the danger of how things naturally cluster in business




 accumulate, amass, assemble, bulk (up), collect, concentrate, congregate, constellate, corral, garner, group, lump, pick up, round up, ball, batch, bunch, cluster, huddle, heap, pile, stack, flock, herd, hive, pack, press, swarm, throng, combine, connect, join, link, merge, pool, unite


Several years ago, in a piece called “thinking, decision making and leadership” I wrote this:

  • Which leads me to declustering. Leaders like to cluster decisions and business like to cluster things. All are typically done because everyone believes it is more efficient and, possibly, more effective. Clustering is the unspoken organizational corrosive barnacle on the progress ship. It is true business, and clusters, arc toward replication. If a leader spots a ‘decisionmaker’ inevitably they will guide more decisions to that person simply. Decisions then cluster thereby making the overall – most clusters – desire to be involved in decisions to decluster. Yeah. For every action there is an equal and opposite reaction. By default, you’ve once again parsed out responsibility, not collaborative sensemaking, and created a different type of hierarchy. I am not suggesting some clusters aren’t better at making decisions than others. What I am suggesting is leaders’ natural instinct is NOT to spread sensemaking and thinking, but isolate so when convinced to distribute and spread, no matter how effective overall it may be, inevitably they will redistribute to a few – that’s the arc. Just ponder.

Several years ago I also wrote: “the natural arc of every business is to mediocrity.”

The two are related.

Yeah. I am saying clustering leads to mediocrity. That said, oddly, business tends to like clustering, or, they do not discourage clustering. Let me be clear. Given an opportunity to be excellent, a business will always choose the path to mediocrity. Yes. Always. What I mean by that is in every situation – customer service, capital investment, ideation, innovation, creativity, planning, strategy, implementation – given an opportunity to choose a Spinal Tap 11 choice, a business will always choose a 9 or below choice. And from there on out that ‘plus number’ begins diminishing bit by bit. And while I imagine I could point to a variety of reasons, let me focus on clustering as the culprit.

As soon as an “11” opportunity is identified, business will circle the wagons around some consensus data, group view, collective wisdom, whatever, which never raises the 11 to a 12, it only brings it down (usually significantly).

That’s clustering.

To be clear. Mediocrity is fairly equal opportunity employer. Data pools, cliques, opinions, specialists, likeminded clusters, all circle the wagons and transactional thinking actually (a) makes teams exclusive and (b) makes value finite and, yet, encourages a belief in exceptionalism (but a reality of mediocrity). From the ‘rational business’ perspective businesses like ‘clusters’ because they appear to be patterns (or some version of mass coherency) which they immediately associate efficiency and scale to. They don’t really care if the cluster benefits lose degrees of efficiency from that point on – they’ll flog the cluster until it is dead of anything good (or true value creating). Maybe worse is that over time clusters become ‘self-sealing belief’ parts within an entire system basically creating limits to tapping into the larger potential of the entire organization. That doesn’t mean they don’t offer value, its just that the value has limits and tends to stratify which makes them a bit more difficult to mesh with other components of the system. Clusters DO produce good things, but while those things seem good at scaling, it is not clear that they are necessarily good at innovation or anything ‘new’. Or another way to put it: clusters may might be good at exploiting (that is repeating a known winning pattern), but are less good at exploring (that is figuring out what needs to be done).

The corollary to that last sentence is clustering limits potential.

Business loves to cluster skills, but clusters tend to be herded into small clusters of micro skills in their macro skill puzzle. In this excellent video with Alain de Botton he points out that clusters are fragmentary in skills and interests and, yet, we are told what we are not good at, what we should focus on and then wealth is rewarded to that focus. In other words, we are told to cluster around a specific thing we may like or be good at and ignore (or relegate as a ‘hobby’) other skills and interests.

This has consequences beyond skills. The ways clusters interpret things do not translate neatly into larger decision trees. In addition. Decisions are not discrete, they are connected. A cluster may get parts of ‘now’ right, but the other parts may gather momentum or strength in the fact they were chosen as inconsequential in that ‘now’. In a cluster’s confidence lies its biggest risk. The truth is that, particularly in business, human expertise is usually tied to knowing what to look for – the nuances within hard data/information. Clusters become increasingly certain and blind to ‘new nuance.’ The repercussions to this is then a cluster is more likely elect to choose solutions or actions inconsistent with the problem/situation at hand but consistent with their view. That inconsistency can be in nuance or it can avoid some significant aspects, regardless, the consequences are often the same as the future unfolds from what was perceived as effective framing, but less-than-effective choice making. You have to decluster things. consciously. consistently, to optimize decisionmaking.

Which leads me to point out growth tends to polarize itself.

