Random Collisions in the Workplace: a driver of innovation

One of the workplace activities we rarely notice—bumping into people in the hall on our way to and from other places—turns out to be an unsung hero of innovation. As we find ourselves coming back to the office, many of us are rediscovering the delightful serendipity of this seemingly insignificant interaction.

But don’t be fooled by its apparent lack of importance. Random collisions are anything but minor players in driving innovation across the enterprise. They are one of several antidotes to address remote-only workplaces (which, if left unattended devolve into siloed organizations). I was reminded of their importance in a recent engagement with a couple of my workmates while we were in the office together.

Back to the Office

The other day many of us found ourselves back in the office. We had a variety of reasons for coming in: kicking off our 2023 goals, getting alignment across teams (up and down the org chart), and working directly with one of our clients with whom we were embarking on a long-term effort.

One afternoon I met with a buddy from a different part of the organization. Usually, we would have met remotely to discuss a completely different project we’re collaborating on, but instead, as we were both in the office for this client, turned our energies to discuss his team’s effort for his part of the joint project.

He walked me through the initial thinking he and his team were considering. As he discussed his plans, I noticed he was focused on a capability an individual on my team had implemented several months earlier. When we dug into that topic, he raised several questions about what exactly, my team had done. I begged ignorance on the details, but realized that, happy happenstance, the individual happened to be outside our conference room preparing for the client visit the next day.

We moved the conversation outside to the adjacent open working area. I made introductions; neither of them had heard of each other or knew what the other was working on. Within a few minutes, my team member was walking my associate through the work he’d done. After a few minutes more, the two of them had calibrated their understanding of the work, and my associate suggested that the work he and his team were doing would in fact accelerate and improve on the solution my team had built.

10 minutes of interaction that will result in applying a solution from his team to accelerate a need on my team. What happened there?

On the nature of Activity Based Work

A few things happened there:

  1. We were all in the same place at the same time, working on a joint project.
  2. I was able to connect the dots in my network because I, alone, was aware of what each person was doing.
  3. We were all from the same discipline (UX in this case), sharing a vocabulary and point of view that accelerated our ability to get aligned.

Technically there were three enablers of that engagement:

  1. Because we were all in the same place, we enjoyed the benefits of a “Random Collision.”
  2. Because I was acting as a “connector node” in my social network, I was able to connect two other nodes who were unaware of each other.
  3. Because we occupied a space specifically designed to enable this interaction, we experienced very low friction in moving from one activity to another.

The first step in answering the question of what happened is to view what we were doing through the lens of Activity Based Work (ABW).

Three taskflows rendered as horizontal lines stacked vertically: Associate, Me and Team Member. 
Each line is labeled with an ABW that connects the taskflows.
A graph illustrating the relationship of Activities to each individual’s taskflow

In the diagram I’ve identified the workstreams associated with each of us, rendered as the horizontal time lines. At the first tick of time, my associate is “Sharing Out” (ShO) to me the work he was proposing, while my team member was doing “Individual Focused Work” (IFW). Soon, my associate and I were actively engaged in his work, “Informal Collaborative Engagement” (ICE), eventually leading to my realization that there was a connection between him and my team member, “Random Collision” (RCo). The event ends when the two of them engage in ICE and I’m out of the picture.

Building off the work of American architect, Robert Luchetti in the 1970s, ABW describes the workplace as a series of activities constructed from physical space, furnishings and technology. ABW was further developed in the work by Dutch architect, Erik Veldhoen and his firm Veldhoen+Company. I became aware of ABW while participating in athenahealth’s Future of Work (FoW) taskforce. In that effort, we identified 16 different Activities our employees engage in when we are working. Note: When I say, “when we are working,” I don’t mean when we are working in the office, but rather, when we do any type of work associated with our jobs (as opposed to activities we do in our personal time). Note also that ABW is like, but not the same as, Jobs to be Done (JTBD); ABW operates at a much higher level of abstraction.

So, the first answer to “what happened there?” is that amid the Activities we were engaged in, we encountered a Random Collision, which shifted us into a different working relationship. A Random Collision is an unplanned, spontaneous meeting with someone you know. Folks who’ve been in offices with greater than a few hundred people recognize this type of engagement. You’re walking from the café back to your office and you bump into someone you haven’t seen for a few months. You stop momentarily to catch up, plan to have lunch, or exchange some tidbit of work you’re doing.

The introduction I made between my buddy and my team member is a variation of the Random Collision, which leads to the second answer to the question.

On the nature of Network Graphs

A different way to think about work is to look at the structure of your network, easily viewed through a network graph. Over the years, the academy has identified specific types of network graphs that improve leadership, collaboration, innovation and productivity. This wonderful article in HBR (Better People Analytics  – HBR Nov-Dec 2018) provides a good summary of the topic.

A network graph: circles (nodes) connected by lines (edges) to other circles.
My node sits between two other clusters of nodes and edges: my teammate's cluster and my associate's cluster. 
Their two networks do not connect except through me.
A network graph illustrating my node connecting my associate’s network with my teammate’s network

Looking at that same situation through a network graph, we can see that my knowledge of two nodes (who weren’t otherwise connected) allowed me to literally connect the dots, enabling them to share vital, relevant information.

