Persistent Compute Objects and The Fabric of Cyberspace and Quantified Everything
Session Topic: Persistent Compute Objects & The Fabric of Cyberspace & Quantified Everything
Conveners: Phil Windley/T. Rob
Tags for the session - technology discussed/ideas considered: Internet of Things, IoT, pCloud, pico, sensors, actuators, devices
Discussion notes, key understandings, outstanding questions, observations, and, if appropriate to this discussion: action items, next steps:
Phil’s Slideshare: http://www.slideshare.net/windley/persistent-compute-objects-picos
Phil started the session with a slide deck explaining the key concepts and their implementation by the folks at Kynetx. He explained that while Kynetx has one implementation, the concepts are generic and could be implemented in other ways. However, Kynetx has a current working implementation which he was able to draw on for purposes of illustration.
Phil referenced David Gelerntner's book Mirror Worlds as having mapped the territory that we are finally able to build with current technology. The book describes computer models that mirror real-world objects and behaviors in real-time and then record those states and behaviors. The result is the ability to correlate events across previously unrelated systems to improve efficiency, safety and comfort and to generate complex automated behaviors.
T.Rob presented in the second half. Although no slides were presented, the deck from Monday's pCloud & VRM Day is relevant. T.Rob asks the question "what architectures can support a world in which the most mundane objects are instrumented?" We currently have smart forks, basketballs, pens, chairs, shoes, switches, outlets, exercise equipment, medical devices, and more – with much, much more on the way.
The current architecture is that all of these things talk through an SSL tunnel to the vendor and the device owner gets whatever data and integration that the device owner sees fit to provide. This isn't the optimal architecture, it is simply inherited from the world in which computers cost millions of dollars and vendors owned all the data because it could not work any other way. But today it can work a different way because most people have the equivalent of a mainframe form the 1980s in their pocket. There is no reason to cling to the old architecture when vendors kept all the data, nor should we.
An alternative architecture was proposed in which devices report data back to the device owner first, then vendors and other 3rd parties are secondary or tertiary users of that data. Rather than trusting the data because it arrives over an authenticated SSL tunnel, device manufacturers should sign the data so it can be authenticated outside the context of a connection. The vendor's economic model should work without necessarily getting the user's data. Then the user can choose whether and how much of that data to allow out to the vendor. The incentive is then on the vendor to provide some actual value to the device owner in return for access to the data.
This architecture is reflected in the Kynetx implementation of picos. Kynetix provides hosting and initial implementation of code to represent different types of real-world object. But the owners of those objects own the data on which the code is run. Rather than pushing your data to the vendor's cloud where they operate on it (think Google Docs), instead the vendors' software is brought to the data and operates on it there.
Mirror Worlds: or the Day Software Puts the Universe in a Shoebox...How It Will Happen and What It Will Mean David Gelerntner, Oxford University Press, USA (November 14, 1991) http://www.amazon.com/exec/obidos/ASIN/B000QTD1HE
Out Of Control: The New Biology Of Machines, Social Systems, And The Economic World Kevin Kelly, Perseus Books; 1St Edition edition (May 1994) http://www.amazon.com/exec/obidos/ASIN/0201483408
Let's Get Cirrus About Personal Clouds http://www.slideshare.net/tdotrob/lets-get-cirrus