What Makes Google Great? Understanding The Difference Between Bounded and Unbounded Complexity Problems

Value flows from knowledge. This has been taught to all of us since we were kids, and it’s the basis of our drive to educate ourselves and to pursue information – something that all professionals do automatically. The value of our businesses is similarly derived from our knowledge– knowledge gleaned from the markets we work and compete within. Figuring out the “Truth of your Market” is always the key to business – beneath the day-to-day operations. Look at what Nike, Apple, and McDonalds have all done: They have figured out compelling versions of the Truth of their Markets and thrive because of it. 50 years ago who knew that athletic shoes or quick-service hamburgers could be multi-billion dollar markets? The more the uncertainty, the greater the likelihood that your competitors have not figured it out, and many times, the more valuable the solution.

Product selection, features, messaging, marketing media and timing

Most of the uncertainty in commerce is found in the interface between a business and its customers; particularly at the interface between a business and customers that are not yet customers.  For managers and entrepreneurs, it is very useful to understand the geometry and structure of the fields of uncertainty that we are solving for, and there are 2 kinds:

Bounded Complexity and  Unbounded Complexity.

Bounded complexity means a field of uncertainty that has limited scope, the details of which can be progressively reduced asymptotically to a boundary of relative certainty. When doing a startup, bounded complexity problems are commonly found in hiring, legal agreements, operations, and engineering. For example, engineering teams do incredibly complex work, however we can easily see the progression from uncertain to certain over a relatively short period of time. The most common software engineering and product questions can be solved for in principle by an engineer with a pen and a whiteboard in the space a few hours (if not a few minutes). A problem is defined, ideas are sketched out, the problem is broken down, and connections are drawn. An abstract problem (building a piece of software with defined features – like LinkedIn or Facebook) can be described succinctly and in a short time from a technical basis. This rapid burndown from uncertaintly to a known process is indicative of a bounded complexity problem. It means that in the grand scheme of things, the problem at hand is likely to be relatively simple, and thus does not represent a fundmental challenge for a well-constructed team of people to solve.

The other type of complexity system is the Unbounded Complexity system. Unbounded complexity systems are those which defy easy restriction and do not easily converge to a boundary of certainty (even despite great effort.) Common examples of this are product definitions, customer segmentation, and market messaging.  Remember the LinkedIn and Facebook examples above? The engineering part is easy – but determining what to build is not. The geometry at play within these systems is really interesting, because the march from unknown to certain is very difficult, and feedback is expensive to gain.  In choosing a product, there are many questions: What do people want? What will the buy? How much will they pay for it? What message will resonate the best? Who is the audience? How can we connect with the right people at the lowest price?

What we begin talking about in exploring these fields of uncertainty is the challenge of optimization. Optimization problems in natural systems like the customer population of an industry or demographic groups online can be incredibly complex.This is true in part because of the difficulty of obtaining feedback – Getting quality data takes time and money, and will always be somewhat limited. (Even Google does not have enough data to efficiently solve many of the problems they are trying to optimize, and they are arguably the most data-rich organization in the world today.)

When we are asking key business questions like “What to sell, with what features, with what messaging to describe it, etc” we find that we are essentially unlimited in the number of options that are available to us. What’s more, unbounded complexity fields will often be non-static. The audience of consumers today is going to be different in a week or month. So companies are often solving against unbounded complexity problems with limited information and a target Truth that is always in flux. That means that even gaining a highly optimized set of answers for what people want or how to message them (itself being a highly difficlut proposition), there is a built-in half-life to whatever we learn. In a world of rapid change, our conclusions today will likely be less true tomorrow, and less the day after that.

As a startup founder, it is very common to advise folks about this distinction between bounded and unbounded complexity problems because a proper framing of the challenges of a business is required for companies to get where they want to be. Because the bounded type of problems are much easier to get a grasp on, they often become the focus of a team’s effort.  Even multimillion dollar companies that I have worked with have frequently fallen into the trap of focusing on things likeengineering, and will push back the more pressing questions of ‘what to build, what to sell, who to sell it to, how much they can charge, etc.’  While companies often push back these questions, or just regard them as ‘settled’, the optimized answers to these business pillars are rarely ever really well understood. Companies know what they know — and don’t know what they don’t know. Paths to huge profitability, innovation, and competitive advantage are out there, but invisible until uncovered by exhaustive and constant exploration. Constant exploration is really the wellspring from which profit and revenue flow. Tracing back to where a startup’s profits come from, we will see that the dollars come from understanding the dynamics of the market, the needs of clients, and how to provision unique value.

So what is your business doing to learn what it does not yet know about its market? It is likely a combination of easy and hard problem solving. The best managers recognize early where they are putting their effort, and realize that bounded complexity problem sets do not drive profit in most cases, despite the fact that they are often so much more easily seen, more easily discussed, and more easily acted upon.

Google is a great example of a company that understands the value of data and feedback, as they are constantly putting forward new product ideas and gathering data to support or invalidate them. Google is a nothing if it is not a continual series of hypotheses in action. A big question like “How can we get the maximum profit from advertisers” had unbounded complexity and is hard to approach directly: So Google creates a series of bounded complexity problems and solves for them in a series. Product ideas like “Customers will love to see a map of local results instead of web search results when looking for hotels” (a bounded complexity problem) can be designed, set in front of users, and validated in weeks or months. By running thesebounded complexity problem-solving exercises in parallel across their business and in a neverending series, Google can make great traction into solving the unbounded optimization problems confronting them in their global, rapidly changing market. This is how Google has become the company it is today – by constant and structured exploration. The other stuff like revenue, fame, glory, and international influence are mere by-products of getting the exploration bit right in their business and learning how to truly give users what they want. Does your company do the same?