Noise, Nudge, and Sludge – These Nobel prize winners want you to change how you manage

Daniel Kahneman and Richard Thaler are both Nobel Prize winners in Economic Sciences1 and they want you to change the way you manage. This is because there are two things afflicting your organisation that you don’t realise are costing you dearly – in their words these are ‘Noise’ and ‘Sludge’. 

‘Noise’ is Kahneman’s term, and it refers to all those extraneous things that can affect human judgement but have no logical bearing on the issue at hand. As he puts it, “The problem is that humans are unreliable decision makers; their judgments are strongly influenced by irrelevant factors, such as their current mood, the time since their last meal, and the weather. We call the chance variability of judgments ‘noise’. It is an invisible tax on the bottom line of many companies.”2 

He explains that noise is distinct from bias, which exists more as a permanent filter on judgements rather than a random factor. The way noise taxes the bottom line is that it causes significant variations in decisions. Imagine that you are a customer and you inadvertently request estimates from two employees in the same company for the same thing. They then reply with substantially different estimates. This actually happened according to Kahneman, and of course the horrified customer decided that the best alternative was a competitor. 

‘Sludge’, on other hand, is a somewhat different but equally pernicious hidden influence. Sludge is Thaler’s word for any aspect of your organisational environment that hinders a person from taking the better course of action, or it may even actively steer them into a poor choice. 

The overall circumstance in which a person makes a choice is called the “decision architecture”, and it is something that can be shaped for good, or for bad, by design or default. Sludge is any part of the architecture that fails to help with good choices or actively steers people toward bad ones. In Thaler’s words, “Sludge … is nasty stuff that makes it more difficult to make wise choices”3

The solution, of course, is to remove both sludge and noise from decision making. But how? In the case of sludge, the answer is to replace it with ‘nudge’. A nudge is a change to the decision architecture that helps a person to act in the best way and make better choices. Nudges often involve rethinking processes to simplify them, personalise them, or make them easier and more efficient, or all of these things. And nudging happens automatically when decision makers are provided with better knowledge and understanding of the intricacies and implications of the tasks they face. 

Reducing noise requires a similar approach, with the main aim being to achieve accurate and consistent professional judgements. Kahneman states that research shows that noise typically causes variations in professional assessments of up to 50% (with instances as high as 70%). Removing noise reduces this to a much more tolerable, and less damaging, 5-10%. Part of his solution is to “adopt procedures that promote consistency by ensuring that employees in the same role use similar methods to seek information, integrate it into a view of the case, and translate that view into a decision.” 

So, for any given task or issue, the way forward is for those involved to: 

  1. Rethink it in nudge terms 
  2. Agree consistent methods, so as to eliminate noise
  3. Subscribe and commit to those methods, and share the associated knowledge
  4. Jointly monitor outcomes and variability
  5. Based on outcomes, go back to 1 and repeat

This five-step approach builds consistency and capability. It also lays solid groundwork for developing effective algorithms for automated methods, which are Kahneman’s preferred ultimate solutions. He states, “Replacing human decisions with an algorithm should be considered whenever professional judgments are noisy, but in most cases this solution will be too radical or simply impractical.” In support of this he argues that “People have competed against algorithms in several hundred contests of accuracy over the past 60 years, in tasks ranging from predicting the life expectancy of cancer patients to predicting the success of graduate students. Algorithms were more accurate than human professionals in about half the studies, and approximately tied with the humans in the others. The ties should also count as victories for the algorithms, which are more cost effective.” 

But the caveats on automated algorithms are serious. They must be developed and implemented with extreme care, and never without close human oversight and gatekeeping. Inaccurate assumptions, flawed methods, inadequate input data, and general haste to jump to a “silver bullet” solution, (especially when highly paid external consultants are involved) can be absolutely disastrous – the most egregious examples being Australia’s Robodebt debacle4 and The Netherland’s welfare fraud and child benefits scandals5 (the latter brought down the government). 

In short, noise and sludge are real, and nudges and noise reduction are proven paths to improved performance. But achieving this is not easy. Fundamentally, it requires commitment, knowledge sharing, and cooperation. And algorithms will be most effective, and least likely to be harmful, when they are an extension of these things and always kept transparently within their control. 