Who knew that there was so much theory hiding behind this seemingly effortless introduction? (It turns out a lot of folks know about this…it’s been studied for years.) But wait, there’s yet another way to answer the question, “what happened there?”

On the nature of the Workplace

I was able to connect these dots very easily because of the explicit and intentional design of the space we were in.

A floor plan of a collaboration studio: a main work area taking up 80% of the space, with two closed rooms to the right.
A floor plan of the space each of us were working in

In the floor plan above, my associate was projecting onto a screen in a conference room with a glass wall bordering the open workspace to the left. My teammate was visible to us, but isolated audially. This physical design enabled the two of us to work as loudly as we needed, without disturbing my teammate, but kept the three of us aware of each other through the glass wall. To get the two of them together was as simple as walking 10 steps.

This space design was intentional. As part of the FoW initiative, we developed several labs to explicitly test our hypothesis that our work in the office would be enhanced if we could engage in multiple Activities in the same space, while enabling fluid transition among them.

The importance of Random Collisions in driving innovation

Now one of the interesting things about Random Collisions is that they are the engine of innovation. Here again we can turn to the academy to learn how organizations can improve their innovation cycles by explicitly enabling Random Collisions in the workplace.

One way to think about Random Collisions is to borrow an analogy from physics: imagine the workplace as a container of people who act like molecules of air. For innovation to work, you want to maximize the likelihood of two different people bumping into each other. The effect is enhanced if the two people are not that likely to bump into each other on a regular basis (that is, they don’t work in the same group). The larger the container, the less likely they’re going to bump into each other.

A thick-lined rectangle with an opening (suggesting an entrance), enclosing six orange dots (representing people)
An illustration of everyone in a single room

Consider the following: you’re working in a startup; there’s a handful of you, maybe working out of one room in a rented office or in someone’s house. Using the physics analogy, you’re bumping into each other all the time. You’re not Randomly Colliding, you’re just Collaborating—a different Activity. Innovation is going on constantly. That’s the mode your group is operating in: bringing ideas to market, the classic definition of innovation.

A rectangle enclosing 12 dots suggesting a larger facility with more people
An illustration of a larger room with more people

Now consider what happens to that startup as it grows, both in numbers and space. When the group grows beyond about 100 people the frequency of interactions between any two random folks decreases. When the space grows into multiple rooms separated by hallways the opportunity to interact decreases. Random Collisions begin to emerge. At this stage, the organization is still small enough that everyone knows everyone, so Random Collisions do happen, but other types of Activities (all hands meetings, lunchroom socialization, and project-based interactions) predominately maintain the group’s connections and interactions.

An rectangle enclosing additional rooms, each of which have a smattering of people.
An illustration of an office floor plan with people separated by rooms

Expand by an order of magnitude: 1000 people and an entire building (or floors in a building, or separate small buildings in an office park). There are likely wholly separate operating units divided by large functions: HR, Finance, Operations, R&D, perhaps. Within each, there are separate divisions. For example, R&D might comprise multiple product lines. This is where things start to get interesting with respect to Random Collisions.

Imagine you are a Product Manager (or a Software Architect, or Lead Engineer, or User Experience Strategist—it really doesn’t matter what the position is, as long as it is focused on bringing ideas to market, i.e., innovation).

You’re working in a product line inside R&D, focused on a particular market (or technology or experience). Unbeknownst to you, someone in a similar role is working on a very similar problem in a different product line. Or, equally likely, someone in a different role from you, in a different product line, is working on an aspect of the problem (or solution) that would be very helpful to your efforts (my example at the top of this article). Or, even more impactful, someone in a completely different role (finance), working in a completely different division, has just the right competency to help you with a problem you have no clue how to address.

As organizations grow, their ability to innovate depends on connecting nodes in an ever-expanding network of people. But that expanding network conflicts with the organization’s natural incentives: our team (the folks close to us in our network graph) needs to work productively; so, we naturally spend more time together. Left on their own, teams will naturally devolve into the “efficiency signature” network graph.

Six nodes are connected to each other with edges, creating a tight cluster. Each node has edges connecting to nodes outside of the cluster, but those "external" nodes are not interconnected with each other at all.
A cluster of nodes tightly connected by edges to each other

Over time, the organization will become siloed.

Several efficiency clusters are loosely connected to each other: a single node from one connects to a single node in another.
Several loosely connected clusters indicate the organization is minimally connecting its divisions

To enable innovation in large organizations, leaders need to deliberately nurture connections outside of their team. Some organizations create ceremonies to help: “innovation days,” cross-organization Lunch and Learns, cross-product-line group meetings. Each of these are useful, but they require individuals to work against their team’s self-interest: completing work on their projects.

