posted 1 Oct 2009 in Volume 13 Issue 1
KM frontiers: Self-signifying knowledge
In the first of a series of articles on self-signifying knowledge, Jan Wyllie discusses the philosophy of interpretation and narrative.
“What the overemphasis on the idea of content entails is the perennial, never consummated project of interpretation.”
“Like the fumes of the automobile and of heavy industry which befoul the urban atmosphere, the effusion of interpretations … today poisons our sensibilities.”
Susan Sontag, Against Interpretation, 1964
Against interpretation 2.0
Both the arts and the sciences have come to be dominated by an endless process of interpretation and reinterpretation by more and more people with biases and vested interests. For the purposes of this article, interpretation is not used in the broadest sense, in which all sensory input is assumed to be interpreted and given meaning by the brain. In this sense, all facts are interpretations.
It is used in Susan Sontag’s sense of, “… interpretation means plucking a set of elements (the X, the Y, the Z, and so forth) from the whole work. The task of interpretation is virtually one of translation. The interpreter says, Look, don’t you see that X is really – or, really means – A? That Y is really B? That Z is really C?”
What is it that is being interpreted? It is content, whether in the form of speech, image, text, even music. Content is king. Its interpretation is deemed to be the way towards discovering the truth (or not). The ever-present danger is that by finding and interpreting the content that we agree with – that is, the content that fits our own preconceived knowledge – sources of information can become seriously discredited and their value diminished. The danger becomes even greater when propagandists with vested interests enter into the discourse.
This kind of echoing process, which takes up so much time and effort by knowledge workers, leads to what Dave Snowden, founder of Cognitive Edge calls “entrainment”, often blinding them to the significance of new knowledge and/or experience. It was Karl Popper’s insight that scientific theories can never be proved; only falsified. For Popper, the high rate with which scientific theories are found to be untrue is a sign of the strength of the scientific method, not a weakness.
Business and government cadres are entrained to manage events by interpreting what they mean based on a mysterious combination of feeling and rationality in order to make the right decisions. Most management systems – from divine-right monarchy and democracy to Taylorism and business process re-engineering – tend to be blinded by their assumptions and prejudices to significant changes in the ‘real world’, lost in the smoke and mirror world of interpretation.
The analytical tools of cause and effect can be all too quickly swamped by complexity, making them a very dangerous basis on which to determine what is significant and thence to make decisions. The consequences are a plethora of misjudgements and mistakes contributing to the environmental and economic devastation that we see almost everywhere we look.
The purpose of the traditional practice of information and knowledge management (KM), not to mention library science, is to provide access to content, so knowledge seekers can find the pertinent information to interpret for their purposes.
Although a great deal of ‘progress’ may have been achieved in certain domains by traditional knowledge content storage, retrieval and interpretation processes, it has been at great cost in the wider context of economic and environmental sustainability. It is also arguable that the main achievements resulted from intuition (seat of the pants) and serendipity (luck) rather than from the enormous infrastructure of interpretation operating in business, academia and government.
Nowadays, though, with the information explosion brought about by the internet, a realisation is beginning to dawn that traditional KM systems are becoming overwhelmed by the toxic combination of volume and complexity. Nobody can possibly gain the crucial bigger picture view by reviewing and interpreting all the pertinent content, even if it was served up by the perfect mind-reading retrieval engine. There are just not enough hours in the day or years in a lifetime.
Google and its ilk are increasingly highlighting the issues of information overload and knowledge incoherence as increasingly acute and unsolved problems. Twitter, with people following each others other’s tweets in the tens of thousands, is turning it into a crisis, while pointing towards an adaptation, which some are even saying is a new reflective faculty of human consciousness.
Self-signifying data requirements
“If excessive stress on content provokes the arrogance of interpretation, more extended and more thorough descriptions of form would silence. What is needed is a vocabulary – a descriptive, rather than prescriptive, vocabulary – for forms.”
Susan Sontag, Against Interpretation, 1964
What if people are missing out on a potentially new knowledge perspective where useful intelligence could be extracted not from the interpretation of content, but by making inferences from information flows? Intelligence in this context would consist of significant information, which would otherwise have been missed, delivered in as close to real time as possible. It is knowledge that otherwise could not be known, whatever the amount of interpretation by any number of people.
Significance would no longer be hidden in the content, but would emerge from the patterns of content without interpretation (a process Dave Snowden calls “disintermediation”). Meaning would no longer be exclusively hidden in content, something to be revealed by interpretation; it is also in the significance inferred from studying form (structure) of content flows – inferring what the data could be signifying and where it points to, rather than interpreting where it came from. The object of the study is the form of information, not its content. And once an information flow had been formalised, it could be analysed statistically in meaningful ways. The output would be literally self-signifying (to use yet another Snowden term).
The most important thing about inferences made from self-signifying data would be that it would enable people to ask different kinds of questions at the same time as providing statistical indicators suggesting (changing) answers. What kind of questions? Questions like: ‘What will the future of consumerism be?’ (See sidebar, page 15).
