The typical CI analyst: Man versus Machine

In her previous post, Anne discussed the role of CI analysts. In this follow-up post I’d like to discuss their raison d’être. Is the human brain replaceable in the competitive intelligence process? Can it be fully automated at some point in time, using processor power instead of brain power?

It shouldn’t come as a surprise that I strongly believe human analysis is a vital part of CI. The more complex the job is – without a lot of structure, routines and a clear and easy-to-follow process – the less likely it is to be automated at some point in time. Have you ever tried writing a work instruction for a CI analyst? Don’t – you will spend ages writing down exceptions for even the simplest of analyses. So if we agree on the fact the human brain is vital (and therefore irreplaceable), the question is: how important is the role of machines then? Do they have a right to exist in a CI process?

To answer that question, I cannot help but referring to an article about chess computer Deep Blue and its close encounter with Russian chess God Garry Kasparov, back in 1997. The article is called Did Garry Kasparov Stumble Into a New Business Process Model? (by Andrew McAfee, Harvard Business Review) and describes the parallel between the game of chess and business models. I would like to more specifically use it to illustrate the role of men and machines in CI.

Kasparov versus Deep Blue

The story goes as follows: during the eighties, chess computers were hardly a match for Kasparov. In 1985 he beat 32 dedicated chess computers simultaneously, no single draw. In 1996 he beat IBM’s super computer Deep Blue 4-2, but when Deep Blue came back new and improved in 1997, it was the first time a chess computer beat Kasparov (3,5 – 2,5). Hurray for Deep Blue, but what does that tell us? Basically, Deep Blue beat Kasparov because of its ability to process tons of possible moves and their outcomes. In 2003, no super computer was needed for that anymore, and Kasparov had a hard time beating commercially available chess programs running on standard servers. So at that point in time, the processor power of a computer overtook that of a rather brilliant human brain.

Fortunately, the story did not end here. One started experimenting with the combination of human strengths and computer strengths. And according to Kasparov that lead to interesting results: First, the combination of a strong human player with a reasonably weak laptop beat super computer Hydra silly. So the combination of human creativity plus some computer power is much stronger than a very strong computer, that’s good to know. But the most interesting and perhaps surprising result of the experiment – and I have to agree with Kasparov (and author McAfee) on that – is the following conclusion: the strongest combination did not contain either the strongest chess player or the strongest computer. In fact it was the process that lead them to victory. Or as Kasparov put it:

Weak human + machine + better process was superior to a strong computer alone and, more remarkably, superior to a strong human + machine + inferior process.

CI analyst versus CI software

So if we move back from the realms of chess to those of competitive intelligence, we can rest assured: we shouldn’t be looking for the Deep Blue equivalent of competitive intelligence to replace our analysts. Instead, we should look for the level of aiding computers can offer and, most importantly, design the process in such a way that the synergy between the two is optimized.

From what we’ve learnt above, we now know we need to do two things: first, decide which human strengths we should continue using and which tasks a computer is better suited to perform and second, design or alter the processes to make sure both strengths are used most effectively.

Human strengths versus computer strengths

‘Analyst’ in fact is a very good job title. Of the operational phases of the intelligence process, analyzing definitely is the phase where the CI practitioner adds most value. Most activities in this phase require creativity, intuition and a high degree of improvising – typical human strengths. In my humble opinion, the only degree of automation in that phase is a ‘smart Office tool’. For example, a CI tool that can serve as a framework for analysis, one that hands you all the information found in such a way that you can apply it easily to your analysis.

The first phase on the other hand, the collecting of information, is one that is very suitable for partial automation. Computers have the reach, precision and power to search a vast amount of sources, either in the company or online – typical computer strengths. On the other hand, a small minority of information may be best collected by a human being. And that is where the need for a solid process comes in, in uniting both worlds.

The process

When it comes to information gathering, you need processes to make sure the automated gathering is always accurate and up to date. No matter how well it is initially implemented, no system can maintain itself in a changing environment. You need to have a process for periodically checking the effectiveness of the automated tasks. Think of both changes in sources and changes in your environment. New entrants, new products, changing legislation, etc. Apart from that, you also need process steps to make sure you are not manually checking sources that are also automatically processed. When designing these process steps (typically part of the ‘setting direction’ phase) , you think about your complete set of available sources (this can also be knowledge in the heads of your colleagues) and how to use them. Systematically interview those colleagues to tap into their pool of wisdom, whereas you let the computer cover the more straightforward sources (such as internet pages, databases and information available in the systems of your company).

Knowing how to best synthesize these pools of information – even if you would not be the most brilliant CI analyst in the world (which of course you are) with a super computer at your disposal – will make you unbeatable by any super CI tool or CI God. Kasparov’s first law of CI.

Did Garry Kasparov Stumble Into a New Business Process Model? by Andrew McAfee (Harvard Business Review)
The Typical CI Analyst by Anne van den Brink

Share and Enjoy:
  • Google Bookmarks
  • LinkedIn
  • Technorati
  • Twitter
Comments: 2 Comments

2 Responses to “The typical CI analyst: Man versus Machine”

  1. [...] Dit blogartikel was vermeld op Twitter door Jeroen van Luik, Dink Intelligence. Dink Intelligence heeft gezegd: RT @JeroenvanLuik: New post: The typical CI analyst: Man versus Machine <– About automation and processes in #CI [...]

  2. Raúl Baños says:

    Great job with the Kasparov example. I completely agree with your post conclusions. We can do a good job automating some processes, like gathering and spreading strategical information across technicians, managers, -or self use-, etc. with an intensive use of RSS feeds for instance, but the most important part it’s the analysis of the CI practitioner. Technology as a resource, not as an end in itself.