About
Projects
Diary
nine earths invites audiences to see patterns and draw conclusions from its presentation of people living their lives around the world. at the project’s core is a creative spin on ethnography. now that’s a particular approach to media art/activism – contrast with the tired climate change imagery of smoke-stacks and polar bears. but beyond those particulars, there could be all sorts of creative uses of ethnography if only there were the tools to unlock the possibility. the tools we need ourselves…
hence the pitch posted before this. that we won funding for! the work of that application was back in march, we found out over the summer, and there’s been much blocking bureaucracy since. but now the project can start in earnest.
the project is two things. an automation technology for tagging, transcribing and translating video. a product built around that technology that gets the video in, then facilitates the insights out.
what’s the minimum-viable version of that? the tech part is straightforward: we can get baseline functionality using commodity computer machine-learning/ai, so that part’s an integration job. the product part could be straightforward, in that we know the pieces needed. i drew them in the slide above. so that part could just be a user-experience job to create a cohesive whole out of them. in the broadest strokes –
the client is another business. their need is cultural insight. our offer is a creative-industries spin on ethnography made possible by
- crowd-sourcing app to gather video data
- automatic tagging (AI)
- data explorer app for analysis
that is the straightforward answer, but i’m not sure it’s sufficient. take for example orbit, a research project to crowd-sourced a video dataset. the premise was people with low vision would better a future where AI could usefully ‘see’ for them. with this project–participant alignment, we had grounds to hope recruitment would go well.
the more examples you have when training ai, the better, so orbit had a further idea. it would have a crowd-sourcing camera app that people would actually want to use. it invested in product quality to unlock participation at scale. squint, and we’re in the ballpark of ‘suitable for the creative industries’.
in the end, participant recruitment was respectable but didn’t make the hoped-for leap of scale. we learnt that project–participant alignment for recruitment and product quality to convert that into productive participation was not enough.
so – it’s not sufficient yet, because there’s a missing recruitment and participation piece. to be commercially viable it needs to scale free of the transactional grind of financial reward and research staff intervention. in short, there needs to be a ‘why’ for people to get involved, and their friends too. this kind of crowd-sourcing is inherently extractive, and we need to turn that around.
