OPINION4 November 2009

Too few respondents? Why not build some?

AI

Do digital “replicants” have any part to play in social media research?

Demographic Replicator - http://demographicreplicator.blogspot.com/ - is an interesting project by creative technologist David Bausola. Basically it’s trying to create automated Twitter streams – composed of retweets of other people’s stuff – that are keyed to a particular demographic. So, for instance, there’s a call on the blog for input in creating a generic 15 year old X-Factor fan, or a generic social media expert (as if such a thing could possibly exist ;))

This kind of experiment has long been of interest to technologists (think artificial intelligence contests like the famous “Turing Test”) and of course there are other, less pure reasons for wanting a convincing artificial persona: the hordes of spambots on Twitter are examples of crude social media ‘replicants’. Most are easily spotted and the technology has a long way to go.But if it did improve, could these replicants have a role to play in research? Here’s a couple of ideas:

- Bringing social media monitoring to life: a replicant could be an artificial ‘face’ for the aggregated findings and sentiments of social media monitoring across a particular target demographic. It’s the same principle as bringing segments to life with pen-portraits – much easier to conceptualise an audience when you have a fictionalised example of it to relate to. But a replicant data-portrait could potentially talk back and provide new information and insights.

- Stimulation streams in online communities: a replicant member of an online community – as long as he/she/it was openly declared as such – could post relevant links and stimulus material which would keep discussions and insights flowing without direct moderator involvement. The replicant could be set to start posting only when no other member had started a thread in a while.

All far away, until the technology improves – though at the rate our understanding of social media data is progressing, “fit for purpose” replicants can’t be too far in the future.

@RESEARCH LIVE

4 Comments

14 years ago

"weighting" gets 21st century!

Like Report

14 years ago

I'm going to reveal my cynical side here but I think target group Furbeys could become very fashionable - a great way to keep the client team focused on what the current customer base are feeling. Rather like those glossy geodemographic charts that look so good on a client's wall - a well tuned DR should come with a monthly tuning fee and a nice little earner for the supplier who delivers it

Like Report

14 years ago

Very importantly "product" should read "project" in the first para! I'll correct it but it's worth a separate comment.

Like Report

14 years ago

"as long as he/she/it was openly declared as such" And if the replicant weren't openly declared as such couldn't he/she/it nonetheless post relevant links and stimulus material which would keep discussions and insights flowing?

Like Report