___________ 

1 Kahneman in 2002 and Thaler in 2017 

2 “Noise: How to Overcome the High, Hidden Cost of Inconsistent Decision Making”, by Daniel Kahneman, Andrew M. Rosenfield, Linnea Gandhi, and Tom Blaser, Harvard Business Review, October 2016 

3 ”Nudge: The Final Edition”, by Richard H Thaler and Cass R Sunstein, Penguin Books, 2021 

4The Robodebt fiasco has been laid bare at royal commission hearings”, ABC News, 4 March 2023 

5This algorithm could ruin your life”, by Matt Burgess, Evaline Schot, And Gabriel Geiger, WIRED Magazine, 6 March 2023  

___________ 

Onno van Ewyk is a Knowledge Management Consultant and author of the book Raising an Organisation’s Collective IQ

CEOs! An acronym you have never heard of could save you millions

Have you ever heard of DITA? Outside a small community of specialist technical communicators, most people have never heard of this acronym. Which is a shame because, when applied in the right way, it can save an organisation millions of dollars. How? By significantly improving the way organisational knowledge is captured and communicated. It makes knowledge capture faster and easier, and it produces an end result that is both easier and quicker for people to assimilate and apply. Employees work more effectively, and new appointments get up to speed much faster. 

DITA stands for Darwin Information Typing Architecture, which is not exactly self-explanatory, but its cleverness becomes obvious when you consider its output – called a ‘topic’. A topic is “a title and content short enough to be specific to a single subject or answer a single question, but long enough to make sense on its own and be authored as a unit”i. In other words, a compact packet of useful information. 

So how will this save millions? The simple answer is by replacing policy, procedure, and instruction documents. These classically rigid and bureaucratic documents, often written more with an eye to lawyers and regulators than employees, can hinder more than they help. Their orientation often tangles the business up unnecessarily in red tape and complicates issues in ways that are counterproductive to their intent. Policies end up as just words on a page with little impact on behaviours and culture, and useful instructions are lost in overblown paragraphs of passive prose. 

Topics, on the other hand, orientate policies, procedures, and instructions to the employee’s needs. Short, sharp, and to the point (single subject, answer to a single question), they are not only far more effective but also easier to maintain and keep up to date. 

Moreover, topics are created and maintained within an ecosystem called a DITA Map. This is a nested hierarchy of categories that are aligned with the business. They ensure that topics are relevant, comprehensive (i.e., covering all important processes), and easily located. The DITA Map provides an overall framework in which topics are related to each other. It creates a knowledge map that employees can use to explore knowledge resources productively, rather than having to blunder through the usual ad-hoc folksonomies found on file servers and Intranets. 

The DITA approach is Darwinian in that units of information (topics) live and die within their use-environment according to their practical value. Topics are also adapted to a particular overall context (your business) and are potentially each related to one another (implemented via contextual hyperlinks). 

‘Information Typing’ means that DITA topics can come in various flavours, the key ones being ‘Concept’, ‘Task’, and ‘Reference’. You can think of policies as being ‘concepts’, procedures and instructions as ‘tasks’, and everything else as ‘references’, but there is no need to get too hung up on these distinctions. Being this fine-grained is only useful when working with detailed technical information that is published in multiple formats from a single source. In a general business context, the greatest utility comes from restricting topics to a single subject or answer to a single question. 

The A for Architecture in DITA, however, is critical. Architecture means not only a clear structure in which policy, procedure, and instructional information makes sense and supports your value chains, but also a commitment to quickly changing that structure as business needs change. In fact, the A in DITA could quite comfortably stand for Adaptation – and what could be more Darwinian than that.

Onno van Ewyk is a Knowledge Management Consultant and the author of the book Raising an Organisation’s Collective IQ

Doug Engelbart and Collective IQ

The great digital pioneer Dr Douglas Engelbart coined the term ‘Collective IQ’*. He identified five key concepts behind raising Collective IQ – Networked Improvement Community (NIC), Dynamic Knowledge Repository (DKR), Concurrent Development Integration and Application of Knowledge (CoDIAK), Open Hyperdocument System (OpHys), and Bootstrap. In his vision, the application of these concepts enhances society’s ability to solve complex, large-scale problems. In an organisation, the value of these concepts is their potential to enhance the organisation’s capabilities to deliver quality products and services to customers, to support innovation and adaptation, and to enrich employees’ work experience.

Networked Improvement Community (NIC)

Engelbart’s concept of a Networked Improvement Community or NIC is one in which a group of electronically connected people with a common purpose use technology cooperatively to tackle the complex problems that they jointly face. They work together to better understand their problem-space, to unearth the best candidate solutions, to decide how best to deploy operational resources and capabilities, to monitor progress, and to adapt effectively to unforeseen complications.
For organisations, the NIC concept is an evolution of the ‘team’. It raises a team’s profile from one in which the members are focused on cooperatively performing a fixed set of tasks to one in which they are also focused on sustaining and improving both how the team works and its outcomes.

Dynamic Knowledge Repository (DKR)

A Dynamic Knowledge Repository or DKR is an online location or ‘container’ in which knowledge is stored, utilised, and developed through direct user interaction. By ‘knowledge’ here is meant codified knowledge, that is, information held in discrete, human-viewable, files that contain text, pictures, diagrams, illustrations, sound recordings, video recordings, etc., or some combination of these.
It is dynamic in the sense that its content is constantly updated by the members of a NIC, who ensure that at all times it is comprehensive, accurate, coherent, authorised, accessible in multiple ways as to the purposes it serves, and linked and marked up in ways that make it explorable in meaningful and useful ways.

