Towards a Processual Understanding of Task Complexity: Constructing task complexity in practice

Noticing primary cues

While monitoring the post-surgery process (path 1), Klaus, the physician on duty, read Fritz’s file and saw an antibiotic that was prescribed for an infection. Klaus could not find an explanation for this, so he collected more information from other departments (path 2; see Figure 2). After taking care of this issue, the nurse on duty during the night shift noticed that the circulatory monitoring alarm system came on; Marie, an assistant physician, treated this with medication (path 4). A few days later, a nurse asked for help when Fritz had problems breathing and had a panic attack. Mechthild, the physician on duty and two assistant physicians supported the nurse and gave Fritz medication; Mechthild then conducted a bronchoscopy (path 10). That same day, a nurse noticed that Fritz had a skin rash and insisted on a dermatologist consult. Mechthild decided against the consult but changed Fritz’s diet (path 11). The next day, a nurse noticed that Fritz was saying confusing things to his wife and was acting aggressively. She immediately asked Fred, an assistant physician for help. Fred involved a physician who provided a medication plan (path 12).

Noticing secondary cues

While treating Fritz after surgery (path 1), a nurse who monitored Fritz noticed a high blood loss. Klaus, the involved physician instructed to order blood reserves (path 3; see Figure 2). During the night Marie, an assistant physician, checked Fritz’s values as a normal post-surgery monitoring step (path 1) when she noticed that the blood values were pointing to kidney failure. She consulted the nephrology department, which advised conducting a dialysis and an ultrasound (path 5). While treating the possible kidney failure (path 5), Marie checked Fritz’s blood values again and noticed a lack of white blood cells, which indicates a higher infection risk. Thus, the physicians transferred Fritz to a single room (path 6). Later that day, the physicians tried to wake Fritz up (path 1). While waiting to see if the trial was successful, a nurse noticed that Fritz was having seizures; these were treated with medication (path 7). Because the physicians could not explain why Fritz was not fully waking up (path 1) and had circulatory problems (path 4), they decided to conduct a CT scan. This CT scan did not provide new information regarding the existing health issues; however, the radiologist noticed a pneumothorax as a new health issue (path 8). As before, the physicians were monitoring Fritz’s blood values intensively (path 4), whereby they noticed that Fritz had growing infection values; this was treated by administering antibiotics (path 9).

Fritz’s treatment shows how new paths formed because hospital staff monitored the patients closely, which helped them to notice cues pointing to a new context. When nurses or physicians took actions in the direction of these cues, which did not belong to any previous path, new paths formed. Physicians are quite aware that patients can exhibit new symptoms all the time, as one physician explained:

A lot can happen, all the time. It is possible that a patient feels all right in the morning, and in the afternoon, he all of a sudden has several health issues.

The example of Fritz shows how several paths emerged throughout the task patient treatment. As per the definition, each new path meant that the complexity of the task changed (increased) in the process of treatment. The more paths that the physicians had to treat, the more complex the task was. As new paths emerged during the process, no one could have determined the number of paths, and therefore the complexity of the task, beforehand. Only working through the task revealed the number of paths and the involved complexity.

Questioning working diagnosis

Soon after Fritz was transferred to the ICU, Marie, an assistant physician, noticed that his blood values were pointing to possible kidney failure (path 5). As Fritz had a new kidney and the team in the ICU did not specialize in this field, Marie consulted the nephrology department. This team advised conducting a dialysis and an ultrasound. The ultrasound was analysed by Monika, a nephrologist who cooperated with the ICU team and worked with them on the entire case. She confirmed the first working diagnosis, that it was a temporary post-surgery problem and that it was best to wait and monitor how the kidney developed. The next day, Monika asked for another ultrasound to verify whether the diagnosis still worked; it was confirmed by the results. The following night, the physician on the night shift noticed growing lactate values, which can signal a lack of oxygen in the tissue – a sign of kidney failure. Thus, he questioned the current working diagnosis of a temporary failure and consulted the nephrology department. The respective physician from that department ordered an ultrasound, which again reconfirmed the current working diagnosis. However, during the morning rounds, Gerd, the chief physician, attended and questioned the working diagnosis: ‘Do we really need to give tacrolimus? Immunosuppression does not make the patient any healthier. Please ask the nephrologists again how they forecast the development of the kidney. I’d like to critically question whether that is still necessary.’ Thereupon, Tim, the responsible physician, and Monika discussed the situation again and conducted a new ultrasound. Monika also involved the physicians who transplanted the kidney. They all confirmed that it was still a temporary problem. A few days later, Monika asked for another ultrasound due to a bad blood gas analysis. The radiologist conducting the ultrasound reported that there was poor blood flow to the kidney; however, he thought it was due to Fritz’s medication. On the following morning rounds, Gerd, the chief physician was present and questioned this diagnosis again. Even though the physicians did not agree whether the kidney failure was only temporary or permanent, they agreed to transfer the patient to the nephrology department.

