantage?”) in last week’s NYT mag about how a minor difference in diffusion patterns can make a big difference to the outcome. Justin’s tunes like Lockdown was so successful, Watts suggests, because it was marginally more successful early on…

In other words, what we think others think and do significantly impacts on individual behaviour. The point being of course, that as each of us is both influencer and influenced, the pattern created is inherently complex…

Mark asks three v good questions:
1. does this explain the “double-jeopardy effect” that gives big brands an unfair advantage over smaller ones? (I seem to remember Jim Crimmins at DDB having a view on this)

2. Is advertising’s role like that of the peacock tail – to demonstrate status through wasteful use of resources? (I tend to think that advertising has always worked between us, rather than on us)

3. How do you maximise the perception of your popularity without telling everyone how popular you are???
(Faris makes a good suggestion, prompted by the same piece, about how shaping perceptions of what other people think being a really fertile area for marketing and media thinkers – he points to the importance of Style-guides and the like. One other odd example I came across last year is the smart shopping trolley which alerts you to what others are buying in the category – damn, can’t find the link today. Does anyone have one?).

Johnnie, meanwhile makes a great point in criticising Scott Karb’s discussion. All too often we see the copying of other people’s behaviour as “dumb” (as opposed to thinking it all through for our selves, like all good individual-decision-making-units should do..???)

I don’t agree with that interpretation. I think it rests on a certain assumption that our intelligence is individual and not social.

Spot on, J.

The more we see this kind of stuff being considered in the real world, the more our ideas about human beings and their amazing capabilities come into focus (or are spotlit, if you prefer the neutral term). Oh, and the more they’re found wanting. Social intelligence IS the big deal and it’s about time we got to grips with it.

For me the real underlying question revealed by the Watts paper is this: if mass behaviour is an example of complexity, then why do we continue to insist otherwise?

Answers on a post-card, please.

Meanwhile, here’s a few thoughts to help Johnnie out:

Complex doesn’t equal random (though it can seem like it to the individual agent thinking about their own behaviour and to those trying to make sense of the mass’s behaviour in normal terms).

Complexity is hard to predict but not unpredictable (you just have to work a bit harder at understanding the mechanism)

Complexity doesn’t reveal it’s mechanisms on the surface (you have to work from the underlying algorithm rather than the accounts of witnesses).

More of this in the Herd book, natch.

3 Comments

  1. robert mcintosh
    April 23, 2007

    Your comment,
    “Complex doesn’t equal random”
    sparked a memory for me. Have you read Philip Ball’s book called ‘Critical Mass: how one thing leads to another’?
    There is much about what he says (that I remember) in that book which chimes with your views it would seem.

  2. Mark Earls
    April 23, 2007

    Yes, indeed. Ball’s book is top.
    I think he’s really good on lots of things (and his style kept me going through some tricky bits) but particularly “econophysics”: the application of the frames of understanding and modelling of particle behaviour to human and animal examples.
    One example is Boids (http://www.red3d.com/cwr/boids/) – an early animation attempt of birds flocking; another is the great work of Prof Dirk Helbing of Dresden (http://www.helbing.org) which models human behaviour in the real world, from mexican waves to traffic systems.
    Both of these assume that human behaviour is essentially complex, based on the interaction of agents according to fairly simple rules. These assumptions then allow the construction and testing of an algorithm which drives both sets of simulations.
    The interesting thing for us marketing and management types is that the agent may not really be aware of the rules or how they learned them or indeed that the behaviour is rule bound(indeed, I’d suggest our minds give us all kinds of bum steers on the mechanism of our behaviour – for example screening out the influence of others as Asch, Bem, Milgram etc suggest in the Conformity research). This disconnect between the underlying mechanism and our individual experience of it is what leads us astray so often in our attempts to explain consumer or employee behaviour(using traditional research techniques which are rooted in the “I” psychology model, largely). It’s not that subjects lie – it’s just that they’ve got dodgy data about their lives and how and why they do what they do.

  3. Johnnie
    April 23, 2007

    Thanks for the kind links, gov.
    I’d suggest that if a system is complex we can see its patterns retrospectively but not be able to predict the future with certainty.
    This sets a well-baited trap for the mindset which tends to assume things will be linear and that the past is a good guide to the future.
    Dave Snowden is great on this topic – there’s a pdf of his thinking here: http://www.research.ibm.com/journal/sj/423/kurtz.pdf