Success and Cognitive Bias
One thing that often strikes me about conversations regarding start-up success is the pervasiveness of the narrative fallacy and hindsight bias.
We can go to Wikipedia’s entry on Taleb for a definition:
Narrative fallacy: creating a story post-hoc so that an event will seem to have an identifiable cause.
Allow me to illustrate. What caused YouTube to grow at phenomenal rates in 2005/2006, eventually leading to a $1.65 billion acquisition by Google in 2006?
Was the cause:
a) YouTube had a better UI/UX than their competitors?
b) YouTube, at the time, decided to largely ignore blatant copyright violations (unlike their competitors)?
c) the External Player and the viral widget-ization of their platform? (Obvious, right? We’ll come back to this in a moment.)
d) YouTube was simply in the right place at the right time? (Growing popularity of social networking, ubiquity of broadband)
e) YouTube made novel and smart technology choices by adopting Flash 7?
f) Mix and match any of the above.
g) None of the above. Other.
All of those reasons pulled from a case study story appear entirely plausible, even logical, to me.
That noted, I don’t how accurate any of them really are. Did/does YouTube have a better UI than the rest of its video-sharing brethren? Were they the first/only to encourage video-embedding? Did they have the best underlying technology?
Even if we assume these to be accurate, we still don’t know whether or not they had any sort of causal relationship with YouTube’s growth.
Back to the original question: What caused YouTube to grow at phenomenal rates in 2005/2006?
My answer: I don’t know (BTW neither does Brant).
Viral Coefficients and Viral Cycle Time
Just recently I came upon an excellent and interesting post about YouTube, Viral Coefficients and Viral Cycle Time. After a persuasive discussion of why one should minimize Viral Cycle Time, the author (@BostonVC) notes:
This [an extremely short Viral Cycle Time] explains why YouTube exploded at a faster rate than ever seen before.
Perhaps I am being overly pedantic, but I think it might be more accurate to state:
This explains how YouTube exploded at a faster rate than ever seen before.
In my eyes, the distinction between ‘why’ and ‘how’ is non-trivial and needs to be recognized to avoid distortion by the biases hardwired into my brain. ‘Why’ implies causation, while ‘how’ implies manner. And I am of the opinion, that we don’t know why YouTube popped, but we may know how it did. Again, I repeat, this is non-trivial. Knowing the difference goes to the heart of the Customer Development methodology.
Please understand me, I am not saying that any of the possible causes I listed above may not have caused YouTube’s spectacular growth, I am simply saying we don’t know if they did. Not only that we don’t know, but pretending to know, while natural, is not cost-less. As start-up founders in efforts to duplicate such astronomical success/growth, all the while pretending to know its true cause, we are prone to draw the wrong lessons, likely to result in wasted money, and more importantly, time. Not to mention the proverbial blood, sweat and tears.
This is why I am of the mind that founding (and investing in) a start-up with the intent of enjoying a future extra-normal financial success is best likened to maximizing one’s exposure to positive Black Swan Events.
As defined by Taleb:
What we call here a Black Swan (and capitalize it) is an event with the following three attributes.
First, it is an outlier, as it lies outside the realm of regular expectations, because nothing in the past can convincingly point to its possibility.
Second, it carries an extreme impact.
Third, in spite of its outlier status, human nature makes us concoct explanations for its occurrence after the fact, making it explainable and predictable.
I stop and summarize the triplet: rarity, extreme impact, and retrospective (though not prospective) predictability.
One could argue that startup founders are trying manufacture positive Black Swan Events, but if we accept Taleb’s definition, this would be a contradiction in terms. (Nivi at Venture Hacks writes about how raising money for your startup is tantamount to searching for Black Swans.)
For the moment, let’s assume Taleb and I are right. We cannot manufacture Black Swan Events, so then how do we, as startup founders, maximize our exposure to positive Black Swan Events?
First and foremost, I posit that we would be better off by thinking about startups through such a lens and accepting the sad fact that we, as humans, are born with cognitive biases (such as the narrative fallacy and confirmation error).
These cognitive biases are evolutionary adaptations that served us well for day-to-day activities prevalent during the Paleolithic, but unfortunately, aren’t appropriate for understanding the nature of technology startups.
Second, by rigorously but not dogmatically, employing Customer Development framework and methodologies which are intentionally constructed such that they correct for and recognize the existence of these distorting cognitive biases.
As Steve Blank writes:
The mistake isn’t having a vision and taking risks. The mistake is assuming you are a Black Swan and continuing to ignore the facts as they pile up in front of you.
Not to mention ignoring a lack of facts as well!
If you haven’t read The Black Swan yet, I suggest reading it (a thoughtful review can be found here). The implicit philosophical overlap and synergy between The Black Swan and The 4 Steps to the Epiphany are stunning.
In the meantime, I hope you have prosperous 2010 and hope you are exposed to a positive Black Swan Event sometime soon!