Searching for working diagnosis

Almost from the beginning, Fritz had circulatory problems. The physicians treated these symptoms with medication and agreed on the working diagnosis of a temporary post-surgery complication. As cardiac arrhythmia also occurred after the wake-up trial, the physicians questioned this diagnosis and conducted an ultrasound to test if it was heart disease. This was not confirmed; however, they assumed that the circulatory problems were permanent. On the next morning rounds, Tim, the responsible physician and Björn, the assistant physician from the last shift, discussed the situation:

Thus, the physicians conducted a CT scan to verify this working diagnosis, but it could not be confirmed. Thereupon Mechthild, another physician, suggested that it might be a virus. Although Tim did not agree and thought it was too early to search for a virus, he agreed to test it. However, the team forgot to do the test, and the idea was dismissed later. The physicians reduced the seizure medication since this medication is also responsible for drowsiness. To adequately wake Fritz up, they inserted a throat tube to ease his breathing, but because Fritz was still drowsy, the physicians again discussed the situation on the next morning rounds. Mechthild assumed that it was an anaphylactic reaction, whereas Gerd, the chief physician, believed that it was a catheter problem (the medication catheter was next to the dialysis catheter, so all medications might be filtered by the dialysis). Thus, the physicians stopped the dialysis to verify this diagnosis, but it could not be confirmed. As the physicians still had no diagnosis to work with, they treated the circulatory problems with medications but still did not know why they were occurring. When Fritz had a new episode of drastic circulatory failure, Susi, an assistant physician who had not worked on Fritz before, analysed his file. The new working diagnosis was that Fritz had a fluidity problem created by the dialysis. Thus, she changed the dialysis mode. However, Fritz had new episodes of circulatory problems, so this working diagnosis was also not confirmed. The physicians were, at this stage, desperate to find a working diagnosis. Tim came back to his initial idea that Fritz might have an infection. The ultrasound to verify an infection on the heart did not confirm this working diagnosis, and the microbiological tests had not been analysed yet when Fritz was transferred to the nephrology department.

Fritz’s treatment shows how paths were actively kept open. When the working diagnosis was not confirmed, physicians were more pressured to search for a new working diagnosis to cure the patient. However, even when the working diagnosis was confirmed, the physicians questioned it time and again. They tried to stay open and not become narrow-minded in their work:

Especially with patients staying in the ICU for a long time, the danger is to say ‘well, we know him or her’. Therefore, we need other physicians coming along questioning what else we could do. These physicians work against routine-blinded behaviour.

Physicians also explained that they always wanted to be up to date, as the patient’s state of health changes all the time:

We are aware that nothing is set in stone; everything is in flux. The status of a patient is always subject to a change process. Thus, I always have to be up to date.

Questioning the working diagnosis means to make repeated efforts to do something. In Fritz’s case, physicians were relatively sure for a long while that the kidney problems were only temporary. As long as no new information was available, they could have also just waited, doing nothing. However, from time to time, they conducted a new ultrasound to question their working diagnosis. In other words, the actors enacted actions to keep the path open. Searching for a new working diagnosis was different insofar as the physicians could not find a working diagnosis to start from. What they assumed was not confirmed by the tests they conducted. In contrast to the first example, the actors were forced to enact more actions to treat the patient respectively since, otherwise, the worst case would be that the patient could die. To keep a path open, more actions by a number of actors are necessary. Per the definition, more actions by more actors means that the complexity changes (increases) in the process of treating the patient. The more actions that are enacted by actors, the more complex the task is.

Raising questions/Making decisions

To form new paths and to keep paths open, everyone throughout the hierarchy was allowed to raise questions, and physicians listened to them, as one high-ranked physician explained:

I have done CT scans because physiotherapists have told me something. [. . .] The same holds true for nurses. I would also react to the cleaning lady if she gave me important information.