CoDIAK

CoDIAK stands for Concurrent Development Integration and Application of Knowledge. It is the concept that you do not ‘stop in order to learn’. The processes used to improve capability should be integrated with, or even identical to, those used to support current action. This has the double benefit of fuelling learning via immediate feedback, and quickly implementing newly devised improvements. DKRs, when properly implemented, are key mechanisms for integrating and applying new knowledge. They also provide the base on which new knowledge can be developed.

Open HyperDocument System

Open Hyperdocument System or OpHys is an ambitious technical challenge set by Doug Engelbart to create a new electronic document format and associated application systems for recording, sharing, retrieving and linking information in a much more effective and collaborative way.
As a general rule, enterprises at the moment tend to manage the documents that contain their corporate knowledge in a fragmented and ad-hoc way. In addition, existing technologies, which are inexpensive and ubiquitous, tend to be underutilised. This is mainly because of a continuing ‘paper bias’ in the way documents are stored and managed.

Bootstrap

Bootstrap or Bootstrapping is the notion that raising Collective IQ potentially operates on three levels, with the upper two levels aimed at boosting or accelerating learning capacity. On the first level (level A), raising Collective IQ increases the capacity to solve specific real-world problems and carry out complex tasks. For example, the problem or task might be how to design and produce a new model car.
The second level (level B) focuses on improving the methods and processes used to improve level A processes. To continue our example, this might involve looking at the way new model cars are designed and produced in general to come up with a better overall approach.
The third level (Level C) focuses on methods and processes used to improve Level B processes. In our example, this might involve improving the way manufactured goods in general, not just cars, are designed and produced.
These levels have implications for the nature of the NICs involved. For our examples above, level A would be addressed by a NIC comprising people working in a motor vehicle company’s design and engineering function. Level B, on the other hand, would require a NIC with representatives from all the motor vehicle company’s key functions (manufacturing, inventory, finance, sales, etc.). For Level C, a NIC involving people from multiple manufacturing companies, academia, and professional societies, not just motor vehicle manufacturers, would be required.
The latter also highlights one of the key aspects of Collective IQ in Engelbart’s view and that is to address complex problems at a societal level that transcend the interests of individual organisations. In the case of our example, working together on improving general new model design and production methodologies benefits all manufacturing companies without impacting their relative competitiveness.
In Engelbart’s words, the essence of bootstrap is “The better we get at getting better, the better and faster we’ll get better.”


*
Engelbart, Douglas C (1995) “’Toward augmenting the human intellect and boosting our Collective IQ” Doug Engelbart Institute (http://dougengelbart.org/pubs/books/augment-133150.pdf)

Leveraging the value chain with knowledge

When a person looks for knowledge to support their work, they do it in response to their circumstances and the problems or difficulties facing them. This means that knowledge needs to respond to questions such as, “What do I do if (a certain situation arises)”, “What do I do when (a certain event occurs)”, “How do I (complete a certain task)”, “How can I understand how (a certain set of activities inter-relate)”. Consequently, knowledge needs to be classified and presented in a way that provides direct answers to these sorts of questions.
The common characteristic of these ‘knowledge-seeking’ questions is that they relate to people’s activity. They relate to what they are trying to achieve in their jobs and, by extension, for the organisation. The most useful basis then on which to classify knowledge is around the way the enterprise is structured to achieve its purpose – in other words, its ‘business model’.
This begs the question as to whether the organisation has clearly identified and articulated what its ‘business model’ is. It needs to do this first before it can provide a useful context in which to develop and utilise knowledge.
The essential theoretical groundwork for this task was thoroughly prepared by Michael E Porter with his concept of a “value chain”*. Porter’s analysis was focussed on the means by which organisations secure competitive advantage. We can apply the concept to managing the knowledge that supports, maintains, and improves this competitive advantage. By applying this value chain model as the conceptual context for developing and utilising organisational knowledge it is possible to link knowledge directly to the task of maximising value. This approach has the additional advantages that the value chain’s primary and support function categories:

  • define the areas of the organisation that need to be covered by knowledge
  • align with the way many organisations naturally divide their activity into specialised functions – this allows employees to locate relevant knowledge through their familiarity with its work context
  • align broadly with the way management responsibility is allocated within an organisation – this makes it easier to identify who needs to take responsibility for curating the knowledge concerned

Porter’s value chain provides an effective overall context in which to develop and utilise knowledge. It provides the foundation for a dynamic taxonomy for organisational knowledge that ties directly to the organisation’s purpose. As a component of Collective IQ, this is the framework in which ‘atomic units of knowledge’ can be both developed and deployed.


