Learning from Surprising Observations in Management – Longitudinal Accounts

4.1 The Role of Owners

In 2020 the role of owners/principals came up quite often. In 1990 it was not a noticeable problem. The reason for that was that most major companies then belonged to a “bank sphere” (“Hausbank” in Germany) which served as a stabilizing moment in the strategic efforts. The ownership in a group of companies was concentrated to major banks and their supporting funds. In 2020 that role had been dissolved by the deregulation of financial markets, the expansion of pension funds, and new forms of activist fund management firms. Several respondents had experiences of corporate raiders mustering funds to acquire enough shares to demand a seat on the board and require selloffs of non-core business etc. Committee work to select new members to the board to provide for a desired variety of competences also indicated increased ambitions towards direction from the board. In 1990 the impression was that the board consisted of elderly former CEOs with considerable experience and an understanding of the need to give the CEOs space to manage the company for him to be held responsible for performance. But the freedom of action will always be related to performance as one respondent expressed it. With poor performance your freedom is reduced.

At the time of the 1990 interviews Swedish industry was also preoccupied with the effects of the legislation concerning the “Wage-Earners’ Fund” whereby large companies would be taxed, and the proceeds used to buy shares in those companies with unions getting two seats on the board. It was a quite tense political issue (“socialization” of industry). Since then, that fund has been liquidated, but today the union representatives on the board remain and respondents in 2020, to the extent that they mention it, say that it has worked quite well. Unions are better informed about the company and there have been no significant “leaks” to the press.

4.2 Information Systems

In the 1990 interviews integration of new acquisitions included installation of the corporate management control system. Some respondents stressed the point that the system represented the way to integrate, manage and report for acquisitions under the new regime – they represented the lens through which newcomers were monitored. In 2020 management control systems were hardly mentioned, and to the extent that they were, it was to note that those “integrated information systems” do not serve top management well enough. In their search for opportunities for improvement (or dealing with “surprises”) top managers deal with expectations for the future due to changes in organization and operations (not “ceteris paribus”). After all information systems deal with history (and can be used for induction at best).

4.3 “Cross-Functionality”

A recurring theme was the need to improve abilities to work “cross-functionally”. In 1990 this was not a topic, indicating that engineering competence will cover most issues. In 2020 it seemed that the problem was that even in particular “functional” departments several professions find a role – like law, IT, risk management, in what was formerly the accounting department. More competences must participate also in problem solving. Here AI is no solution (Larson 2021) since the problems top management deal with typically are “surprises” where expectations, and consequently, the beliefs on which those are built, have been falsified. Therefore, inductive analysis cannot capture the issues in historical data. And improvement propositions are, by definition, new designs, re-configuration of resources. Dialogue, and respect for other knowledge areas are required.

4.4 Closer to the Customer

In 1990 quality, in the business-to-business industries that dominated the material, was a matter of development and production of superior products to be delivered at the factory gate. In 2020 quality is to be found in the value-creation process of the customer. Different for different customers and requiring participation in the customer’s problem-solving processes. What was earlier termed “internationalization” (seeking a stronger market position by acquisitions) has become “consolidation” (partnerships, in terms of service contracts etc. with individual customer firms). This requires extensions of “supply chains”, more of local decisions, and greater complexity. One consequence is that what was earlier a matter of design, efficient production, storage, and transportation of components has now bifurcated into similar processes for sub-components that can be combined for different solutions by having compatible interfaces. Suppliers of machinery and tools have moved into the very production processes of the customer by service contracts also concerning re-cycling and certification processes.