Moreover, they were not just allowed to speak up; everyone was expected to monitor the patients and question whatever was suspicious. Physicians were quite aware that nurses, especially, had the most direct contact with patients:

No other professional group in a hospital has as much contact with patients as our nurses. That means no one is observing them as intensively and frequently besides the technical monitoring devices. [. . .] No one has the possibility to collect so much information about patients. In consequence, this means that [. . .] they are our detectors, who collect and report all information.

Another physician confirmed this:

They are also medical professionals. [. . .] They have never been to a medical school, but sometimes this is not a disadvantage. They have learnt much better to observe and to actually get to know the patient and, thus, have expertise in a field in which we are novices.

Even though everyone is asked to speak up, final decisions are made by the physicians. Nevertheless, nurses see this process as a team decision, as they can provide their opinion:

[A] transfer is a team decision. Experienced physicians also listen to us when we say ‘better not’. They may not do what we say right away, [. . .] but they take to heart what we say and question their decision.

This interactive process between physicians and nurses is exemplified in the following situation in which Tanja, a nurse, makes Mechthild, a physician, aware of a skin rash:

Following the patients’ wishes, although they are oftentimes not able to articulate themselves due to unconsciousness or poor health conditions, was important to the nurses and belonged to their identity as nurses, as the following quote indicates:

We always think ‘if that were our mum or dad, we would like to have someone doing everything they can’. [. . .] I think that is what drives us.

In sum, all personnel were called in to monitor patients, speak up and question treatments. Nurses are of particular importance, as they are very close to the patient. In the process of decision-making, they can also provide their opinion; however, the final decision is up to a physician within the hierarchy. Speaking up and questioning specific treatments or providing information for possible new treatments is of utmost importance in forming paths or keeping them open, thus maintaining or increasing the task complexity.

Interdependent interruptions

An example of this occurred in the beginning, when the ICU team treated Fritz after surgery with the expectation that he would wake up soon (path 1). However, a nurse noticed high blood loss (path 3). Klaus, the physician on duty consulted Fritz’s surgeon to discuss the further procedure. Fritz’s surgeon provided a medication plan, but moreover, advised not to wake Fritz up until they were sure that Fritz had no secondary bleeding. Another example is that Fritz had seizures (path 7) when the physicians tried to wake him up (path 1). Tim, the responsible physician first administered medication and then checked Fritz’s file for further information. However, the physicians did not try to wake him up again until the medication had taken effect. Only then did they try another wakening approach.

In these examples, actions in one path were interrupted and stopped by actions in another path for a certain time. The treatment in path 1, waking Fritz up, was interrupted by the high blood loss (path 3). The physician who operated on Fritz decided that the on-duty physician should not take any actions to wake Fritz up (path 1) until the bleeding (path 3) was stopped. Thus, path 1 was interrupted by path 3, and the actors stopped any actions related to path 1 until path 3 was cleared (see Figure 5). Later, path 1 was again interrupted. This time, Fritz had seizures (path 7) when the physicians tried to wake him up (path 1). Thus, path 1 was interrupted by path 7, and actions within path 1 stopped until the physicians thought that the medication would prevent the seizures (see Figure 6).

OPEN IN VIEWER

Figure 5.

The process of enacting parallel paths – enacting interdependent interruption.

OPEN IN VIEWER

Figure 6.

The process of enacting parallel paths – enacting interdependent interruption.

Interdependent consultations

At the end of Fritz’s time in the ICU, he became increasingly confused and aggressive (path 12). Tim, a physician decided to consult psychiatry for advice. They provided a medication plan. However, assistant physician Susi was not sure if she could give a certain medication because Fritz’s kidney was not working properly (path 5). Thus, Susi called nephrologist Monika to determine which medication could be given to Fritz.

In this case, psychiatry advised a certain medication to treat Fritz’s confusion and aggressive behaviour (path 12). However, the attending physician was not sure if this medication would have consequences for path 5. Thus, Marie consulted an actor of path 5, in this case, the attending expert, nephrologist Monika (see Figure 7).

OPEN IN VIEWER

Figure 7.

The process of enacting parallel paths – enacting interdependent consultation.

