I’ve been hearing a lot about
citizens recently. The iPaaS vendors are all talking about citizen integration, and at the Big Data London event this week I heard several people talking about citizen data science.
Gartner’s view is that development can be divided into two modes – Mode 1 (conventional) and Mode 2 (self-service), with the citizen directed towards Mode 2. For a discussion of how this applies to Business Intelligence, see my post From Networked BI to Collaborative BI (April 2016).
There is a widespread view that Gartner’s bimodal approach is outdated, and that at least three modes are required – a trimodal approach. Simon Wardley’s version of the trimodal approach characterizes the roles as pioneers, settlers and town planners. My initial idea on citizen integration was that the citizen-expert spectrum could be roughly fitted into this approach as follows. If the town planners had set things up properly, this would enable easy integration by pioneers and settlers. See my recent posts on DataOps – Organizing the Data Value Chain and Strategy and Requirements for the API ecosystem.
But even if this works for citizen integration, it doesn’t seem to work for data science. In her keynote talk this week, Cassie Kozyrkov of Google discussed how TensorFlow had developed from version 1.x (difficult to use, and only suitable for data science pioneers) to version 2.x (much easier to use and suitable for the citizen). And in his talk on the other side of the hall, Chris Williams of IBM Watson also talked about advance in tools that made data science much easier.
That doesn’t mean that everyone can do data science, at least not yet, nor that the existing data science skills will become obsolete. Citizen data scientists are those who take data science seriously, who are able and willing to acquire some relevant knowledge and skill, but are performing data science in support of their main job role rather than as an occupation in its own right.
We may therefore draw a distinction between two types of
user – the citizen and the tourist. The tourist may have a casual interest but no serious commitment or responsibility. An analytics or AI platform may well provide some self-service support for tourists as well as citizens, but these will need to be highly constrained in their scope and power.
Now if we add the citizen and the tourist to the pioneer, settler and town planner, we get a pentamodal approach. The tourists may visit the pioneers and the towns, but probably isn’t very interested in the settlements. Whereas the citizens mainly occupy the settlements – in other words, the places built by the settlers.
I wonder what Simon will make of this idea?
Andy Callow, Exploring Pioneers, Settlers and Town Planners (3 January 2017)
Jen Underwood, Responsible Citizen Data Science. Yes, it is Possible (9 July 2019)
For further discussion and references on the trimodal approach, see my post Beyond Bimodal (May 2016)