Guess what Thomas the Tank Engine and information filters have in common? Both allow for the presence of a Fat Controller – the well-fed, all-powerful head of the railway system who can gather information in context. And then take action on it.

While attending Think G-Cloud 2013 in London, Dennis caught up with JP Rangaswami, Chief Scientist at, to hear more about how filters are becoming industrialized – and, in the process, delivering value.

JP says that, for years, people talked about fire hoses, with guys like Clay Shirky saying there was no such thing as information overload. Just filter failure. But JP still felt that a critical piece was missing: meta data — the placing of information in context through date, time, circumstances and the people involved – allows information to live.

Only that wasn’t enough. People on the move with tons of information streams needed information to become more actionable and valuable. E-mail filtering began to loosen things up that way — though, as Dennis notes, filtering is still a chore. JP agrees, but says that it’s getting easier, and that we’re seeing the early stages of this for the common man through notifications. JP’s informational “magic kingdom”? A structure where everything is allowed on the network and users can set conditions to test against thresholds so that certain actions are taken.

The long view, the guys agree, looks like this: Rather than standardizing to a level of six things programmers think people like to do, instead specify thresholds, wrapped in context, that let you dictate certain actions. The end result? Information reaches you when, where and how you need it, precisely as JP and his colleagues have envisioned it for 21 years.

Show Notes

:44 Growth in metadata over last 4-5 years allows information to live

1:34 It’s getting easier to make information actionable and valuable

2:54 Now in early stages of information labeling — through notifications — for the common person

4:15 Contextual info headed for a “magic kingdom” – thresholds trigger actions

5:24 Long-tail view goes beyond a handful of standardized actions

6:37 “Data is well-wrapped in context because it’s cheap to do so”

7:22 The goal? Information reaches you when, where and how you need it, as envisioned 21 years ago