Interdependent actions

When the physicians attempted once more to wake Fritz up (path 1), they continued to monitor him for seizures (path 7). On the next morning rounds, physician Tim ordered a CT scan to figure out why Fritz was not properly waking up (path 1) and why he had circulatory problems (path 4). Moreover, nephrologist Monika wanted a new ultrasound to evaluate the current state of the kidney (path 5). Tim told her that Fritz was going to the CT scan anyway and that they could scan the kidney as well. The next day, Gerd, the chief physician ordered an X-ray to evaluate reasons for the circulatory problems (path 4), kidney problems (path 5) and the pleural drainage (path 8). The last incident was the moment a nurse noticed that Fritz was confused and aggressive (path 12). Physician Klaus immediately developed a working diagnosis that this might be a side effect of the anaesthesia; it could also explain the breathing problems and panic attacks (path 10). Thus, he provided a medication plan for both paths. The physicians were still handling these paths separately because Fritz had panic attacks from time to time, which needed immediate treatment, and the laboratory found a fungal infection, which could also explain the breathing problems (path 10). After a while, Fritz became more depressive; this was treated with antidepressants meant to help against the panic attacks (path 10) and aggressive behaviour (path 12).

These two examples illustrate that one action was enacted for several paths. In the first example, paths 1, 4 and 5 needed that CT scan (see Figure 8); in the second example, paths 4, 5 and 8 needed the X-ray (see Figure 9). Thus, the physicians were able to conduct only one examination instead of three. The last example shows how medication was given for paths 10 and 12 (see Figure 10). This action, giving medication, belongs to both paths; however, it was only this one action that was enacted in relation to both paths. The actions before and afterwards belonged to only one of these paths, which is different to enacting an interdependent course of actions.

OPEN IN VIEWER

Figure 8.

The process of enacting parallel paths – enacting interdependent actions.

OPEN IN VIEWER

Figure 9.

The process of enacting parallel paths – enacting interdependent actions.

OPEN IN VIEWER

Figure 10.

The process of enacting parallel paths – enacting interdependent actions.

Interdependent course of actions

When the physicians started a second wakening trial (path 1) after Fritz had seizures (path 7), Fritz was waking up during the night. On the morning rounds, the physicians discussed how to continue and if Fritz was sufficiently awake to breathe on his own. Physician Mechthild suggested reducing the Keppra, the medication that Fritz had to reduce the likelihood of seizures (path 7), because it also made him drowsy. Physician Tim agreed. In another instance, chief physician Gerd assumed that the medication to treat the circulatory problems (path 4) was being filtered out by the dialysis to treat the possible kidney failure (path 5). His working diagnosis was that the catheter to give medication and the one for the dialysis were located too close to each other. Thus, he decided to relocate one of these catheters. The executing physicians Tim and Mechthild decided to execute this in a slightly different manner and did not relocate the catheter but did stop the dialysis to evaluate this working diagnosis. Nurses monitored this process, reporting that the dialysis was not the reason for the circulatory problems.

The examples show how actions in two different paths influenced each other on an ongoing basis. Whereas in the first example, the physicians only reduced the medication in path 7 to continue actions along path 1, in the second example, the interdependencies continued over a course of actions. Gerd, the chief physician assumed that there was an interdependency between paths 4 and 5 (see Figure 11). While reflecting on path 4, he assumed that the medication for path 4 was being filtered by the treatment for path 5. Following this, he gave the order to relocate the catheter in path 5; however, the attending physicians discussed the situation related to path 4 again. They made a different decision to stop the dialysis for a certain time. The expert belonging to path 5, the nephrologist Monika, had to execute this action. Nurses had to monitor the results, which were important for path 4. Thereby, actions were moving back and forth between the two paths.

OPEN IN VIEWER

Figure 11.

The process of enacting parallel paths – enacting interdependent course of action.

To sum up, these examples show how different paths are interdependent. However, the manner of interdependence is quite different. Dependent actions merge different paths into one path for one specific action, thereby reducing the number of actions, as only one action is enacted instead of three. Per the definition, this reduces the task complexity for the moment. The underlying logic for the other three interdependencies is quite different. In none of these paths could actions continue without a connection to one of the other paths. From a path management perspective (Goh & Pentland, 2019) a new path is formed which, per the definition, increases task complexity. When one path was interrupted by another one, a new path was formed because the first path could not continue unless the interrupting path was cleared. Thus, the two paths are, for a certain time, inseparable. The same holds true if actors from one path need to consult actors from another path. The two paths are only connected for a short moment, but they form a new path based on the inseparable connection. The broadest interconnectedness is created if two paths are interwoven for a longer course of action, going back and forth.