The Random Collision is a natural antidote to the silo tendency, but it requires two key enablers:

  • People must be in the same physical space
  • The space must be physically designed to enhance the likelihood of Random Collisions

BMW, when they designed their innovation centers, absolutely understood the value of enabling Random Collisions in service of driving innovation. Knowing that separating people into rooms, across floors and across campuses decreased the potential for Random Collisions, the design team carefully crafted the centers to overcome those barriers, maximizing the likelihood of disparate folks bumping into each other. (For a wonderful account of the design process behind BMW’s FIZ1 and FIZII innovation centers, read The Organization and Architecture of Innovation.)

The impact of remote work on Random Collisions

An array of squares (representing home offices), each with a single dot inside
An illustration of individuals in their separate home offices

Now, let’s look at the pandemic and dispersing everyone to work from home. The container has stretched to an entire country or worldwide, but our container is our home office. Regardless of the size of the group—10 people or 100,000—Random Collisions have effectively been driven to 0. So long to seeing Georgina in the hall. Goodbye to running into Tracy at the lunch counter.

We’ve tried to adapt, but it’s definitely not the same. We look on our chat system to see someone’s status indicator (red? green?) and decide to bother them with a message. Depending on the power dynamic we’re more or less self-conscious about reaching out. Maybe they’re multi-tasking and will answer us. Maybe they’re away from their system. The moment passes, we’re pulled into a meeting, and the exchange is attenuated over time…perhaps hours, or days. Or, we decide it’s important enough to schedule a meeting and talk about it. But it’s anything but random. It is far from effortless.

In fact, it’s so far from effortless that even this attempt at overcoming barriers in our network graphs happens infrequently. We use the limited time slices we have to meet with the folks with whom we’re already connected. And that’s absolutely the opposite of Random Collisions.

On the nature of Random Collisions

Random Collisions happen to us, not by us

Ignore the effort for the moment; there’s something else about Random Collisions that’s qualitatively different from our online attempts to engage. Random Collisions happen to us, not by us. Our current digital technologies are not built to randomize connections in our network. (At times it feels that our technologies can’t connect us to the nodes we want to work with, a purpose for which they are ostensibly designed. But I digress.)

For organizations that were “born remote,” they recognize the importance of getting together in a shared physical space to do the kinds of Activities that just work better when people are physically adjacent. Consider an annual sales meeting, or customer conference. Consider a scrum of scrums or other multi-divisional event when folks fly in from their remote hideouts. Random Collisions happen all the time in those cases, as do the types of conversations that lead to innovative breakthroughs.

So, back to that interaction with my buddy and my teammate.

10 minutes.

Could I have gotten them together if we’d all been remote? Sure. Sort of.

But given I’d been working with each of them for the prior two years, over which time I hadn’t connected them, clearly something happened that day which made a difference.

Let’s say we hadn’t come into the office. Let’s say my buddy shared his approach with me online instead. I would likely have raised the same observation and mentioned my teammate. And then what? I would have created a group chat, and perhaps the two of them would have taken it into a conversation. Given my experience over the past several years, however, I know that that conversation would not have happened over the span of 10 minutes. Maybe more like three or four days. In practice it appears we need to do something very differently to enable Random Collisions when we’re all remote.

What we can do to improve our ability to innovate

From my FoW initiative work, I’ve concluded there are three options for organizations trying to drive innovation in a neo-pandemic world:

  1. Get back together in real space. Let’s be clear: when your teams are the engines of innovation you will simply need to get them together in a shared space. There’s no real substitute. But we’ve also learned that these collocated Activities don’t need to be full-time.  And of course, not all teams drive innovation. But as many of us have discovered when we return to the office: for certain types of Activities, we get more done in a few hours together than we could do remotely over days or weeks.
  2. Implement remote-based practices to enable Activities that drive innovation. Forget about Random Collisions in this option. Remember Random Collisions happen to us, we can’t, by definition, make them happen by taking an action. But there are several other Activities that drive innovation: Formal Collaboration, for example, is another Activity we identified in athena’s ABW framework. Using online whiteboards, with carefully crafted exercises and facilitation, we can get a group of folks to do many of the same things they do in person. But let’s not kid ourselves. Online collaboration tools and interactions fall short of what we get done in person. Still, something is better than nothing.
  3. Invent (or repurpose existing) digital technologies to enable Random Collisions. Maybe something like this already exists and I’m not aware of it (to invoke Gibson’s notions of the future: it’s already here, just not equally distributed). The closest technology I’ve played with is ambient chat rooms – like Gather.town for example. It’s not too surprising that by making a 2D floor plan (and hinting at where others are hanging out), these types of applications enable Random Collisions using the same dynamic that happens in real space. As I (virtually) walk around the space, I literally run into other people who I might not have expected to see. Sort of a digital cocktail party. No doubt there are other ways to enable Random Collisions that don’t rely so heavily on reproducing a physical analog. Perhaps by looking at our network and knowledge graphs a friendly AI agent can “bump” us together in some digital way to help us make those odd connections.

Regardless of how an organization grapples with their innovation pipeline, one thing is certain: it will not take care of itself if we choose to work only remotely using the technologies we have today. We must take explicit and intentional action to enable people from far flung parts of our organizations to spend time together, whether in a physical space or through (yet unknown) technology intermediation to leverage the power of Random Collisions on innovation.