5.1 Consequentiality

The ideology behind consequentiality (March and Olsen 1989) is that, given the premises (facts and goal criterion), deduction guarantees the truth of the conclusion. It is the form (cf syllogism) rather than the content that is in focus. We think in terms of if-then. The quality of the decision depends on the quality of the “if”. The concern in complex situation is to get the facts right and determine what decision criterion applies. The problem here is that what counts as a good reason is situation dependent. Scanlon (2014) suggests that we should be “realistic” about reasons and claims that a reason is not just a matter of facts and values – it is a matter of will where the agent is included. It is a relation (p. 31) between the agent (a), the situation (c), and the act (x). R(p, a, x, c). P is a reason for the agent a to do x in situation c. That would require a narrative (i.e. an account) to justify the action. We coded consequentiality as the orientation of an account of a learning event using the “keys” (keeping in mind that the speaker- as a responsible agent – is present in the account):

1. Justification – the “Why” question (p)

2. Construction of action (operative causality – the “How” question (x))

3. Situational demands (the situation (c) requires action now)

5.2 Appropriateness

In the logic of appropriateness, the agent is at centre stage. March and Olsen (2008) start from the famous quotes from Sophocles’ Antigone and Martin Luther “Here I stand and can do no other” to illustrate our tendency to develop rules/codes of conduct to justify action “in terms of something more than expected consequences”. It is a universal tendency in all kinds of organized human life that stems from the indeterminacy of roles, identities, rules and situations and the need to justify purposeful action. Due to this indeterminacy, actors use criteria of similarity and congruence rather than calculations of likelihoods and value. They proceed according to the institutionalized, collective practices, which will underlie atrocities as well as moral heroism. At the centre of it all stands the individual person with a “practical identity” (Korsgaard 1996) to attend to … “and can do no other”.

We (humankind) have a long tradition of deliberation on what it means to be virtuous. Aristotle (c.350 BC) pointed out that virtue always includes an appropriate balance between extremes in view of circumstances. Courage, for example, is to act somewhere between recklessness and cowardice in view of circumstances. If you do it right, you gain recognition and honour. That is what makes us social.

We follow March and Olsen (1989) and Smith (1759), by proposing that most of the time humans take reasoned action by trying to answer three elementary questions: What kind of a situation is this? What kind of a person am I? What is a person such as I supposed to do in a situation such as this? This will also be our coding orientations:

1. Situation

2. Practical Identity

3. Action in context

5.3 Scientific Management

Taylor (1911) saw inefficiency and waste in all human activity. To remedy the problem, he presented scientific management as a science and as such it constitutes a sub-group of the consequentiality logic, only that the value dimension was taken for granted – to avoid waste.

It is forward-looking and focuses on a rational choice between different options. But Taylor goes a step further and breaks these options down into tasks. Scientific management is often referred to as task-management.

The process (motion study and experimentation) draws heavily on record-keeping and measurement. Output is the main efficiency indicator (although Taylor also warns against any decrease in quality). When the optimal way to perform a task has been determined the process should be standardized, and the most suitable person to perform each task should be chosen.

In coding we look for:

1. a focus on output,

2. statements indicating a systematic search for the one most efficient way,

3. statements indicating thinking along the lines of standardization.

5.4 Sustainability

The fourth kind of logic we have chosen is the most recent one to emerge, and perhaps also the most elusive. It is “elusive” in the sense that it has come to mean different things to different people in different contexts.

The Brundtland Report (WCED 1987) is often mentioned as the place where sustainable development entered the policy arena (Bebbington 2001). That report identifies sustainable development as – “development which meets the needs of the present without compromising the ability of future generations to meet their needs” (WCED 1987, p. 8). Many of our respondents recognized the significance of sustainability, but also mentioned how it makes the job more complex. During the last decade the reports on climate change and the need to act now, not only by establishing “goals”, have had significant effects on strategic orientations (electrification of vehicles, green steel, devices to combat pollution). There is a sense of urgency, as well as increased complexity. Initially, the term sustainability related primarily to the physical environment, but now it is often seen to comprise three intertwined elements – the economic, social, and environmental (Dillard, Dujon, and Brennan 2013). In the debate questions have been asked whether the profit motif of capitalism will always supress the other concerns (Mäkelä and Laine 2011). As priorities are problematic, sustainable development has become a legitimate question for the wider managerial community (Prasad and Elmes 2005).