*
Porter, Michael E (1985) “’Competitive advantage: Creating and sustaining superior performance’, The Free Press

An atomic unit of knowledge

The idea of knowledge packaged as an ‘atomic unit’ was first suggested by Michael H. Zack*, who stated “… knowledge-as-object becomes knowledge-as-process. The basic structural element is the knowledge unit, a formally defined, atomic packet of knowledge content that can be labelled, indexed, stored, retrieved and manipulated.”
This ‘atomic unit of information’ needs a name or label to support its use as a fresh form of information currency. For this, organisations can leverage a concept developed and refined by the technical communication profession. As part of their transition to increasing digital distribution and management of technical information, they coined the term ‘topic’, which is defined as “a title and content that is short enough to be specific to a single subject or answer a single question, but long enough to make sense on its own and be authored as a unit”. Qualifying this concept to cover just information about organisational activity, gives us the term “Business Information Topic” or BIT.
Replacing policies, procedures, and work instructions with BITs not only breaks down out-dated cultural influences, but knowledge resources can be built so that they are much better suited to digital communication and interaction.
Individual managers can begin to raise the Collective IQ of their teams immediately simply by abandoning conventional approaches to writing policies and procedures and, instead, use a topic-based approach. As a manager, this means:

  • changing your mindset from ‘command-and-control’ to ‘knowledge sharing’
  • writing to address specific tasks or circumstances and how best to handle them (by combining all the relevant elements of policy, procedure, and instruction)
  • employing a writing style that is easy to understand and engage with (this means using the active voice and other ‘plain English’ techniques, as well as following the Behavioural Insights* precepts of ‘simplify’ and ‘personalise’)
  • developing the content collaboratively with team members
  • enabling and actively encouraging team members to provide feedback, criticism, and improvement ideas
  • using the documented information as a basis for common understanding, coordinated action, and collective improvement

Making these changes immediately improves the capacity for written information to bind and focus a team.


*
Zack, M. (1999) “Managing Codified Knowledge” Sloan Management Review, Volume 40, Number 4, Summer, 1999, pp. 45-58

Nudging the world toward smarter public policy: An interview with Richard Thaler McKinsey Quarterly June 2011

The problem with policies and procedures

The problem with policy and procedure documents is that they are an anachronism. They belong to a pre-digital world in which the prevailing ethos was ‘command and control’.
The concepts of ‘policy’ and ‘procedure’ as discrete documents are relics of a past world of exclusively paper-published documents. They were bundled into ring-binders and updated sporadically. They were often widely ignored because they were difficult to access (because of limited numbers of copies) and often out of date because it was cumbersome and logistically difficult to update them.
The language typically used was also problematic. There was a general belief that what was written down and published should sound important and ‘official’, but this often resulted in a tone that was bureaucratic, stiff, and remote. Readers struggled to connect with it, and to clearly understand what it meant for them and their work.
The networked digital environment that now prevails has the potential to overcome these shortcomings, but organisations fail to take advantage. Documents that were once distributed on paper are simply converted into PDFs, with the same use of bureaucratic language, the same ineffective writing style, and the same inflexibility. Here are the tell-tale signs of this anachronistic approach:

  • PDFs with little or no use of hypertext links (usually an indication that content is developed in isolation and published sporadically)
  • scanned hand-written signatures (despite almost every employee now having an online identity that is much harder to fake than a scanned signature)
  • ‘document control’ tables that are manually updated in the document (despite the ready availability of content control systems that include version histories)
  • file names that contain version numbers (again, despite the ready availability of content control systems that automatically assign version numbers, which can be made to appear in the document and update automatically)
  • documents that begin with a separate title page that must be scrolled out of the way to get to the content
  • documents with standardised headings, which, under the pressure of filling an otherwise blank space, result in content that is repetitive, generic, and, in many cases, redundant
  • long documents that are subject based rather than short ones focused on users’ needs and circumstances
  • documents in Intranets grouped under the headings of ‘Policies’ and ‘Procedures’, instead of the categories of organisational activity to which they relate

The starting point for addressing these problems and taking advantage of networked digital environments is to replace policies and procedures with Business Information Topics. These are documents that combine policy and procedure information such that they are:

  • user-focused and easy to access
  • confined to a specific issue, decision-point, or task (combining, where applicable, elements of policy, procedure, instruction, and explanation)
  • written in a way that is concise, direct, and straightforward (so that it is quickly assimilated to support action or decision-making)
  • part of a connected network of inter-related units of information that can be explored in one direction to gain broader understanding, or in the other to focus in on greater detail, depending on the links followed (that is, it should be ‘atomic’ in the sense that is a building block for more complex information structures)