One of the factors determining how the agents in a complex adaptive system interact is proximity: the positioning and movement of the agents through space. The term ‘space’ can refer to physical location but also to any form of conceptual space. In that case the notion of shorter or longer distance refers to higher or lower probability the agents interact.

   The actions of agents can create, shape and alter the spaces in a system. The bazaars in Persia and souks in the Middle East formed when merchants set up shop to commercially interact with the travelers in caravan trade routes. Initially outside of city walls – but not too far as the city offered protection and production – these open-air market places gradually took on a permanent character and were eventually absorbed within the city as it expanded. This is an example of agent action creating a space, but that space then increasingly started shaping agent action. People go where other people are, and the souk gradually layered on social and cultural elements in addition to its original commercial function. These location dynamics also influence agent strategy. During a stay at Marrakech, on a walk to the souk near Jemaa el-Fnaa square, we first looked for a sandal shop as the searing hot street asphalt was literally melting the soles off my partner’s (old) sneakers. When we saw all sandal makers were grouped together in the same district I was surprised at first. Why didn’t shopkeepers spread out to all areas in the souk, to avoid direct competition in their immediate vicinity? But then I realized this would be a losing strategy. As soon as two or three sandal makers are clustered together somewhere, that’s where all buyers will go to easily compare product offerings and prices. The other sandal makers really have no choice but to also go where the buyers are. There are additional factors, ranging from tradition to efficient logistics of supplies and waste product, but it is no coincidence districts independently formed in the majority of souks.

   Conceptual spaces substitute the physical location element by some other dimension of easy-to-reachness between agents. Two computers might be literally stacked on top of each other, but if only one is connected to a network the other might as well be on the moon for a computer virus that seeks to propagate onto it. The Tinder app combines both physical and conceptual space in an interesting way to broker interaction between its users – or so I’m told. Physical proximity being a necessary condition for the more practical forms of romantic meetup, the app will only present a user with candidate partners within a certain radius around their current location. But the app also bridges an important conceptual barrier: it only offers interaction with other users open to meeting potential partners. Other dating applications, designed to facilitate deeper and longer lasting relationships, use additional methods to lower conceptual barriers between agents. Matching algorithms seek to optimize the odds of a successful coupling using a richer set of variables in the hopefuls’ profile information. Technology-enabled or not, these factors in relationship formation are of course as old as the hills, since mating is one of the earliest forms of agent interaction. Ask around with the couples you know, and you’ll probably hear most of them grew up quite close to each other. Even so, conceptual spaces – such as clan membership or caste – are often a more formidable constraint on relationship formation than physical distance. Members of European aristocracy are likelier to marry aristocracy from a different country than an ordinary citizen living close to their residence.


   In complex systems, altering distance is an important mechanism by which new technology transforms the agent interaction patterns. Large parts of the world are currently involved in a gargantuan technology-facilitated proximity alteration experiment in one of humanity’s biggest complex adaptive systems: business. The global pandemic displaced millions of knowledge workers, whose jobs are not hardwired to a specific production place, both physically and conceptually. At the time of writing, the worst of the outbreak appears to be behind us and the pluses and minuses of a mass return to the office are intensely debated. Many knowledge workers want to preserve the “work from home” routine they have become used to, up to 100% of their working time, while others advocate a full return to office based work. The opinions are sometimes strongly held, but it is hard to generalize whether or not office-based work is better than remote working. To better unpack what will happen, a better question to ask is how different agent types will adapt to the proximity changes, and how various work locations are likely to evolve as a result.

   Proximity changes will inevitably transform low trust work environments. Many knowledge work environments still function on the basis of suspicious managers keeping close supervision over employees they don’t trust. And the feeling is mutual: according to BambooHR’s “Bad Boss Index”[1] 44% of employees attribute their resignation to their manager. These managers push hard for a full return to office based work. But their employees have options now to keep their distance. Many companies with a more trusting culture offer (partial) remote work, which has rapidly become a competitive advantage in attracting new employees. This puts pressure on low trust environments to also evolve to a more collaborative and trusting work culture. But some are adapting by doubling down and using technologies, such as keyboard loggers, activity trackers and screen recording software, to maintain or even intensify tight supervision over employee activities. As Brookings reported in January 2021[2], employees “have little expectation of privacy while on company grounds or using company equipment” – in other words, wherever they are in company concept space. The employees will adapt to this adaptation though. It won’t be long before an entire cottage industry of software emerges to block logging or mimic activity on unused laptops. Spending less time together will never increase trust between unmotivated employees and narrow-minded managers.