The gap between current practices and a sustainable way of life makes it difficult to “ground” arguments. One can question whether sustainability should be seen as a “logic” – it is more an area of concern or a state of “being”, which means there can be many different sustainable positions (Gray 2010). No doubt an element of ethics is involved in sustainable development, which cannot be simplified into defining “responsibility” as making profit for owners (Gray, Adams, and Owen 2014).

We have chosen the following “keys” for our coding of accounts given in the life story interviews:

1. Taking more than one “element” into consideration (economic, environmental, social),

2. Responsible behaviour (moral?),

3. A long-term perspective (future generations?).

It has been pointed out to us, in seminars, that these four logics are not distinct categories. “Scientific” is a subcategory of “Consequentiality” etc. Point taken! But we have worked with the premise that appropriateness and consequentiality are not crystal clear either. We view these logics as discursive, i.e. as tools of deliberation, individually or in a team, to classify arguments or justifications under these four labels to “see” the different aspects of reasoning. We expect, for instance, “scientific” to appear inside an engineering culture, as well as the portfolio choice arguments inside a finance culture.

5.5 Coding Results

This kind of coding is not very precise. We tried to improve precision by each of us first coding individually, and then resolve differences by discussion. Still uncertainties remain. For example, an account could include a particular situation, the solution of which was then summarized into a general management principle like this one by a CEO:

They did not dare come and report that this is a problem, as it were! So, you need to create a culture where mistakes are put on the table as a source for improvement. Of course, you cannot say “Excellent! We have made a mistake!” But we can say “Good that you reported this! How do we solve this? What can we learn for the future?” Obviously, that is easier said than done! I can get raving mad and irritated, but you must live by the principle ‘you should never go to bed with hard feelings’ and reach an agreement with the guy about what we can learn from this. (Coded under “appropriateness”)

Or the principle of pushing for stronger efforts in product development by a board member:

I have said that the greatest challenge is that if we want to expand (this company) we cannot do it within this particular business area. You cannot grow above 50 % (of the world market) by acquisitions! We must find new areas. So, I push hard for new ideas – innovations that can be turned into products that make good use of our core competence (using XXX for doing YYY). (Coded under “consequentiality”)

Coding for statements that can be classified under the four chosen logics (and their sub-dimensions) yields the two tables (Table 1) below with the identities of the respondents in the left column (“interview”). The variety in the use of arguments is quite large. The fact that the 1990 interviews generate a higher total is mostly due to these interviews being longer (on average). Extremes being P2 in the 1990 series, which lasted 3 h, and Q1 in 2020, which was done over a quick lunch on a strict time limit. A few interviews in 2020 were done by phone.

The variety can partly be understood against the background of the specific line of business of the company, and the education of the respondent. The number of engineers (often with an advanced degree) in the first round was 12 out of 22 (seven with a business or law degree), and in 2020 it was seven out of 19 (11 with a business degree). One should not attach too much significance to these figures since manufacturing industry dominated in the first round while the composition of industries was more varied in the second.

There were three women in the 2020 interviews – only men in 1990. The distribution of statement logics does not seem to differ between sexes, except that there is a somewhat higher concern about sustainability for women.

On both sets of interviews the reason why Swedish firms are doing so well internationally (coming from such a small economy) was an early topic. The core argument remained unchanged; Swedish managers’ willingness to listen and respect the opinion of the other. “Respect the other” being a more advanced form of individualism than sheer egoism.

Number of learning events: 1990 = 279; 2020 = 405.

It is curious that the number of coded “logics-in-use” was much higher in 1990 in spite of the fact that the number of “events” (or accounts) was much higher in 2020. We think that this observation is somehow related to the changes in context between the two sets of interviews discussed earlier, especially the increased complexity and tempo facing managers.

The overall impression when surveying the frequencies of the two series of interviews is that there is a more “even” distribution across the logics in 2020. The variety across individuals was, however, considerable.

