Consider a young high school couple. Odds are they won't get married; maybe the odds are 5%. If you have to bet whether they will get married or not, you'll make money in the long run by betting that they won't. However, if you have to bet on the single most likely person in the world for either of them to marry, their current partner is surely your best bet.The same is true for knowledge. Paradoxically, even though we can safely predict that any given piece of knowledge will eventually be overturned or rendered obsolete, we can be almost sure that any given piece of knowledge is our current best bet.
Knowledge has a half-life
I recently read Ronald Bailey's article on Samuel Arbesman's book, The Half-Life of Facts: Why Everything We Know Has an Expiration Date and was sufficiently taken by the idea that I bought and read the book. The idea is a compelling one: just like half the amount of a radioactive element will decay during its half-life, in practically any field of knowledge, half of what we know will be replaced after some predictable amount of time.
In this post, "knowledge" refers both to facts that change over time or are proven wrong, and conventional wisdom or best-practices that are overturned.
Consider the following knowledge that was at one point considered true:
The Sun orbits the Earth. DNA has 48 chromosomes. Cigarettes are not bad for you. There is a dinosaur called the Brontosaurus. Red wine is bad for you. I mean good for you. I mean bad for you.
What gives one body of knowledge a shorter half-life than another?
I would argue there are four main characteristics that give a body of knowledge a short half-life.
- The field is young and poorly-understood.
- There is a limited amount of data from which to draw conclusions.
- The knowledge is difficult to test and variables are hard to isolate.
- The knowledge is more best-practice than fact.
Modern medicine has been around for less than a couple hundred years, the knowledge is part of a chaotic system where variable isolation is often impossible, and much of the knowledge is best-practices for treatment rather than provable facts. Medical knowledge can have a half-life as short as 10 years.Nuclear and plasma physics are even less mature and well-understood, the data is hard to collect, and complicated conclusions are more often deduced than observed. For these fields, the half-life of knowledge is only 5 years.
This lead me to wonder: how does startup knowledge stack up? Probably not very well.
The half-life for startup knowledge must be very short
The startup industry in its current form is relatively young and poorly understood. Compared to other fields, there is a small amount of data about cause and effects for startups. What's worse, the data exists in an incredibly chaotic system where isolating individual variables is often impossible.
In other words, most of today's startups knowledge is probably wrong.
There are two type of startup knowledge that I'm talking about here, which I would describe as startup facts and startup wisdom.
Startup facts are just that; cold hard facts that are relevant to startups. These can lose their relevance because they are either proven untrue, or more likely, because they change over time. Examples of startup facts include:
Price of data storage. Number of people that own mobile devices. Government regulations.
Startup wisdom refers to generally accepted knowledge about what startups should do to succeed, a set of best-practices analogous to the generally accepted treatment methods in medicine. Current examples of this include:
Fail fast and iterate. Raise 12-18 months worth of runway. It's easier to enter an existing market than create a new one.
How likely does it seem that all of these insights will stand the test of time? The odds are pretty low.
All startup knowledge is susceptible to a half-life decay, and is turning over at a rapid pace. Although there will never be unanimously accepted best-practices for startups, consensus changes at a rapid pace. Similarly, trends like Moore's law and global mobile penetration are rapidly making yesterday's knowledge obsolete.
If today's startup knowledge will be obsolete tomorrow, how should founders proceed?
I think there are at least two takeaways from these (far from perfect) thoughts:
1) Founders should realize that the startup industry is immature and constantly-changing, the knowledge tends to be more best-practice than fact, the data exists in a chaotic environment and variables that are difficult to isolate. In short, founders should realize that the half-life of startup knowledge is short.
Of course, just like the high school couple, today's knowledge still represents the best guesses we currently have. Even though we know it's more likely than not that any given piece of current knowledge will ultimately be supplanted, this is still where we should place our bets unless we know otherwise. However, founders should never be afraid to challenge assumptions, because odds are the assumptions will turn out wrong.
2) Perhaps even more interestingly, it follows that good ideas for new startups may be found at the place where knowledge is overturned. Peter Thiel effectively made this point when discussing the idea of secrets at Stanford; startup ideas may exist where you believe a truth that most people do not.
As data storage became cheaper and security improved, Box gambled that enterprise companies would keep their data in the cloud. HelloSign bet that e-signatures could work. Founders should try to see around corners to anticipate what knowledge today will be obsolete tomorrow, and find their place accordingly.