• Joseph Hunt

Complexity Series: Part 1 of 4

May 14, 2021 | Blog

Complexity. The word alone is enough to disturb business owners and investors alike. Constantly reminded of the harsh reality of the unpredictability in the market, we are inundated with tools for analysis and organization of data as a linear series of causality. Reducing complexity, it seems, is key to optimal productivity and creating predictable & repeatable patterns of success. It isn’t easily avoidable, however. Even in 1983 academics and businesspeople alike were aware of the accelerating rate of globalization and its effect of increasing the complex interdependence of businesses. And very early on in the development and adoption of the internet, it was noted that trans-national communication via digital medium wouldn’t just become the norm; it would be central to success. This poses a unique problem for negotiators; how do you adapt a business model and contractual relationships to be effective in a space of ever-increasing uncertainty? But complexity isn’t to be feared and minimized; it in fact holds the key to growth and success in the modern marketplace. To provide a solution to this increasingly intricate system of relations, though, it is important to first specify the problem.


Defining Complexity

Complex systems are systems in which each node in the network has a set of nonlinear interactions with other nodes. This means that there is a level of uncertainty in the behavioral properties of each nodal point in the network. With each added agent, the expected arrangement of the system as a whole over time gets exponentially more unpredictable because the uncertainty at the small scale is compounded. This is what is commonly known as the butterfly effect. If you have a 50% chance of guessing the impact of one node on another, once you add a third agent the dynamic spirals out of the realm of predictability as unexpected impacts compound. Obviously humans as agents in such a system are far less predictable, so that determining the particular relational features of a group of people after a complex interaction process is even more daunting. This is why complex systems (particularly of people) are often called random: they exhibit emergent properties which make predicting their dynamics across time next to impossible for traditional analytics methods. Operationalizing their behavior would mean a wholesale simulation of the planet, with all its lines of connecting causality. Minimizing complexity, it would appear, is therefore in the best interest of industry. Maximizing predictable profitability means reducing any variables that jeopardize the validity of the models.


Building Toy Soldiers

Traditional behavioral and economic modeling seeks to predict the actions of individual agents in the network. They aim to arrive at the gestalt properties of a system by accurately modeling the relational dynamics of each node. In this regard such processes are doomed to failure on some level when applied to human interactions, as the nonlinear nature of such relationships induces an inherent level of unavoidable randomness. But building a model at the level of the node, rather than the relation, would allow a system to sidestep many of the problems of inherent nonlinearity and arrive at concrete, predictive conclusions about the output from such relationships. Such a system demonstrates a unique type of randomness, different from the traditional definition of random.

The key issue at play here is the equivocation of the word random. Typically the term is used to describe when all possible outputs of a function are equally likely, such that it is impossible to determine what the result is probable to be from the outset. Comprehensive unpredictability in a system meets the criteria for randomness, as by definition it means there are variables that cannot be factored into the model and thus all results are equally possible (from the system’s perspective). However, the unpredictability leading to inaccuracy in the application of behavioral modeling in human organizations (such as the system dynamics model) is a failure largely of the type of information to be achieved, not necessarily a feature of the population itself.

The Wrong Kind of Random

The unpredictability present in systems designed at the level of the node is the same as the earlier designs; it is impossible to predict the orientation of the system, the ways in which the nodes will be interrelated after a period of time. However, building up from the node allows a level of certainty about the net result (output) from an interaction (relation) without the necessity to specify how those interactions occurred.

For a simple demonstration of this we can observe global economics. While it is impossible to predict the success and failure of individual businesses (relation), there are a number of features of agents that allow a level of predictability on the net result (output). For example, each node in the global system makes decisions predicated on two conditions: (1) profitability and (2) sustaining the system. They need to maintain both their own existence through profit (1), as well as encourage stability in the system as a whole such that their supply chains and consumer base continue to exist (2). Through these two elements, it can be determined that the net output from the global economic system is extremely unlikely to collapse at the global scale. This is due to the nodal motivation toward stability that leads to the maintenance of each of those meta-goals through profitability (individual) and sustenance (global).

So while there is still unpredictability at the level of relation, node-based modeling techniques allow degrees of certainty in limiting the types of output from dynamic systems. This changes the definition of randomness when applied to these tools, as the set of possibilities randomly selected from is reduced in scope by features at the level of the node. As the accuracy of representation of the node increases, this scope will conversely shrink. Thus a system built from the node up exhibits set randomness, wherein the true unpredictable “randomness” of the system of relations is narrowed by predictions derived from the features of the agents modeled.

An Opportunity for Growth

Complexity isn’t necessarily something to be minimized, but harnessed and directed. The caveat is that while new modeling techniques can represent complexity more accurately, not all expressions of complexity can be manipulated to be conducive toward business success. In later articles in this series I will tackle complexity at the larger scale of business-to-business, demonstrating how tightness / looseness in business relations can lead to both mutual thriving as well as combined self-sabotage. At the smaller scale, node-based analysis provides insights into how businesses might manage expansion and self-organize in ways that increase resilience and decision-making effectiveness. And finally, I will demonstrate a number of specific tools that can enable the practical implementation of these abstract concepts in actionable ways, allowing negotiators to be more efficient, cohesive, and effective in communication and deal formation.

A New Breed of Business

Success in business is no longer determined by output data analysis; economic reports can’t tell us why something happened, only what happened. The success of modern business is considering each link in the chain carefully; modeling the desires and tendencies of each agent in the system. Customers, coworkers, business partners — each with individualized agenda and behaviors interacting to form the emergent properties of these complex systems, analyzed and simulated by the power of machine learning. This is the future of negotiation, and the future is now.

About Intellext™

Intellext is an AI startup that is revolutionizing the way contracts are negotiated, accelerating time to close, and improving deal terms. Intellext’s Intelligent Negotiation Platform™ eliminates the complexities of contract redlines and stakeholder collaboration and optimizes deal terms by applying machine learning during the negotiation process.