Skip to content

Automation, feedback loops and the redundant human

May 24, 2012

AKA Who watches the watchmen?
OR Dirty deeds done dirt cheap
OR Sit back and enjoy the ride?

I’ve been to a number of analytics, big data, smarter planet (IBM’s initiative to make the world a better place using technology) and automation presentations in recent times, and I wonder whether there is a chord un-struck with most of those involved in the fast pace of progress. The attention is focused on the technology; how cool and smart it is, how much better it is than people and their error-prone fat fingers and sleepy attention span. I wasn’t always a technologist, and in fact my undergraduate degree was in Psychology. I have the same background as Mark Zuckerberg of Facebook infamy, but please don’t get us confused – I have different hair.

The last ‘AKA’ title up top was the subtitle to my undergrad thesis, which described a phenomenon known as the “out-of-loop performance problem”: an inherent and very human issue in automation, brought about due to the changing nature of human involvement in sophisticated and complex systems such as cars, planes and nuclear reactors; as they fast move towards closed feedback loops of many systems of systems. IT is fast moving towards the same human problem, and here’s why.

The technology emerging today is incredibly exciting and the time we live in is so poignant. What I mean by that is that many different streams of technology are coming together – and I don’t mean converged devices or the consumerisation of IT per se – but that we have emerging clusters of expertise. Whereas previously we had a whirring box that did everything (the jack of all trades), now we have defined and segmented architectures that break out of the enterprise, like a living, breathing entity. Applications have emerged that are optimised for providing one specific service incredibly efficiently and effectively, and the capability to integrate these applications together, and to integrate applications to underlying layers now means that we are building a super-intelligent mesh; a matrix of information.

We’ve enabled it – by architectural segmentation, partnering, outsourcing, SOA, BPM, cloud, obviously the internet at the core of it all. Software-defined networks mean that we can even morph and change virtual connectivity circuits in a relatively touchless manner. We talk of round-tripping from data capture, analysis, interpretation, action and back around – these are essentially and fundamentally just closed feedback loops. Smarter Planet relies on this premise – that we have pervasive intelligent devices capturing vast amounts of data for us to harvest, analyse and act upon. More and more information, more accurate models, greater insight, better decisions – and now we have all the pieces of the puzzle to start making that a closed-loop reality! The architecture is rapidly becoming an automaton. Not one that can quite take over the world or express a genuine desire for your clothes and your motorcycle, but still… so my main warning surrounds how this changes the role of people.

The fact is – and this isn’t meant to be an Eeyore view of the world – things go wrong. Stuff breaks. Models have flaws. Software has bugs (very, very occasionally 🙂 ). Nothing is 100% reliable. Even those systems conceived to monitor others can themselves break. How often do you hear of someone taking their car in to a garage because a red light came on, only to be told it was a faulty sensor?

So if the watchmen have flaws, who watches the watchmen? Just as the changing role of the IT Manager is now one of supplier management; the changing role of the ops engineer means that instead of watching and receiving alerts on simple metrics; the watching is now a considerably more complex task.

Humans make terrible monitors. We are truly rubbish at it. We get bored, distracted, fall asleep, chat, think about dinner… and we miss stuff. Important stuff. So we need to be aware of the automated changes being made, trace the reasons for those changes, view information in ways that directs us to the most important first; correlates information relating to one ‘thing’ in the real world that we can make sense of. Watching log files roll by is not an option. How to present information in the right way? Technology is edging into the science of ergonomics. How do you govern a semi-automated system? How about a system of systems? How about if the boundaries of that system span the globe and out of the walls of your office, or the safe, familiar and friendly ‘vendor of choice’ boxes in your datacentre?

Did you think I’d have all the answers? Maybe I do, but I’m keeping them to myself for now! Or maybe I don’t… but one thing’s for sure – Governance is still a human concept, in a world of high technology. When the machines become self-aware, we better be sure we had that automated fallback policy backed-up… we’d never need it they said… it’s all just dumb metal and silicone they said……..

Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: