A Tale of Three Startups… and how they mess with our brains

Do you recognize this story?

It was the best of times, it was the worst of times.  Entrepreneurs we’ll call “Alan,” “Ben” and “Colleen” started competing companies around the same time.  In the chart below you can see their competitive rises and falls over a period of several years.  The X axis is time. Consider the Y axis each firm’s level of overall success at any point in time.

Alan Ben Colleen

Chapter 1

At first, Alan, a normally shy and awkward guy, somehow signed up some huge, sexy customers that gave him a big early lead.  Ben (in second place) kept changing his strategy, making him look unfocused and scatterbrained.  Colleen was in last place.  She couldn’t land any big deals and had to make ends meet with some oddball customers in weird little niches.

Chapter 2

Alan was still the market leader, but Ben finally got his act together and was gaining momentum.  While Ben’s company didn’t have Alan’s fat profit margins or big customers, Ben was scrappy and started spiking up to take deals from Alan by low-balling on price.  The two became ferocious competitors and traded blows almost daily.

Chapter 3

One customer at a time, Colleen grew from a sapling to a mighty oak.  Before Alan and Ben knew it, Colleen overtook them and captured the bulk of the market.  Colleen remained the market leader thereafter – by such a wide margin Alan and Ben would never again be more than obscure, bit players.

Who does this story remind you of? Google, Yahoo and AOL? Borders, Barnes & Noble and Amazon? Blockbuster, Netflix and Redbox? Someone else? Take a moment. Think about it.

I took several artistic liberties, so this isn’t really a fair exercise, but if you’ll forgive me I did it to illustrate a point.  While the characters in this story (Alan, Ben and Colleen) were “real” in a sense, they weren’t actual companies per se.  They were virtual bots in a computer simulation.

We’re researchers who predict business survivalor failure using data science.  We use simulations and other tools to both predict what’s going to happen, and to study the way market phenomena work.

If all you have is the graph above, it’s surprisingly hard to tell the difference between a real-world business case and a simulated one.  Simulations can be quirky, textured, unpredictable and surprising – just like real life.  Run the same model twice and you often get a different result.  Just as in the story here, simulated businesses seem to fight, have egos, get distracted and behave opportunistically as you watch their plots unfold.  Other times companies can appear lazy, shortsighted or even confused.  As we watch virtual businesses compete it’s almost impossible to resist our human impulse to see personalities develop, to judge their actions and to yell at them when they do something stupid (as if they can actually hear us).  If you were listening to us watch a simulation you might think it was the Super Bowl.  We don’t get out much.

Of course the virtual bots don’t have personalities or egos.  They’re just mindless little calculations.  On one hand, we know this.  On the other hand, it’s easy to forget.

Human tendencies to overweigh personality and visceral drama is more than a triviality; it often holds back scientific discovery.  What if business success or failure is often driven by subtle quantifiable dynamics, played out to logical extremes?  Before you dismiss this idea offhand, consider this: if I told you the graph for Alan, Ben and Colleen was from three actual startups, could you tell the difference?  Would your intuition detect deep mechanical forces or is it more likely to gravitate towards the more animated storyline?

Skepticism aside, simulation techniques like these have led to far more accurate “real world” business predictions than most people fathom.  That’s just a fact.  No two ways about it.  We’re not saying humanity, personality or emotional drama aren’t important factors in business.  They most certainly are.  I’ll say it a third time – qualitative stuff can be really, really important.  However our research suggests those variables are frequently a lot less predictive than most people assume.

This isn’t about defending any one specific model. Rather, it’s meant to pose a question (albeit a heretical one in many circles): do you think it’s possible that businesses succeed or fail largely due to market forces beyond what human drama and qualitative anecdote adequately account for? If you scratch below the surface of anthropomorphic narrative, might there be highly powerful, subtle mechanics at work? If so, might these forces be obscured or even concealed underneath juicy plot lines? Our research indicates this is very much the case, but we’d love to know what you think.  As much as we try to mimic humanity with our bots, sometimes they’re the ones teaching us what it means to be human.

This Post Has 6 Comments

  1. Ron

    All I can say is WOW Thomas. F@#ing awesome!

  2. Árpi Jakab

    Even a broken clock is right twice a day. How accurate is the average prediction? Are you transparent about the factors considered?

  3. Mike

    Just saw an article about these guys work in Fast Co. http://www.fastcolabs.com/3021903/this-prediction-algorithm-can-tell-if-your-startup-will-fail
    By all accounts it’s the real deal. Rumored accuracy more than 80% and heavy on the statistics so seems a lot better than a broken clock, but very secretive about the special sauce. Obvious use in venture capital so I wouldn’t share either if I were them. I guess if we can simulate weather and atom bombs and biological systems we can simulate business too

  4. Logan

    Human psychology has its biases and this shows some of them. We like neat and tidy anecdotes however this usually does not adequately represent factual matter. It is just an educated guess with coherent story power, which are often wrong but sometimes right. It takes science research and statistic follow through to know the difference. Most people don’t do this. Nice article.

  5. Tod

    Very interesting simulation. The thing that strikes me is that from the initial start up you’d have to have the guts and stamina to be in the game a long time to invest in Colleen. My guess is that most ROI models, perhaps even with a 7 year outlook, would lead one to invest in Alan.

  6. Thomas

    Great point Tod! I hadn’t thought of that… really provocative observation

Leave a Reply