Smart Machine, Smart Human

At the risk of being simplistic, I'm starting to wonder whether the evolution of information tools might not be something like the following:

1. Dumb machine, dumb human - hammers, shovels, etc. (caveat: I'm not dissing the craftsmanship of a Michelangelo...I'm talking about information)

2. Smart machine, dumb human - monolithic systems (e.g., mainframes) with a priesthood of operators that tend to them. "human on the loop"

3. Dumb machine, smart human - distributed systems (e.g. networked PCs, pre-mobile Internet) with humans using the machine as a "speed of need" information tool in a fixed location. "human in the loop"

4. Smart machine, smart human - adaptive agent-like chunks of IT that we weave into our everyday sensemaking activities everywhere we go. "machine on the loop"

The last stage is just emerging...Apple's amazing market cap is evidence of the potential value of such a combination. However, it is a complete paradigm shift (in the strong Kuhnian sense of the word) from machine-centric to human-centric

It raises a fundamental issue: how do you enable humans to easily create, monitor, and manage automated micro-models/narratives? And, do all this within the daily flow of sensemaking?

I mentioned one possible approach a couple of years ago...it seems awfully crude in retrospect and it does not begin to address the create/monitor/manage challenge (which is the real "magic" in the ecosystems that Apple has created).

The only widely used general user IT *modeling* tool I can think of is a spreadsheet (Project is a specialist tool IMO). However, spreadsheet modeling is not done "on-the-fly", so it provides no insight into how to weave micro-modeling into sensemaking flow.

The challenge posed by a smart human, smart machine capability is daunting; the potential is incredible. The good news is that the necessary pre-conditions are largely in place, the needed tools are emerging, and we have some good frameworks for thinking about how to go about architecting this kind of IT (e.g., Klein's Data-Frame, Cynefin).

Meshing Exploration and Exploitation

Earlier this year I posted a couple of items discussing how information technology seems to cycle through complex-complicated-simple-chaotic as a new capability moves from (a) a one-of-a-kind monolithic structure to (b) a partially decoupled structure to (c) a commoditized fully decoupled structure which then (d) provides a chaotic swirl of standardized components that are then used to create a brand new capability with a monolithic structure (Techno-Apocalypse and Business Models, IT, Architecture, Cynefin). I also alluded to the possibility that the time it takes IT to traverse this cycle is rapidly dropping...which may be one reason the whole Explore-Exploit contrast is getting more attention.

Anyway, I just ran across an OSCON (July 2010) presentation by Simon Wardley entitled "Situation Normal, Everything Must Change") that covers much of the same ground...though in a far more entertaining fashion. Highly recommended (even though I'm not a big fan of the "flashing/slashing graphics" style of presentation).

Explore and Exploit - some background

Although it's been several years since this compare-contrast first hit me, I suppose I should make a note of a few of the resources that helped me see that this view/pattern is probably widespread.

1. This first came from my thinking about the Cynefin framework (as a taxonomy, in this case) in conjunction with business processes and innovation...I first saw this as Discovery and Execution. This morphed into Exploration and Execution.

When I went searching for literature with these terms, I came across a number of items, including the following:

2. John Hagel and John Seely Brown's disucussion of Push Programs and Pull Platforms.

3. "When Learning and Performance are at Odds: Confronting the Tension", Singer and Edmondson, Harvard Business School Working Paper. See especially Figure 3.

4. Various HBS working papers by Tushman, et. al. Two that I found useful were:
"Organizational Designs and Innovation Streams"
"Ambidexterity as a Dynamic Capability"

5. "Strategy and Your Stronger Hand" by Geoffrey Moore (Harvard Business Review, December 2005.

6. "Organizing for Innovation in the 21st Century" by Deborah Dougherty (Rutgers Business School September 2004) - this has a nice summary of various approaches to innovation

7. "Planning: Complex Endeavors" by Alberts and Hayes - one in a series of publications about complexity and NCW that was published by the DoD's CCRP. This publication is the follow-on to Alberts and Hayes' "Understanding Command and Control", also recommended. If you're completely unfamiliar with how complexity science relates to organizations, Czerwinski's "Coping with the Bounds" is not a bad intro.

I'm sure there are lots of other (and much earlier) writers who've discussed this...these just happen to be some of the folks that helped me see what seems to be a basic pattern. Now that I think about it, Klein's Data Frame model covers this space well (i.e., "elaborate" is Exploit, "question" and "reframe" are Explore).