Finding the Best Way to Fail

Nancy and I were talking about a kind of strange newspaper article that her sister sent her discussing the upcoming release of the DSM-V (the official diagnostic manual for mental illnesses). The author of the article was a psychiatrist advocating going back to the 19th century definition of depression – melancholia. I joked that we might as well go back to using phrenology.

If you’re not familiar with it, phrenology was a diagnostic system [sic] based on the idea that the bumps on your head could tell you something about the brain structures underneath the skull. The theory goes on to suggest that the different brain structures reflect different personality traits. As a science, phrenology was discredited a long time ago – around the same time we stopped talking about “melancholia”.

But Nancy had a fascinating response to my joke about phrenology. She said (approximately):

Phrenology was actually really important because it was the first time that people started thinking about the localisation in the brain. Before that, they thought of it as a pretty undifferentiated organ – like a kidney – where each part did the same thing. So phrenology was actually one of the first steps towards modern neuroscience.

That reminded me of a great post by Randy Haykin about the Apple Navigator (which I talked about earlier here). Haykin talks about how many of the key features of the Apple iPad were first introduced in the Apple Navigator – a prototype from 1987 that never launched as a product.

The stories of phrenology and the Navigator show how both science and the economy are evolutionary processes. They both build on earlier ideas to create new ones – usually through creating new combinations. Phrenology failed as a scientific theory, and the Navigator failed as a product, but both contained ideas that could be combined with others to form new, better ideas. We learned from the failures.

That’s why a lot of people, including me, advocate developing a tolerance of failure when we’re innovating. Failure gives us a chance to learn, and it helps us execute ideas that might form building blocks of better ideas in the future. If at least some of our ideas aren’t failing, we’re not trying out enough new things.

However, failure also has consequences – something that venture capitalist Mark Suster forcefully points out in Why the ‘Fail Fast’ Mantra Needs to Fail. His key point is that when fast failure is encouraged, it can have several major drawbacks for start-ups. It can encourage poor business model development, premature abandonment of start-ups, and a cavalier attitude towards the money that others have sunk into the venture.

All of these are valid points. But I think it shows that we are using ‘fast failure’ to cover many different things. One of the key quotes in Suster’s post is this:

You want to talk about the ultimate “fail fast” – how about if you fail before you’ve spent any money building product because you validate there isn’t a big enough market or you can’t make money?

This got me thinking. I think that what we need is a taxonomy of economic failure. We can actually think of failure as a hierarchy that looks something like this:

  • System failure (the collapse of communism)
  • System component failure (stock market crashes)
  • Major firm failure (Enron going out of business)
  • Start-up failure ( going out of business)
  • Product failure (New Coke tanking)
  • Idea failure (Apple Navigator prototyped but never launched)

As you go down that list, failure gets less expensive. When I talk about tolerating failure, I’m talking about trying to set up systems that encourage cheap fast failure. This is usually at the level of ideas. I agree with Suster that encouraging failure at higher levels can be irresponsible.

Innovation courts failure. Not every great new idea will work – and since it is nearly impossible to tell in advance which ones will work and which ones won’t, we have to find cheap, quick ways to test them out. This can be done through the use of experiment as in rapid prototyping combined with iteration based on feedback, through the use of modelling or other simulations, or through the use of a screening tool like the stage-gate process.

The main point is that we need to try to encourage failure before new ideas get too embedded into the economic network. At the top level of the failure hierarchy, failure causes enormous disruption and pain, because those parts of the system are so deeply interconnected. It is much better for ideas to fail than it is for products, firms or economic systems to do so.

(photo from flickr/evansville under a Creative Commons License)

Student and teacher of innovation - University of Queensland Business School - links to academic papers, twitter, and so on can be found here.

Please note: I reserve the right to delete comments that are offensive or off-topic.

7 thoughts on “Finding the Best Way to Fail

  1. Thanks for stopping by Peter! Two great quotes – and both definitely apply here. Edison in particular is a great example of figuring out which ideas won’t work as quickly as possible.

  2. Hi Tim – phrenology was an early indicator of localization of brain functions. However, because it was so completely wrong in ascribing the discoverability of localization through contours of the skull, I’d argue it hindered the scientific discovery of localization. A great idea, expressed through the wrong metrics, becomes a point of ridicule, blinding prospective discoveries for fear of peer response.

  3. Hi George – I see your point. But it’s a bit of an open question in the end. I suppose in some ways it parallels the situation with Lamarckian evolution…

Comments are closed.