Cycles, belts and cores: applying lessons from science to building startups
Today we have a guest blog post to share with you. Max Brodie, a VeloCity resident and founder of Inflo (one of our VeloCity Garage startups) wrote an interesting piece about startups and the value science can bring to startup life. Enjoy!
Applying scientific concepts to startups is hardly a new idea. Beliefs about problems, product and market were only assumptions until converting them into hypotheses became popular with Steve Blank’s The Four Steps to the Epiphany (written more than a decade ago). Forming hypotheses forces you to admit the existence of uncertainty – that no, you probably don’t know your true market on day one, and yes, there’s a good chance few people care about the benefits of your product  – but how exactly do we know when a hypothesis is validated or refuted? Science is full of examples where multiple hypotheses jockey for supremacy. Many modern pillars of science were at one point underdogs:
– Classical mechanics when transitioning from Aristotellian mechanics
– The Einsteinian Relativistic worldview from the Maxwellian Electromagnet worldview
– Quantum mechanics, which redefined classical mechanics
How exactly did a set of theories prevail over the other? It’s useful to consider at this point that forming a hypothesis is only a small piece of conducting research to generate new knowledge. The simple idea of a hypothesis can be unpacked and expanded into a much more powerful tool, a power tool resulting from hundreds of years of scientific refinement. We could ask ourselves: if there’s benefit in applying the idea of hypotheses to startup development, could there also be benefit in applying thinking and frameworks typically applied to research programs?
There is a well developed field especially concerned with understanding the process by which we generate new knowledge: Philosophy of Science . I contest the concepts resulting from over a hundred years of development apply to startups as well to help answer critical questions such as:
– How do you know when to pivot?
– How do we handle data that refutes a hypothesis? What actions are justified?
– Are all hypotheses equally important?
How to determine when to pivot
Thomas Kuhn  proposed a cyclical model of science, claiming all scientific activities can fit into stages in a cycle. Let’s take a closer look at each of these stages and see how they might map to the activities of a hypothesis – validating startup:
Left: a model of science, right: a potential model of key startup activities
“Normal Science” refers to discovering new knowledge that fits well with your previous discoveries – your customers are validating your hypothesis, all is good.
Now, let’s suppose your early stage startup is happily validating the hypothesis when it encounters a customer that wholeheartedly disagrees. Uh oh. We’re now at the ‘anomalies’ stage. Now what? How important is the hypothesis compared to other hypotheses? Are there dependencies? What action should be taken based on one customer’s feedback? Enter Imre Laktos and his model of the Research Program.
In order to answer when to pivot, we must first understand:
How to handle data that refutes a hypothesis
Lakatos suggests the research program (read startup) is composed of core hypotheses with a surrounding protective belt. The protective belt contains theories with few or no ‘dependents’ – that is, other theories that rely on them. These would be relatively new theories or relatively unimportant theories; the core meanwhile is essentially held to be true. It is built up over time and gradually consumes theories from the protective belt as the protective belt expands.
I believe this is a useful tool for answering our opening question of how to respond to an anomaly. If ‘Disagreeing Customer A’ attacks a theory positioned at the fringe, it constitutes a different response than if DCA were to attack a theory within the proactive belt, which is a very different response than if DCA were to attack a theory within the core. Theories at the fringe are far more malleable than those closer to the core in a linear manner . Practically, what does this mean? While a barrage against a fringe theory might be a reason to spend an all-nighter coding an alternate solution that’s split tested the next morning, any attack on the core should be taken with significant salt. For the core to be wrong, much of the protective belt would also have to be wrong. An attempt on the core probably indicates a situation where the customer’s opinion should be ignored – however, it is a successful attack on the core that constitutes a crisis and begins the countdown to revolution (pivot).
Gauging the relative importance of hypothesis
Now, we could also consider if certain hypothesis and their associated theories are more inclined to be located at the core or fringes of Lakatos’s model. Let’s consider some of Steve Blank’s startup hypothesis categories: product, customer problems, distribution and pricing, demand creation, market type, and competition. One possible standard arrangement in Lakatos’s model might be a configuration something like:
The Market type hypothesis, for example, is fairly fundamental in creating the competition and demand creation hypothesis. This idea of theory interrelatedness also plays nicely into another Philosophy of Science landmark: The Duhem-Quine Thesis. The DQ thesis places significant emphasis in understanding the environment in which any given theory stands, and notes that no theory is stands alone. Theories only make sense in the context of other theories.
The Revolution (pivot)
So we now have tools to help us appropriately react to invalidated hypothesis. Let’s say the core was successfully attacked and a pivot is in order. What exactly occurs when a startup pivots? Perhaps an epiphany – you see the data in a new way. Kuhn was often known to employ this optical illusion to demonstrate how the same information may be perceived drastically differently:
(Out of curiosity, leave what image you saw first in the comments. I have a theory about why we may be disposed to seeing one option first)
Did you see both versions of the image? The moment you realized the existence of the other animal could be considered an epiphany. While the process by which that exactly occurs could be a much larger conversation, the important thing about revolution is that it embodies a tipping point – the point at which what was done before is no longer relevant, or if it is, relevant only to the extent of supporting that which lies beyond the revolution. Revolution in the context of Kuhn’s model parallels pivoting in the context of lean startup.
Startups are often seen as fuzzy and ambiguous, the sum of sixteen hour work days and luck. I strongly believe that scientific logic can come into play when dealing with startup strategy. Startups don’t need to be all about spontaneity – applying external knowledge about the workings of hypotheses can streamline their improvement.
– Max Brodie Steve outlines a few interesting and well known product failures where fundamental assumptions were not properly validated by customers:
Webvan: Groceries on demand – the killer app of the internet. The company spent money like a drunken sailor. Even in the Internet Bubble costs and infrastructure grew faster than the customer base. Loss: $800 million.
Motorola’s Iridium satellite-based phone system – engineering triumph and built to support a customer base of millions. No one asked the customer if they wanted it. Cost: $5 billion. Yes, billion. Satellites are awfully expensive.  The Philosophy of Science basis of my commentary originates from Dr. Katie Plaisance’s Knowledge Integration course, “The Nature of Scientific Knowledge” at the University of Waterloo.  The term “Paradigm Shift” has achieved a height of cliché perhaps in eclipsed in tech jargon. It’s emergence as a buzzword is riled far and wide, to the point of articles referring to it as “abused and overused to the point of becoming meaningless”. Yet – it’s original significance was never rooted in describing the world changing effects of technology or modern communications, but rather was a technical term coined by Thomas Kuhn to describe a stage in his model.  It’s worth noting that in the very beginning of a startup’s life, there is not much of a protective belt surrounding the core and the small cluster of theories that do exist could all be considered fringe.
A huge thanks to Max for all of his insight – we hope this has inspired you to start thinking about startups from a new perspective.
With So You’re A Startup this Thursday and Demo Day next week, we have a lot of blogs lining up! Be sure to check here, on VeloCity’s Facebook, Twitter, Pinterest, Google Plus and LinkedIn Group to keep up with us!