Solving an issue

When Fritz was first transferred to the ICU, physician Klaus noticed an antibiotics order which he did not understand (path 2). Thus, he collected more information, asking the nephrologist for an expert opinion, which he followed. The issue was solved by stopping the antibiotics. Shortly afterwards, the nurses and physician Klaus noticed a high blood loss (path 3). Klaus gave Fritz medication, and the nurses monitored the blood loss. As this declined during the night, the issue was solved. Later Marie, an assistant physician, noticed a lack of white blood cells, which is an indicator for a higher infection risk. Thus, she transferred Fritz to a single room to reduce the risk of infection (path 6). However, after a few days Tim, the physician on duty, noticed growing infection values, which he treated with antibiotics (path 9).

Phasing out

When the physicians started the first trial to wake Fritz up, a nurse observed that Fritz was having seizures (path 7). Thus Tim, the physician on duty, gave him medication. Moreover, the team waited for the medication to work. When Tim and Klaus started another wakening trial, the nurses monitored Fritz for further seizures, but he did not have any. However, he was also not fully awake. Two days later Mechthild, the physician, suggested reducing the seizure medication, as the medication also made Fritz drowsy. Tim agreed to follow that suggestion. Shortly before, the radiologist had noticed a pneumothorax due to a wrongly positioned pleural drainage tube (path 8). Tim and two assistant physicians put in a new drainage tube. Over the next few days, Fritz still had problems breathing, and his kidney values were poor. So Gerd, the chief physician, ordered an X-ray. He also ordered to check the drainage in the X-ray. After another few days, Mechthild decided to change Fritz’s diet because nurse Tanja made her aware of Fritz’s skin rash (path 11). In the next round, Jupp, an assistant physician asked how to continue with the skin rash. Mechthild decided to wait. A few days later Tim, Klaus and two other physicians discussed the skin rash again and assumed that it was an allergic reaction due to different medication. Tim instructed that Fritz be closely monitored.

These examples illustrate how paths dissolved because health issues were solved or phased out. When a treatment helped to overcome the health issue and it was assumed for the moment that the health issue would not come back, it was considered solved. Discontinuing or administering a medication, transferring Fritz to a single room, and giving antibiotics generated direct results. In contrast to solving a health issue by generating a direct result, in other examples, it was unclear for a while if the treatment helped to overcome the health issue; thus, the path did not dissolve right away and rather phased out (no further actions). Giving Fritz the seizure medication helped to prevent further seizures. However, it took a while for the physicians to decide to stop giving him the medication. When the physicians put in a new pleural drainage tube, they could not be sure if it was located correctly, as it was not visible from outside the body. Not until Gerd, the chief physician, decided to order an X-ray anyway did they check to see it was located in the right place. Similarly, to treat the skin rash, Mechthild decided to change Fritz’s diet and then waited to see how Fritz would react. A few days later, other physicians discussed the situation again and decided to just monitor Fritz closely. All these examples indicate that some treatments do not generate any direct results.

In sum, actions either dissolved because the health issue was solved or stopped for a while because the physicians waited to see if the treatment worked. Either way, paths eventually dissolved or at least fewer actions were performed, both of which changed (decreased) the complexity of the task. The fewer paths that were enacted by physicians or the fewer actions that were performed, the less complex the task was.

New paths are formed

In line with research on unexpected events, actors were driven to act upon new cues to start or continue working (Bechky & Okhuysen, 2011; Christianson, 2019; Klein et al., 2006). As outlined in our findings, actors reacted to cues but not with a simple and rushed approach, as this could lead to inferior consequences as research on decision-making has convincingly shown (Kahneman, 2012). Instead, actors started to enact ‘meaningful action’, which is action beyond the level of mere guessing (Kramer, 2007, p. 77). Kramer outlines that meaningful action is a response between the extremes of being absolutely sure in a closed system of rules and completely guessing with no rules whatsoever. Meaningful action in that sense means working with hypotheses. Hence, meaningfulness is not to be confused with knowing exactly what to do but acting on the basis of ‘fallible, partial, and preliminary knowledge’ (Kramer, 2007, p. 77). Nevertheless, starting to act with a hypothesis in mind generates new information and makes cues available through acting (Rudolph et al., 2009). In our findings, being open to new cues and acting upon them formed new paths, in other words, new sequences of actions to accomplish the task of patient treatment. The formation of new paths increased the task complexity.