Very good observation and analysis. Other biases are Optimism bias – starting a venture despite the odds and Downward Counterfactual
Outstanding! Great tutorial on Black Swans!
I am a great fan of Black Swanism. Problem is, I am not sure it is actually of much practical value – other than in following the advice of Taleb on investing: Minimize your exposure to (negative) black swans by holding -mostly- riskless treasuries, but, still, have some exposure to positive black swans (by, if you are, as I am, attracted to the Double-Dip Recession hypothesis, buying cheap, far out of the money puts on SPY’s with a few percent of your money).
I don’t see, though, a direct application of Black Swanism to Startupism. It seems to me that a startup is a whole-hog bet on the come, knowing that the odds are against you. A startup does not provide much opportunity for hedging.
The more indirect Swanian exhortations to see facts as they are, to avoid Monday Morning Quarterbacking and to not mistake coincidence for causality are deeply apropos everywhere in life, but, since the folks most in need of this advice will more than likely not be readers of Taleb, we are somewhat back where we started… Sigh.
Bill
@Bill
Thanks for your comment. You write, “I don’t see, though, a direct application of Black Swanism to Startupism.”
Unfortunately it looks as if I have been neither persuasive nor eloquent enough to convey the application/understanding of Black Swan Events to startups. To wit, YouTube is the epitome of a positive Black Swan Event, given that it meets Taleb’s criteria:
rarity, extreme impact, and retrospective (though not prospective) predictability.
And my point is that YouTube often seen as anything but.
IMHO, knowing this (and of our cognitive biases) is of great practical value.
There are people working on “hitting” more positive black swans like youtube. Check out http://www.rafefurst.com and Kevin Dick here,
http://emergentfool.com/2008/10/31/what-im-working-on-supercharging-innovation/
I am having a little trouble following the thread of your exposition. What are the top three to five lessons learned that you take away from YouTube? I understand that you can’t explain their growth. Is there anything we can learn from what happened? Can we even learn what actually happened?
I am having a little trouble following the thread of your exposition.
@Sean Murphy
Completely understood. I am not terribly happy with the way this post turned out.
What are the top three to five lessons learned that you take away from YouTube?
I don’t know. Are there any? My point is that we are better off admitting we don’t know than pretending to know and applying made-up “lessons” to our own startups. Sometimes the lesson is that there are no lessons.
Can we even learn what actually happened?
IMHO, I think if any of us knew the actual cause of their growth and whether or not it was easily replicable, our present net worth would be a lot greater.
The reason I used YouTube as an example is that they had witnessed insane growth and people drew what I think were the wrong “lessons”. For example, from a comment on the LSC Group:
“we are struggling with the same issue on lead/buzz generation.
We are open to “failing fast”, but not before we have seen real user activity and corresponding dropoff after that.
I heard interesting stats that in early YouTube days they found that only 2% people loaded videos, 15% commented and rest just watched.
So the 2% “seeders” is what we need to get at.”
Really? Even if that is true, is it causal/relevant? And how did this person come up with this “strategy” — IMHO, by looking at a Black Swan and making up a story that seemed to make sense.
Does that help?
What I take away is that you believe YouTube “won the Lotto.” It’s a one time non-replicable event.
I respectfully disagree, at least from my perspective it was the result of dozens if not thousands of decisions, not one or two lucky ones. I don’t know that I have puzzled them out but I think there are lessons there.
My sense is that they picked a format the looked OK for lower bandwidth than competitors had selected, they made it easier to contribute than existing competitors, and they could scale infrastructure successfully when they got their first few growth spurts. But these are my hypotheses and I wouldn’t promise you I could build another YouTube.
Also, where I think you could predict that auctions would be a “winner take all” kind of application/market-space because people selling goods want to have as many bidders as possible so they gravitate to the largest site, it’s not clear that there are “winner take all” properties to showing video.
As to your 2% seed, 15% comment, 83% watchers I have seen roughly similar figures for other communities:
1-4% original content
10-16% comment/rate/tag
80-90% lurk/read-only
There are certainly interactions/incentives to be monitored or managed between seeders and commenters/raters/taggers and even the decision to display views or other “lurker activity” is probably encouraging to both “seeders and commenters.”
Net net I think YouTube’s success was not the result of one or two key decisions or a flash of genius. I don’t think it was a Black Swan that we can’t learn from, but I freely acknowledge that understanding what happened and deriving some useful rules of thumb for other on-line communities is a very complicated undertaking.
at least from my perspective it was the result of dozens if not thousands of decisions, not one or two lucky ones.
@Sean Murphy
That is not mutually exclusive with what I am arguing. I picked a set of causes that I have heard people refer to often.
My sense is that they picked a format the looked OK for lower bandwidth than competitors had selected, they made it easier to contribute than existing competitors, and they could scale infrastructure successfully when they got their first few growth spurts. But these are my hypotheses and I wouldn’t promise you I could build another YouTube.
Sure, sounds reasonable to me. As much the causes I listed above sound right to me.