Cluster-driven growth is often spectacularly unprofitable. Business chases success/growth. Its inherent reaction to success is to milk that success until it is dry (albeit many businesses refuse to let go of even dry growth paths). Business uses math, numbers, relentlessly to pursue this objective. Reminder. It was Bernoulli who stated “mathematics alone does not capture a situation.” This is also true of a graph/dashboard. They simply inform or become a guide to a grander discussion of what exists and how to make sense of something, yet, business people cluster around dashboards and cluster energy/focus based on them.

In a world glutted with information, the capacity to ignore selectively becomes ever more valuable … a system that is to control its environment successfully must adapt by constructing models that allow it to decide what information to get and how to act on it.”

Seth Lloyd, MIT

Business loves numbers. Which leads me to remind everyone one of businesses/s biggest lies is “the numbers never lie.” Numbers lie all the time. Even beyond how people torture numbers until they say what is wanted, numbers create clusters. Yeah. As soon as you find a number you like it becomes a magnet for other numbers, resources, energy, focus, etc. What this means is business relentlessly clusters resources against a diminishing growth opportunity. Invariably ROI can never really improve, in a meaningful way, but intrepid business people will always find ways to suggest things are good and getting better. Once again, it never hits 11 and is only getting closer and closer to 1. The only way to get off that slippery slope of lessening growth is, well, declustering.

Which leads me to power.

The issue occurs with, well, power. Those in power like evenness and replication and they view the only way to keep balance is to enforce it. They believe that no one else understands how to balance or maybe can even see the value of balance. They see themselves as being in the middle doing what’s best – for their sake – and they use ‘clusters’ as hammers. They cloak their hammers in ‘safety’, i.e., we keep clusters safe and the business safe. This gets tricky because most people clusters have an unhealthy trust in in larger systems with regard to what is beneficial TO the cluster which makes them less likely to realize when they (the cluster) has made a mistake AND less likely to realize when the larger system truly has something to offer that is beneficial to the progress/development of the cluster. All that said. Power clusters. Even though an individual in power is either (a) obliviously overconfident or (2) constantly paranoid, most individuals in power recognize they need other power players to dominate others. So whether it be by salary/$, titles, perceived meritocracy, education, or even psychotic authoritarianism, people will cluster for power – in implementing as well as in holding onto. Obviously, businesses should actively decluster power as much as possible to attain full potential.

“It’s a game of nanoseconds. Judgement calls have to slide through impossibly narrow doorways.”

Carolyn Barclay

Which leads me to data, technology and humans.

I have written a number of times about how technology can be used to decluster. In my unpublished book I even dedicate a chapter to how algorithms and technology can be used in a business infrastructure to nudge declustering to the benefit of sensemaking and gaining knowledge. That said. Technology amplifies everything – strengths & vulnerabilities – all of which stress the boundaries of any system. The corollary to this is the more you depend on technology the more likely a catastrophic even will occur (resolves little problems hiding the lurking catastrophe inside) because technology attacks/addresses things systematically while systems shape themselves organically (creating dissonance at some point). What that last mumbo jumbo sentence really means is that technology clusters shit by amplifying what people like and or are interested in. Sure. Machines and technology will often suggest potential actions and we may calibrate different paths with probabilities of success. The problem is that everyone will have the same input and if everyone does exactly what the input suggests everyone will, well, cluster.

Data and technology are inputs, but it is experience, wisdom, patience, and emotional fortitude that are the key ingredients when added to the inputs that result in differing results. And this is where clustering gets, well, fucked. Yeah. I just said clusterfuck. If people are clustered, they apply the same experience, wisdom, emotion against the same clustered data and technology, hence, the clusterfuck.

Experts always suggest the key to an algorithmic future is to be sure you understand what an algorithm is optimized for. I believe this is too simplistic.

“We know from biology that relentlessly preventing small problems can increase the risk of future big problems.”

Seth Blumsack


Not only are algorithms typically built to head off problems immediately (eliminate small problems), but how they get advice, information, determines how they behave. This has some very practical consequences in that, generally speaking, is more likely to present only catastrophic complex problems to the attention of clusters. Basically it ignores small problems/issues. Well. Anyone remember: “its never the elephants that kill you, it’s always the ants”? We design our systems, process and even encourage clusters, to focus on building elephants. We even build algorithms with that in mind. Clusters are elephants.

Which leads me to end with cluster narratives.

I don’t really know if clusters of people, or data, create narratives or if narratives encourage clusters. What I do know is a good story told well gets repeated over and over again creating a myth – and people cluster around myths. Generally speaking, myths aren’t very good for business. Myths get in the way of reality and are stubbornly resilient against alternative narratives (learning). So, yeah, I am suggesting declustering narratives (which is actually deconstructing narratives). This is a semi-important point as I circle back to an opening point: the arc of business is toward mediocrity. I cannot remember one business narrative myth that extols mediocrity, yet, by following most business narratives you are simply following the lure of mediocrity. Quite the paradox. It kind of feels like the only way out of that paradox is, well, declustering. Ponder.

Written by Bruce