Keeping paths open

Through acting, attention becomes focused on specific aspects of the task. An inherent danger in these situations is that the hypothesis with which actors are working becomes ‘closed off’ from other cues, which means that actions become increasingly influenced by existing insights without being open to new ones (Kramer, 2007; Weick, 1979). In order to prevent actors from only working with those existing insights, the insights must be updated repeatedly. As Christianson (2019, p. 46) mentioned, the essence of sensemaking is captured in the question ‘What’s the story?’ (see also Weick, Sutcliffe, & Obstfeld, 2005), whereas ‘updating also raises the question of whether the sense that has been made still makes sense’. Moreover, research on updating stresses that it differs from initial sensemaking, as it is more constrained due to earlier sensemaking (Christianson, 2019; Weick, 1988). For example, Rudolph and colleagues (2009) showed that updating only took place when actors were open to new cues and the plausibility of explaining the new cues was in favour of the current sensemaking process. This research stresses that updating the current sense can be a self-reinforcing feedback loop in favour of the existing sense. Thus, Kramer (2007) stresses that discrediting one’s existing insights helps to overcome the risks of a positive feedback loop. Discrediting existing sense is not only the attempt to update existing insights but an active doubt of current beliefs. Our findings underpin this research by showing that a current questioning of a working hypothesis is a constant mechanism to doubt current sense. In addition, the process of searching working hypotheses uses doubt to question initial assumptions.

Our findings indicate that the process of discrediting existing sense was supported by allowing every actor throughout the hierarchy to raise questions. This is in light of research pointing to a discussion-friendly culture (Schulz-Hardt, Jochims, & Frey, 2002) or an open spirit of contradiction (Kramer, 2007). Our findings showed that superiors asked unpleasant questions by challenging existing treatments or that nurses took over the role of patients’ advocates, also challenging earlier or future procedures. However, as in situations of time pressure, actors cannot engage in endless time-consuming discussions; it is important that decisions are made in the hierarchy and not necessarily through mutual agreement. Our findings pointed out that even though everyone was allowed to speak up, in the end, decisions were made in the hierarchy.

In contrast to previous research stressing the importance of reducing task complexity as much as possible (Campbell, 1988; Wood, 1986), our research points to the significance of increasing task complexity as a productive encounter. The two mechanisms for forming new paths and keeping paths open are consequential for this increase. This finding is in line with research pointing out the importance of complex thinking (Tsoukas, 2017; Weick, 2007). Accordingly, in unique situations, paying attention to ambivalence and newness, that is, preserving ‘disorder and confusion’ (Weick, 2007, p. 17), is indispensable. Mechanisms to keep or increase complexity are necessary to capture unique situations and obtain sufficient information to make decisions (Schneider, Wickert, & Marti, 2017). Hence, nontrivial agents engage with an open world, which might be the trigger for more complex forms of inquiry (Tsoukas, 2017). Adding to this line of research, our findings demonstrate how complexity can be increased in the process to create disorder and confusion. The permanent questioning of hypotheses produces more actions, which also creates increased task complexity. Thus, the process of keeping a path open, not by just waiting and doing nothing but, instead, actively questioning the existing actions time and again (Rudolph et al., 2009) increases task complexity. Moreover, our findings show that updating is not something that needs to be balanced with tasks as highlighted by Christianson (2019). Instead, updating is – to a greater or lesser extent – part of any task, and thus, updating is enacted through enacting the task.

Paths are dissolved

In addition, our findings indicate that paths dissolve when the issue is solved or a path phases out, meaning no actions are taken for a while. Fewer actions and paths create a decrease in task complexity. This indicates that after being open to complexifying, decomplexifying is also indispensable. Staying open just for the purpose of being open (Tsoukas, 2017) does not help to complete a task. Whereas solving a task is a proactive action, phasing out is a rather passive process. For a while, no actions are enacted along a path, and actors simply wait for a result or do nothing.

Thus, our research provides insights on how task complexity changes in the enactment of an organizational task. Whereas previous approaches could only say that task complexity changed (Hærem et al., 2015; Wood, 1986), our findings point to the different mechanisms that form the enacted task complexity: how paths form, dissolve and phase out. Thus, task complexity is not seen as a ‘thing’ out there, which waits to be measured; quite in contrast, the construction of task complexity is a social practice. In the process of socially enacting a task, actors construct task complexity. Thereby, the focus is not on reducing task complexity; to the contrary, task complexity can be increased to accomplish an organizational task many times in the process.