But what about the content of the material being analysed? It would still exist as the source of the analysis, but its role would change from being something to interpret or agree or disagree with, to being its significance as an indicator in a meta-analysis. Of course, if the analysis is not to occur in a complete vacuum, reference would have to be made to what sources are saying.
Here great care would have to be given to ensure that the content is reported as evidence, rather than interpreted as meaning something. Quoting representative sentences, rather than rewriting them, would be a useful strategy. Indirect language where the ‘I’-word is never used would be another.
There was a time when journalism expected its practitioners to be objective reporters. Now, the fashionable thinking is that nobody can be objective, and reporters are biased in their own interests (whether vested or not). Unfortunately, instead of using this knowledge to resist bias as much as possible by taking a disinterested perspective, much journalistic practice is now increasingly about interpretation, show-biz and broadcasting entrained ideas, rather than straight reporting and critical thinking. Nevertheless, a clear distinction between reporter and interpreter should be a vital one for journalism, and would be a necessary condition for compiling self-signifying data.
As for the inferences that can be made on the basis of this data, they would derive first from the patterns of information in the flows of data. The ability to compare flows over differing periods of time would enable a high degree of cross checking between flows as an ongoing means of verifying or falsifying inferences.
A bit like markets and indeed the ‘real world’, the world of self-signifying data would not be controlled. It would be a world of both trends and surprises independent from the hurly burly of interpretation. That is not to say that it would have any greater claim to truth, which is a questionable concept in itself. Like any scientific data, self-signifying data would be simply another facet of human reality.
Susan Sontag was asking for a new language to describe the forms of communications flow. Unlike entrained thought, which closes off possibilities and questions in its quest for truth through interpretation, the magic of language is that it can describe and classify instances, while at the same time being open so that it can describe new unthought of possibilities. Even highly simplified, artificial languages, such as computer programming codes, can create vast numbers of applications that were not even conceived by the language creators.
So, self-signifying knowledge would require new artificial languages that describe the form of the content, rather than the content itself.
“To interpret is to impoverish, to deplete the world - in order to set up a shadow world of ‘meanings’. It is to turn the world into this world. The world, our world, is depleted, impoverished enough. Away with all duplicates of it, until we again experience more immediately what we have.”
Susan Sontag, Against Interpretation, 1964
Self-signifying knowledge practices do, in fact, exist. They have been outside human experience until recently, other than by a few pioneers who, by definition, were ahead of their time.
Various disciplines under the heading of content analysis have been have been working successfully in this way since the 1930s. For example, Allied intelligence agencies were able to infer German troop movements in World War Two from a statistical analysis of public train timetables.
Eugene Garfield’s Science Citation Index, which counts the number of times scientific papers are quoted in other scientific papers – when, by whom and about what – has been an important influence on scientific thought for nearly 40 years. The self-signifying data that it generates has enabled scientists to see new patterns of thinking emerging, which would have been simply impossible using traditional methods of interpretation.
In the UK, two companies have been working in the field of generating self signifying data for years – Cognitive Edge (formerly Cynfyn)1 and Open Intelligence (formerly Trend Monitor).2
According to Snowden, Cognitive Edge conceives of language, not as a deep structure, nor a mental model, nor even a semantic network, but as another example of a co-evolving, complex adaptive system. People use language to make up and tell each other narratives of their experience using a wide variety of media. For Cognitive Edge, these narratives or stories, along with how the tellers feel about them, are the source of the self-signifying indicators about a group’s thinking.
How is it done? The first step is the collection of stories, ‘narrative fragments’, as they are termed. Narratives are collected from groups of people and/or (I presume) different types of media. At this point, all that exists is content that would, in the traditional way, have to be read and interpreted for any of its significance to be appreciated.
What is promised, though, is self-signification. Cognitive Edge has invented and patented some nifty intellectual/software tools for doing just that, the latest of which is called a triad.
A triad is a triangle. Each side represents a scalar question about the associated narratives. Here is an example of a triadic question from the Cognitive Edge website.
“I would characterise the leadership behaviour in this story as: altruistic (top of triangle); assertive (bottom left); analytical (bottom right).”
All the user has to do is to move their mouse to position a the round indicator at the place within the triangle which best represents their view of the significance of the underlying story. The neat thing is that this self-signifying triangulation yields a three-dimensional graphic, which signifies without any intermediaries or interpretation what different groups think about questions, where they might be open to change, and where they might resist. This knowledge can then be used by client organisations to manage change using what are called ‘safe-fail’ interventions.
All a user needs to know is how to do it, requiring a three-day training course, and SenseMaker software. The business model seems to consist of an income stream from signing up trainees for accredition, then using the accredited Sensemaker practitioners to sell bespoke software applications into big organisations. Come to think of it, doesn’t IBM do that? This is not surprising since the company was originally a 1980s management spin-off from IBM.
These self-signifying techniques have been applied in a huge variety of different contexts from museum staff improvement programmes to anti-terrorist intelligence research.