Sustainability was hardly mentioned in the early 90ies (it was not a widely recognized concept yet), but has gathered momentum, albeit somewhat unevenly, in the 2020ies. The managerial problems with sustainability seem to be that there is a delay between action and effects, and that the general goals, which everybody accepts, must be translated into concrete actionable proposals – which, usually, changes the focus of the discussion into, for example, an engineering problem. The former problem can be illustrated by one of the respondents telling the story about an open pit mine that is scheduled to be closed in 2046. The company put a lot of effort into designing a solution that would provide the local population with an attractive (large, and deep) lake with a surrounding hiking area. Local councillors were delighted. The question came up when the lake will be waterfilled to serve its future purpose (tourism). The answer from the environmental manager was “about 200 years”. What if we lead the neighbouring stream into the lake? 150 years.

The frequency of statements that can be classified as belonging to the “scientific management” logic has increased over the 30 years. This could possibly be related to the prevalence of arguments about “share-holder value” and “finance speak” these days. Discourse on managerial conduct cannot avoid this.

Appropriateness and Consequentiality dominate in both series, but the order between them has shifted from “Appropriateness” in 1990 (646 larger than 453) to “Consequentiality” in 2020 (259 larger than 229) (Table 2).

Going through the “keys” used for the two logics we find a noticeable shift for Appropriateness from “Practical Identity” being the most frequent in 1990 to “Action in Context” in 2020. For the Consequentiality the proportions between keys are virtually unchanged.

Caution: as mentioned, the coding relies of a searching for the “meaning” of an account by examination it in context (Cooren 2000). This means that there is no “truth” to be found in an account, only “discursive” intent. The speaker wants to “make a point”, and the coder wants to “make sense” of it.

6.1 Comparing Two Generations of Managers

Both Adam and Bert earned their reputation as successful managers in a sequence of “trouble shooting” and improvement projects that required heightened attention over time. Their accounts often start with being thrown into an unfamiliar situation or a “discovery” of an opportunity for improvement. The arguments they used to explain and justify action spread over logics, with Adam leaning toward finding new, appropriate solutions for production systems, while Bert, being an early computer nerd, describes initiatives and new designs like establish a service division to improve the value-in-use of (in site) of products by closer cooperation with customers. Almost as an afterthought there is, in most accounts of the projects, a statement like “and we had to re-structure the organization accordingly”. This shift in attention brings with it an awareness of the presence of several other interests that must be catered for. Bert, it seems, must be ready to use a more varied portfolio of arguments to pilot his projects to fruition. The personalities of Adam and Bert are different, as are the contexts in which they operate, but in both cases communication across knowledge areas (cross-functional, cross-cultural, between organizational levels) seems to be a key tool.

6.2 Searching for an “Inference to the Best Explanation”

The more evenly distributed use of logics in the 2020 interviews combined with the greater frequency of self-contained, and therefore more “unique”, accounts of learning events pointed us in the direction of searching for explanations in the singular accounts rather than in some “general” factors. However, there seemed to be a disturbing ambiguity in accounts. Looking closer at the justifications used in them we noted that there were very few references to the dominating modes of drawing conclusions in science – deduction and induction. Instead, we found frequent examples of “abduction” or abductive reasoning (Peirce’s concept (), see Brent 1993). This was a central concept in the times of “pragmatism” around the turn of the previous century. James (1907) indicated the significance of abduction in pragmatism by stating that ‘If you, after having hesitated, are ready to act on this information,’ it is “true” in a pragmatic sense. Larson (2021) points out that abduction starts from an observation – a “surprise,” or a deviation from the normal – which initiates a search of your mind (“world knowledge”) for an explanation (“sensemaking” – which AI cannot do). Misak (2016) provides the history of pragmatism in Cambridge and the core of abductive reasoning:

“The surprising fact, C, is observed.

But if A were true, C would be a matter of course.

Hence there is reason to suspect that A is true.”

(Misak 2016, p. 41)

This was intriguing since the explanatory power of the accounts by these senior managers came from the “meaning” the explanation (A) gave to the surprise (C). Time is short, relevance questionable, and expertise knowledge come from several knowledge areas. “Associative coherence” (Kahneman 2011) will have to do. Abductive inference being more akin to how human intelligence works (i.e. in “leaps” and “guesses” involving what we already know, and what we observe, to update our prior beliefs). This differs from inductive inference (i.e. acquiring knowledge from experience/historical data), or deductive inference (which builds on what is already there in premises, and does not add new knowledge), which are more “calculative” (Larson 2021).

