The (not) unreasonable effectiveness of negative discovery

A technique often used by mathematicians for proving a statement is contradiction. The thought experiment is relatively straightforward: if the opposite of the statement you are focusing on leads to inconsistent, contradictory, or absurd results, then the original statement must be true.

The Four-Color Hypothesis

This approach is not as common in other domains of science, but has historically served mathematicians well, occasionally even resulting in the discovery of new math. An example is the "four-color hypothesis": first proposed in 1852, the Four-Color Hypothesis states that any map drawn on a sheet can be divided and colored with four (or fewer) colors so that no two contiguous regions present the same color.

To validate the hypothesis, mathematicians tried to create maps where two or three colors were not sufficient, and maps colored with five colors that always had neighboring regions do not have the same color. So, essentially, the problem was solved by eliminating the incorrect solutions: three colors are too few and five are more than enough, with the consequence that the remaining question (four is just enough) had to be true. Decades later, in 1976, the problem was finally solved directly with the aid of computer calculation by researchers at the University of Illinois, and it's now known as the Four-Color Theorem .

Process Discovery

An emerging field of application of negative discovery is data science. Process discovery is a learning technique that, given a set of training examples, has the goal to derive a process model which encloses the behavior underlying the training set. Process Discovery helps organizations identify business processes, record all possible variations using machine learning algorithms and make recommendations for automation.

Process discovery relies on various techniques, and usually employs machine learning-based tools for deriving models; in a 2018 paper - Incorporating negative information to process discovery of complex systems - the authors have explored the use of negative information in process discovery. Process Discovery distinguishes business processes that can be automated, but also help design automation workflows, making the mapping, planning and implementation of automation quicker and more efficient. The researches discovered that by focusing on the parts of the process that can NOT be automated, and rule them out, they could generate significant benefits, especially with a reduction in the complexity of models, without trading off in any of the four quality dimensions:

• fitness?(ability of the model to reproduce the traces in the event log)
• precision?(ability of the model to avoid reproducing undesired behavior)
• generalization?(ability of the model to reproduce desired behavior not found in the event log)
• simplicity?(the well-known?Occam’s Razor?principle).

Negative discovery and product incubation

This approach can be very efficient when dealing with very ambiguous problems, such as product incubation and business acceleration, where usually the common path taken is trial/error. The idea, again, is to focus on ruling out what does NOT work from the very early stages of validation: what customer expectations where not met, what features weren't used, what problems weren't solved, what types of customers weren't the right fit, and so on.

If a company wants to successfully take advantage of opportunities that fall outside of its normal strategic boundaries, it needs to be able to experiment in a non-deterministic, non-linear way. This means being able to tweak the design of a new opportunity, as well as challenge and reshape its underlying value proposition and business case throughout the development process. In order to successfully pursue strategic innovations, companies need to adopt an iterative cycle of experimentation, testing, learning, and adapting until they get it right (or decide to abandon the project altogether). This period of experimentation requires a certain mindset, specific behaviors, and a methodology that includes the appropriate tools. Additionally, companies need a guide to help them determine if they are on the right track. This is what is known as 'incubation'.

Incubation is the period between having a well-reasoned concept and an actual artifact (product, service, platform, business model, etc.) that can thrive in the world. The reason for having an incubation period is to design and conduct experiments (and use the results) to get as close to a true, valuable manifestation of a new artifact as possible. Negative discovery is vital during incubation, as it helps to navigate through the potential risks and uncertainties associated with the project. This is especially true for businesses that are looking to create something that is completely new and different from anything they have done before. It is often difficult to achieve success due to the lack of experience and expertise but, by carefully planning and considering all the potential risks, chances of success can be maximized.

Disclaimer

What's above represents?my personal views?and not the opinion or policy of my employers or any other company, organization or individual I can be associated with.

The facts expressed here belong to everybody, the quoted sections, stetements, and imagery belong to their respective authors, the opinions only to me: the distinction is yours to draw...