A picture of me standing at a lectern, working on a laptop computer, on the stage of the FWD50 digital government conference

Hi! I’m Alistair. I write surprisingly useful books, run unexpectedly interesting events, & build things humans need for the future.

The perils of local maxima

Preamble: It’s been nearly two years since I wrote something here. In that time, I’ve worked with a dozen or so big public companies trying to understand how to balance keep-the-lights-on, milk-the-cash-cow marketing with the pursuit of disruptive innovations. I’m trying to assemble these thoughts into something—maybe a book, maybe an online course, maybe a workshop—that represents what I’ve learned. I often talk about the perils of local maxima, so I thought I’d explain it in a bit more detail

Why isn’t Blockbuster Netflix?

Blockbuster knew that streaming video was going to happen. But the company failed to execute on this near-certain future, despite the fact that the global rental chain had tremendous advantages as the incumbent. For example, Blockbuster knew the rental history of all its customers. It knew where they lived. It knew when they watched movies. It even had their payment information on hand.

The problem was that Blockbuster was too addicted to the current state of the world.

When Blockbuster went bankrupt, two of its biggest sources of revenue were late fees and the sale of in-store concessions. Both of these revenue sources vanish when you stream movies online. This meant that the new state of the world looked much less attractive to Blockbuster than it did to a startup with nothing to lose—namely, Netflix.

Netflix, on the other hand, knew that the world of streamed content was coming. Armed with little more than a spreadsheet, its founder figured out how to get to a new business model.

There was one major hurdle any early streaming company had to overcome. At the time, North American broadband was neither fast enough, nor widely available enough, to make streaming a reality. But, convinced that streaming video was the future, Netflix dug deeper. In doing so, the company discovered that there was a ubiquitous, high-bandwidth digital network already in place: The U.S. Postal Service. By mailing DVDs to customers, Netflix figured out how to get to online streaming first.

This was a big, brilliant hack, and many of today’s big companies have a hack like this in their history: Facebook launching on campuses; Twitter using SMS before everyone had a smartphone; Uber starting with licensed limousines; Amazon focusing on a long-tail, easily-shipped commodity (books.)

The problem large companies face isn’t a lack of hacks. It’s more systemic and deep-rooted. Big companies aren’t even looking for these hacks, because they don’t see future changes as inevitable. The warning signs of tomorrow’s upheavals are everywhere, from the rise of coastlines, to the end of hydrocarbons as an energy source, to the ability of machine learning to outperform most humans at common tasks, to the ruthless efficiency of robotics and automation. Like Blockbuster, today’s big organizations don’t plan for the future because they’re addicted to the way things are.

The problem of the local maxima

We’ll come back to business in a minute. But first, it’s time for some math.

If we have a function (say, y=2x) then we can plug in various values for x, and get values for y. We can draw these points on a grid, and get a plot of that function:
Simple function

Mathematical functions often have a maximum value. This is a particular set of constants that produce the largest value. For example, if I plot the function y=-x^2, I get an arc of possible values. And the maximum value of y, in this plot, is when x=0:

local maximum

If it’s your job to squeeze the most y out of something, you’re going to get x as close to 0 in order to make that happen. You’re going to try to maximize y.

But the local maximum might not be the highest value. Functions can have more than one maximum. Let’s plot a more complicated onex^5 – 10x^3 + 30x:

Global maximum

This plot has two maxima, shown by the two red dots. It has more than one hump. Depending on where you are on the line, “uphill” means something different. And the business world, with a constantly changing landscape, seldom has one hump.

Big incumbents mandated moving uphill

Now let’s get rid of the math, and consider the worldview of Blockbuster and Netflix:

Three maxima

What Blockbuster had was a local maximum problem. It was too busy optimizing for the present to commit to disrupting the future. Everything about a big, public company is designed to optimize for the local maximum. Employees are rewarded, and investments are approved, based on their ability to move the company towards the local maximum.

Blockbuster was playing single-hump strategy in a two-hump world.

Blockbuster local hump

Everything the company did was focused on getting to the top of that hump. That’s what compensation, management, and the public markets encouraged. And from that one-hump vantage point, going downhill is bad business.





One response to “The perils of local maxima”

  1. Janine Avatar

    Great article. It really helps to explain why most large companies who once had the market leading position go out of business. The fail to continue to innovate. Innovation was what got them into the market in the first place, but instead of continuing to evaluate the market trends and innovate for the next opportunity, they become too inward focused. This happed to many companies who have either gone out of business or lost significant market share to new innovative competitors, I can think of several off the top of my head like Dolly Madison Ice Cream, Kodak Film, Xerox Printing, Wang Word Processing, and so many others.