Re-examining all the 2020 accounts for indicators of abduction (relating “observation” to “explanation”) we found that about half (out of the 405) could be classified as abductive reasoning (Obviously this classification is spurious and might be influenced by our enthusiasm over this discovery.) The 1990 accounts gave similar proportions (out of the 279 accounts). Still, the question is whether this use of abduction is stimulated by current discursive practices in Swedish industry, or if the current stage of development of a certain firm drives it. Comparing Adam and Bert (in the same company, above) gives the impression that, certainly, the problems were somewhat different, but in most cases, it was a matter of “understanding” a new situation, working out (with others) a possible solution, and implementing it (through others). In 1990 it was more often engineers talking about engineering and automation. Context plays a role.

Abductive reasoning (c.f. “Clinical reasoning” in medicine (Higgs et al. 2018; Yazdani and Hoseini Abardeh 2019)) should be understood as a process with loops with an intention to construct coherence (Kahneman 2011) that justifies action. As such it serves as a bridge between thought and action, which is essential in management (and for doctors). Arriving at a conclusion on “the state of affairs” as “associatively coherent” (Kahneman 2011) is a constructivist undertaking. Hacking (2000) warns us about the risks in social construction, but managers and doctors are primarily concerned with the well-being of their protégé, and about action now. Misak (2016) provides an account of how the early pragmatists promoted “clarity of thought” and “degree of belief” (a concept Keynes (1921) discussed) as criteria in early parts of scientific inquiry. Popper (1963) would call it “conjectures”. March (1991) discusses the mindset of exploration (open) and exploitation (closed). The early pragmatists claimed that inquiry is driven by doubt and stressed “clarity of thought” that Misak (2016) list as stemming from:

1. familiarity and habitual use

2. clarity of the parts of the definition of an object

3. clarity of the object’s practical implications

Admittedly, coding the learning events of life stories for logics-in-use did not help much, but in combination with analysing accounts as narratives it led us to consider accounts as applications of abductive reasoning.

The problem when considering abductive reasoning as a communicative tool in teams finding ‘the best explanation’ to a “surprising observation” is that words always come with an interpretation. This was a central topic in Cambridge at the time pragmatism arrived. Ramsey opposed Wittgenstein’s (1922) conception of philosophical propositions as unambiguous (ready to enter the mind for analysis). He was charged with translating Wittgenstein’s Tractatus (1922) and claimed that the words always “get in the way” of common understanding (Misak 2020). Wittgenstein resisted, but finally gave him right in Philosophical Investigations (1953), which gave rise to the “linguistic turn” in social sciences with new concepts like “language games” and “forms of life”.

When the words “get in the way”, incommensurability is a consequence which Kuhn (2022) discussed in his last writings (with inspiration from the pragmatists?). Science wars (cf Hacking 2000) may be stimulating in academia, but managers must do something about it. Mumford (2000) reviews the literature on creativity in teams and points to the role of leaders.

6.3 The Abductive Process

Searching for an “inference to the Best Explanation” (McCain and Poston 2017) to why abduction seems to lay behind so many accounts of learning in our interviews we return to the credo of pragmatism given by Misak (2016) (above).

1. The process starts with a fact (surprising observation) that deviates from what was considered “normal”. Boltanski (2011, p. 68ff) describes it as “worth a second look” – it needs to be explained since it violates expectations, and, possibly, the beliefs behind those expectations (What is going on here?). To register such an observation as “unnormal” you must have seen many examples of “normal”. But you also need to check whether the observation is a “fact” (truth test according to Boltanski and Thévenot (2006)) and whether it is relevant (reality test). If the “surprising observation” is indeed a relevant fact, uncertainty will be at maximum.