   There’s inspiring managers too, and while most of their employees will thrive in a more flexible work arrangement, that is not the case for their apprentices.  Historically, young boys were placed with a master in crafts ranging from blacksmithing to painting. They started their training with menial tasks and built up expertise by observing and working alongside the master. This is not just a Renaissance thing. It’s still the way young cooks train for years in the kitchen of an accomplished chef. Students at James Dyson’s Institute combine formal study with project work alongside Dyson engineers. Some expertise or know-how can’t be picked up by reading a manual, but needs to be absorbed and assimilated through interaction and observation. Picking up the tricks of the trade is especially critical in the early years of somebody’s career. The starters themselves may not realize this, but their managers and mentors should – and make themselves available in the office environment even if that doesn’t serve their own personal productivity. This also creates opportunity for those starters who do get it to get more exposure and visibility: with less people in the office, there’s less competition to get some time with senior leaders and decision makers. For high trust environments, loss of apprenticeship is an existential risk. Some companies with a strong culture understand very well how it can destabilize their system. A Goldman Sachs memo even explicitly used the word ‘apprenticeship’ in a May 2021 memo calling employees back to the office[3].

   If the agents in the knowledge work system drift further apart in physical space, they also huddle closer together in concept space. For some, like quite a few software engineers, this is merely vindication for what they’ve said all along. Because of subcontracting and offshoring, they’ve worked on complicated projects across regions and companies for decades. They are obviously very comfortable with the enabling technology. Being forced together into a single physical location, often with an open plan layout to make matters worse, is counterproductive to them. They literally try to recreate spatial isolation by wearing noise canceling headphones and hoodies. Jason Fried said it well in his October 2010 TED Talk: the office isn’t a good place to get work done as the “M&Ms” of managers and meetings get in the way. Today that’s not just the coders any more: all kinds of other knowledge workers have moved a portion of their work to concept spaces ranging from Sharepoint to Slack. From there, it’s a small step for an individual worker bee agent to create some productive work in another concept space. The modern day equivalent of what used to be the startup garage is now virtual. Knowledge workers don’t need to leave their current company yet, they are gathering digitally to create new enterprises silently and out of sight. And those enterprises of tomorrow may not even be conventional companies any more. A Decentralized Autonomous Organization (DAO) is an early example of knowledge work systems built more in concept than in physical space.

   Spaces shape agents and agents shape spaces. Like the souks, changing interaction patterns between the agents in the  behavior will transform the workplace itself. Companies shrink office space, reduce the open office plan consisting of endless rows of desks or cubicles, and develop spaces designed for collaboration and other forms of human interaction. But adjusting the physical space is easier than deliberately designing around employees’ needs in concept space. Important information flows which previously relied on a physical proximity networks and workplace rituals need to be rebuilt or enhanced. Some of these support employee productivity, but other, subtler ones influence longer term employee development – as we saw with apprenticeship, for example – or engagement with the enterprise. But where most will struggle to play defense, the best organizations will play offense and create entirely new ways of putting knowledge work to productive use. Imagine an enterprise radically redesigned or rebuilt from the ground up in concept space first. Imagine an enterprise that doesn’t lose a portion of its employees productive time to stealth work for others, but instead gets contributions from outsiders attracted by its concept space. The future is already here, just not evenly spread: projects like Wikipedia, Linux or the first forms of DAO are early forms of knowledge work systems in a different form than the traditional corporation. Some knowledge workers too are monetizing their production in different arrangements than a labor contract and monthly salary. It’s still early days, but that’s the shape of things to come.



[3] Per “We know from experience that our culture of collaboration, innovation and apprenticeship thrives when our people come together, and we look forward to having more of our colleagues back in the office so that they can experience that once again on a regular basis”

(First published: March 2022)

Download in PDF format

Leave a Reply