Deep Learning and Systemic Insight

A number of months ago, I staked the following claim:

Through a combination of thoughtful physical space planning coupled with elements of rich mobile, network, and sensor data we can engineer the randomness of human interaction, in the hope of enriching for serendipitous outcomes. Such outcomes will be driven by engaged actors contextualizing previously unknown but knowable information/data/knowledge.

There’s an emerging community of thinkers/doers exploring the intersection of data collection, modeling, and people analytics - in the service of engineering randomness. In the context of ‘work’, this paragraph represented the best synthesis of my noodling at that time on the topic.

Wait, why is work in single quotation marks?

When I talk about ‘work’, I typically refer to actors engaged in activities that contribute towards something that is bigger than any one of them. Interestingly this is inclusive of profit and non-profit ‘work’ … and it’s also inclusive of, perhaps, civic engagement.

Now, hold that thought for a moment, while we turn to Deep Learning …

Earlier this year, some researchers from Google DeepMind, published some research in Nature which demonstrated ‘Human-level control through deep reinforcement learning’. In short, a deep learning architecture was fed images from classic Atari games. In addition the architecture was fed the game score at that time. In general, as the number of learning iterations is increased, the architecture determines ever-increasingly optimal strategies for increasing the game score.

For example, in Breakout, the ‘machine’ learns the ‘shimmy’ (the strategy of applying a little noise to position of the paddle), before figuring out that popping the ball over the back of the wall is a quick way to get rid of a lot of blocks, and increase your score. A mind-blowing, accessible, and visual introduction to this research can be found here.

Think about this for a moment. An algorithmic infrastructure, given minimal input, subsequently determines the underlying rules of a system - and then learns to navigate that system with a view to increasing an objective function.

Now, hold that for a moment while we turn to distributed computing …

With services like Amazon Web Services we have unprecedented compute power available to us, but it’s actually really difficult to write software that works optimally across hundreds-of-thousands/millions of machines. Adding compute power works when the underlying problems are separable, or embarrassingly parallel, but less so for problems that are intractably complex - you know, like social systems.

Interestingly, companies like [Improbable]( are exploring exactly such architectures - with a view to simulating complex social problems. Improbable CEO Herman Narula explores this in a recent a16z podcast and alludes to the potential power of enriching simulated worlds with sensor and Internet of Things-enabled devices …

Now, let’s bring it all together …

What if we:

  1. Find a space to explore (space could mean workplace, city etc.)
  2. Create a sensor rich data abstraction of the workings of the space
  3. Build a simulation of the space, enriched by the rich sensor data

Such a simulated environment might be a real boon to exploring cause/effect in a way that mirrored the true complexity of the social nature of the problem(s) faced in spaces containing people.

Now, imagine the following (and I admit this is all tenuously sci-fi stuff …): what if those simulations were fed into a deep learning architecture? Could a machine then ‘learn’ the unwritten rules for how people navigate the underlying space? (I fully appreciate that the generation of the simulation data is an additional unnecessary step - this is perhaps most useful when the simulation is exploring the effect of an as-yet unimplemented intervention - prospective questions might explore whether ‘the rules’ changed in (un)expected ways).

Such algorithmically generated insights could be amazingly powerful tools for objectively exploring the spaces that contain our interactions - and, most excitingly, all of the pieces needed are currently available, or in active development. It’s an exciting time to be thinking about people analytics.

Thanks for listening,



Now read this

Randomness, Serendipity, Bounded rationality … and Light Cones

Kudos to Greg Lindsay for continuing to Work Out Loud as he explores the role of place, network, and data in engineering the conditions for enriching for serendipitous outcomes. He has a wonderful recent piece on Medium, where he... Continue →