2. The next step, the search for an inference to a Best Explanation must, by virtue of the uncertain state of affairs, be based on Kahneman’s (2011) “fast thinking” (associative coherence) and include values as well as operational facts (Weber’s (1922/2019) “Wertrationalität” as well as “Zweckrationalität”) and can be expected to be recursive (including using simple cases as basis for definitions). It can be understood as building a scenario in which the observed surprise makes sense. It also provides space (“framing”, Goffman 1974) for other members of the team with different knowledge bases to act rationally in this particular situation.

3. Persuading others to go into action based on an explanation of ‘one observation’ is not easy since acceptance of the surprising observation as factual renders the situation complex, as well as uncertain. Engagement with the proposed explanation will be driven by doubt, curiosity and/or hope. So, it must be “communicated” to others, with different expertise, to make them able to “see” the point and how they may contribute (which requires “keys” (Goffman 1974) to re-framing). Snow and Benford (1988), taking inspiration from Goffman, argue, in a similar vein, that in a process of mobilization, leaders act as “signifying agents”, where the signifying work involves assigning meaning to relevant events by framing (“motivational framing”), and by aligning frames.

These three dimensions – noticing, explaining an observation, and persuading others to act – constitute a complexity inherent in the accounts given in our interviews. The challenge for these leaders was to switch from an (open-minded) exploration mode to an (focused) exploitation one (March 1991), not only in their own mind, but for the whole organization. Coding on single logics does not seem to clarify matters (much), even if shifts in logics indicate shifts in intentionality (persuading others). It seems like the complexity itself gets in the way of “clear” thinking (at least for us!). However, the accounts we have studied seem to be able to accommodate considerable complexity, while at the same time allowing the speaker to articulate her or his own “practical identity” (Korsgaard 1996).

The picture that emerges from the 684 accounts of learning experiences by prominent leaders in Swedish industry is confusing (in the sense of the pragmatists (“a surprising observation has been made”)). Our inquiry, driven by doubt, took us, first, to the observation that the logics (justifications) in the accounts tended to shift inside the same account. When we took a step back and looked at the account as a description of a process of deliberation, we chose to focus on the narrative itself. Considering what it means to learn from one unique and complex event, as the accounts indicate, it cannot be “the truth” in some well-defined sense, but “meaning” that is constructed. A meaningful narrative about a situation, and the way out of it, may be a basis for hope, and hence a readiness to go into action. Since the narrative is the basis for hope – the situation is characterized by uncertainty – the success of action depends on the adequacy of the operational parts of the emerging action plan. Hope provides for the attribution of success to ego, while operational adequacy secures success in context. There are two logics here! The “practical identity” of the actor, and the operational conduct of actors in context. March and Olsen (2008) summarize this in the logic of appropriateness (“what should a person like me do in a situation like this”), which, by definition, engages “me”^[3] as well as “a situation like this”. The “situation” requires knowledge about causes and effects in the real world and about how chains of action connect. Engaging oneself in causing effects in the world requires switching between perspectives (inside-out/outside-in) and, consequently, shifts in logics-in-use. This many-sidedness can only be captured in narrative form. The researcher expecting one logic in rational action will be surprised.

The crucial steps in the “real-time” processes of deliberation are the shifts from “thought” to “action”, and from “Outside-” to “Inside-” orientation. That process may be dis-orderly in many ways (recursive), but when members of the team approach “reflective equilibrium” (to use Scanlon’s (2014) term), they have a basis for constructing an “account” in a narrative form that appeals to different audiences, with different knowledge bases, inside the organization. When the interviewee gives an account of a learning event to a curious researcher the narrative will be provided with keys that are deemed suitable in that context. One example is a respondent who claimed that he preferred “Management by Means” (? – he knew that the academic will be familiar with MBO (Management by Objectives)) and argued that one discusses the problem thoroughly with the project team, develops an indication in what direction the solution might be found, provides the team with the required resources, and gives them the go ahead. His point was that it is easy to follow up in such a setting; you just pick up the phone and ask, “How are you doing?” (The report will be geared to the earlier provisional narrative, and easy to grasp.)