Join us tomorrow in Livermore

Core DI reading: W. Brian Arthur's overview of the history of complexity economics.  Complexity economics is very closely related to DI: the difference is DI's tighter connection to machine learning, and our emphasis on visual models that make complex models accessible to the masses.  Also, DI has a tighter focus: on the decision - than economics, which seeks to model larger systems. We share identical roots however: the core themes of emergence, systems, and complexity are essential for solving complex problems, and complexity economics is way ahead of the pack in terms of fields that understand this.

The explosion of big data has created an adjacent "desert": knowledge and best practices about how to translate that data into value.   This is especially true in large organizations, in areas like telecom (as shown here), health care, and government.  This article is one of many showing that this resource is vastly underutilized. 

Indeed, there isn't even a widely-followed practice of how to write a specification/requirement/objectives document for a project that uses big data.  Indeed, most companies begin with the data, and only later get to the use case / machine learning spec / decision modeling that will translate that data into their business objectives. 

This is what we do for our clients (http://bit.ly/1Ukqi98).

A great illustration of antifragile principles.  And a hope for the future: the complexity we face everywhere will make us stronger.

In multiple arenas, including economics and more, an older approach - which we might call "linear" for lack of a better term - is giving way to a way of thinking that better reflects the world.

This article gives a good summary of what's going on from the point of view of economics.

For instance, the Federal Reserve did an analysis in  2006 that predicted "not much" of an impact on the economy if housing prices dropped by 20%.  This was, of course, wildly wrong.  As have been many other predictions that don't take networks, virtuous and vicious cycles, and emergence into account.

This is, in essence, the core of what we do in Decision Intelligence: bringing the same next level of nonlinear systems thinking to organizational decision making that this article says is starting to creep into economics. 

There is every reason to believe that both approaches have tremendous promise.

Different verticals are adopting data, analytics, and machine learning at different rates.  Digital marketing is near the start of the pack, so other industries like healthcare and telecom can learn a lot by watching its progression.  Here's a great article from my friend Karsten on the state of the art.

Exciting! Take expensive-to-obtain stock market data, encrypt it, then allow anyone to download it and make predictions.  Run a tournament, create a hedge fund, and share earnings with the winners.

One question: if the data is encrypted, it's going to preclude one of the most powerful techniques for maximizing performance: feature engineering. 

otoh, unlike Kaggle, you don't have to disclose your intellectual property when done.   That's a very good thing.

See more discussion here: https://www.reddit.com/r/MachineLearning/comments/3wdr9e/numerai_a_global_ai_tournament_to_predict_the/

Slides from a great talk from H2o's Erin LeDell at Oscon last week.  Good view of how machine learning and AI help organizations to extract value from big data investments.

A great insight from Barry Kort.  And a beautiful little gem of a video trailer about Gregory Bateson here.  A couple of my favorite quotes:
"...He was interested in larger patterns, he was interested in how things are connected"

and

"The major problems of the world are the result of the difference between how nature works and how people think."

I hope to see you in Livermore in June.

More evidence for unconscious rationality: this time it's the "ideomotor effect": you can move things - and even do so in cooperation with others - without the conscious mind realizing it. 

When I throw a ball at an audience during a DI talk, and when I say that decision intelligence is meant to train our intuitions to align more effectively in complex situations, this is what I'm talking about. 

There's a myth that "conscious == rational" and "unconscious == irrational".  I think that Kahneman and behavioral economics have debunked the former.  And now there's an increasing understanding of the validity of may aspects of the rational unconscious. 

Of course, this is what Malcolm Gladwell covered in "Blink".  DI takes the cognitive science insights there and transitions them into practical applications that solve "wicked" problems.

In DI we do this using a combination of methods, from systems thinking, to AI, to big data, and the interconnectedness insights of Buckminster Fuller, all on a basis of improved understanding of neuroscience and cognition.

A grand tour of decision intelligence, with a focus on art and visual reasoning, also including what's happening in the Silicon Valley SimCenter and the Reality Stack vision

Thanks to Don Kimber of FXPAL for making it happen

This is a cautionary tale for machine learning: by detecting weak signals we can discover way more than what is apparent in a raw data source.  And in this article the researchers found that it didn't even require that level of sophistication to deduce a heart condition, a gun buyer, and a woman planning an abortion: not from the content of a phone call, but from the identity of who was called as well as the timing of those calls (aka metadata /  CDRs for my telecom friends).

Old-fashioned symbolic AI was based on the notion that reasoning happened consciously (and could be captured by logic and language).  Subsymbolic AI - of which today's (primarily neural network-based) machine learning is a subfield, recognizes that to emulate human thought, we must go beyond the consciously introspectable.

So what are the implications of subconscious rationality on decision intelligence?  If decisions are made "below the water line", then how can a consciously accessed thing - like a dashboard or visualization - help?  By training our subconscious reasoning. 

That's the core premise of DI: that we "train the brain" through rich, interactive, whole-body visualizations.  This article is good recent evidence that this is on the right track.

Congrats to my friend Bob, quoted here.   Thing are getting interesting :-)

Yeah, Bucky's back for sure

Another bucky follower  / systems thinker working on wicked problems

Mark your calendar: I'll be speaking June 25 at TEDx Livermore.  More soon.

Watch out, Bucky’s back

The latest on the Silicon Valley Sim Center