Snowden emphasises that the really hard work is in designing the questions, especially ones with a meaningful triadic structure. He speaks of hiring anthropologists by the man-year on one project. On the other hand, he says, children use implementations to produce significant results as well as, if not better than, adults.
Like Cognitive Edge, Open Intelligence also has stories as its source. The raw material, from which the self-signifying indicators are extracted, is what is being said traditionally in published sources, now increasingly in blogs and potentially in real-time tweets.
Key phrases and links to the source publisher are classified using a faceted schema currently designed to capture data on the subjects of economy, the environment and energy. The result is to be able to measure information flows through channels pertinent to useful questions, such as consumers/attitudes/change, or economy/risks, or business/opportunities.
Unlike most taxonomies which are designed to help users retrieve documents to be interpreted, Open Intelligence’s schemas are designed with the purpose of posing useful questions to the sources, generating indicators that directly show significant changes in coverage patterns. Trends in coverage can be reported and inferences drawn by analysts.
Open Intelligence is also developing its own Web 2.0 database application, which not only enables groups to share intuitively common facted classification schemas when collecting, analysing and synthesising source material, but also displays information flows in real time, in standard-time series graphical formats. The software is now in its alpha test phase. Until this software becomes available, Open Intelligence (www.openintelligence.wordpress.com) is using Amplify, (www.amplify.com) as a Web 2.0 repository for its current self-signifying data collection. Even in collection mode it provides a free key quotes clipping service for those interested in the subjects that it covers.
When its software becomes bullet proof, Open Intelligence aims to tap the intellectual potential of social networks to build collections of classified self-signifying data, open and free for all to see.
While both Cognitive Edge and Open Intelligence have been around in various forms for years, in the past few months a whole slew of self-signifying data providers have sprung up in the ‘Twitterverse’ with names, such as NewsTrendz and Trendrr. Although as applications they are very primitive, they are bringing the issue of self-signifying knowledge to the wider public. These software services and their implications will be reviewed in the next article on self signifying knowledge.
1. The description of the work of Cognitive Edge is based on an interpretation of a talk given by Dave Snowden to the International Society for Knowledge Organisation (ISKO UK), on April 23, 2009. Given that all interpretation is subject to degrees of misinterpretation, the sense that I made of it might not be quite the same as Dave’s. So apologies for any misinterpretation, but thanks for the sense;
2. The description of the work of Open Intelligence derives from direct experience, so is not an interpretation, but is a report;
3. Susan Sontag was one of the most important American thinkers of the secnd half of the 20th century. In a long career, she authored numerous books and articles about Western culture and modes of communication. She was always mistrustful of fashionable beliefs and once wrote that: "The most interesting ideas are heresies".
Jan Wyllie is a founding director of Open Intelligence and Trend Monitor. He is also author of one of Ark Group’s most successful reports, Taxonomies: Frameworks for Corporate Knowledge. He can be contacted at email@example.com
Sidebar: Self-signifying knowledge in action - A true story
In 2003, we published a report with the title Consumers: Going for Broke. Its sources were a representative sample of the best English language business and daily press in the UK and the US. From the analysis of the data we were collecting then, it was obvious that consumers’ balance sheets would not add up. Even then it was not just obvious, but demonstrable, that excess debt would pop the housing and spending bubbles.
Below is the consumers’ collection schema that we used to create the pertinent information flows. We wanted to draw inferences about consumers’ economic behaviour, hence the balance sheet type organisation of the schema.
Pay and conditions.
In October 2002, in the early days of the ‘Goldilocks economy’, the report inferred that:
Indicators suggest that consumers suffer from a ‘money illusion’. They have not understood the difficulty a low inflationary (or, even worse, a deflationary) environment causes for the repayment of their unprecedented level of debt;
Indicators in the INCOME/employment category suggest that jobs will be lost at an increasing rate, so personal debt defaults are liable to become a growing problem for banks and credit suppliers. These problem loans will be in addition to the already growing number of insolvencies and non-performing loans in their corporate loan portfolios; and
Although we were right on the money with our inferences indicating a credit crunch long before it happened, one of the most interesting findings at the time (under ATTITUDE/Change) was a few stories that had been classified as ‘thrift’ and ‘poverty’. There was even one instance classified as ‘profligacy’.
Now six years later, we are in the process of writing an update using a year’s worth of material from more or less the same sources and using exactly the same schema. It is hardly surprising that the consumer balance sheet looks a lot worse now than it did in 2003. However, once again, it is the ATTITUDE/Change category that holds the most interesting data.
Not only has the category grown enormously, both in absolute numbers and compared to consumer coverage as a whole, but thrift is now portrayed as a mainstream, rather than a marginal activity. Even more fascinating is that analysis has founded new information flows under ‘conscience’, ‘satiation’, ‘self-reliance’, ‘personal development’, ‘health’, ‘crime’, ‘post-consumer economics’ and (gulp) ‘revolution’.
One possible inference from this data is that the ‘recovery’ is under increasing threat from significant consumer attitude change.
Note how these kinds of self-signifying indicators can flag up important changes in behaviour long before they can be detected in conventional opinion polls, which is why they should be used before framing polling questions.