BTW Take a look at the video I linked to where one of the founders talk’s about the growth of YouTube. I don’t he makes any strong claims as to the causes of their spectacular growth.
What I don’t buy is how history has been “re-written” and now we indulge in retrospective (though not prospective) predictability.
As to your 2% seed, 15% comment, 83% watchers I have seen roughly similar figures for other communities:
1-4% original content
10-16% comment/rate/tag
80-90% lurk/read-only
There are certainly interactions/incentives to be monitored or managed between seeders and commenters/raters/taggers and even the decision to display views or other “lurker activity” is probably encouraging to both “seeders and commenters.”
Of course, that those sort of stats should be monitored and managed! But I think you have miss my point, namely, that those stats may be an indicator/measure of engagement, rather than a cause of engagement and/or growth. My point was that this individual was clearly using YouTube as a template to follow when they didn’t know if this was applicable or causal to success YouTube’s success:
We are open to “failing fast”, but not before we have seen real user activity and corresponding dropoff after that.
I heard interesting stats that in early YouTube days they found that only 2% people loaded videos, 15% commented and rest just watched.
So the 2% “seeders” is what we need to get at.”
Maybe this person should employ CustDev techniques to figure out why they have not seen any real user activity versus widening the funnel/increasing throughput and just blindly pushing to get 2% seeders?
Net net I think YouTube’s success was not the result of one or two key decisions or a flash of genius. I don’t think it was a Black Swan that we can’t learn from, but I freely acknowledge that understanding what happened and deriving some useful rules of thumb for other on-line communities is a very complicated undertaking.
Fair enough.
The point about “winner take all” was that I didn’t see path dependencies ( http://en.wikipedia.org/wiki/Path_dependence ) at work to create a single large winner. Although it could be that viewers want to go where there is the largest variety than puzzle out which of a dozen specialty sites will have what I am looking for.
If there are some strong path dependencies or ‘winner take all’ effects at work then YouTube is not so much a Black Swan as inevitable in the that I believe a single large auction site (that happened to be Ebay) and a single large on-line bookstore (that happened to be Amazon) were.
But even in that situation the sites have undergone considerable evolution (many decisions, not just one or two lucky ones) and there is a lot of wisdom that can be gleaned from their current user interface and mechanism design.
@Sean Murphy
A lot of things look inevitable after the fact. Not to get to philosophical, but the way history is most often taught, at least to my mind, is imbued with false inevitability.
A lot of interesting work has been done in using bond prices as proxies for prediction (prediction markets, you may have heard the term).
From http://www.midasoracle.org/2007/04/23/black-swan-predicting-world-war-i/
…Ferguson is intrigued by the behavior of the financial markets on the eve of World War I because stock and bond prices at the time registered scant concern about the impending cataclysm. This contrasts with the conventional view among historians that the war was all but preordained because of a decade of escalating great-power rivalries that erupted into violence after the assassination of an Austrian archduke by a Serbian terrorist in June 1914
[That is how I remember it being taught to me.]
Ferguson was most surprised that the people who had more to lose from a war — bond investors — didn’t see it coming (…).”
When Nassim Taleb spoke in New York last week he mentioned how, according to the analysis by Ferguson, war bonds did not anticipate WWI. Consistently, Taleb wrote in his reaction to a New York Times review of “Black Swan”: “(…) I gave a cohort-style exhaustive account of the four highest impact events in the history of the last two thousand years–the two great wars, and the rise of Christianity and Islam –and showed that they were unexpected based on contemporaneous accounts. If, as Niall Ferguson showed, war bonds did not forecast the great war, it was a Black Swan (…).”
Swans sound like heavy-tailed distributions to me. There are so many distributions to chose from, but we do most of our predicting based on the normal distribution, which really is very rare, and thus very wrong.
We can predict prospectively, but we cannot predict when. Get used to an asynchronous clock. You can always decide that an even has arrived and act accordingly even if it hasn’t. That by itself mitigates the negative event.
In installed software, the transition to late market killed public companies. Growth from their initial product line is over at this point, so the investors bail out. It should be obvious when a company has consumed half of its market, but somehow no one measures this, so surprise they miss a quarter. Is it going to happen? Absolutely. When? Depends on how fast you are consuming your market.
@David Locke
I am not as sanguine as you regarding the accuracy of our predictive powers. 🙂
If you haven’t read The Black Swan, I highly recommend it.
This is a well written article and there is a lot of truth to it.
HOWEVER.
The irony here is… that people, INCLUDING investors/angels/incubators are such sheep that they will very easily have a black swan Blatantly right in front of them… and will ignore it in their search for the perfect ABC silicon valley model of success. The ideal ‘team’, the ideal ‘pitch’ the ideal idea.
How do people come up with the ideal pitch?? Well they plan What People Want to Hear you fucking idiots. You know this.
Yet you sheep want to be ‘dazzled’, you want to be given what you want and expect to hear. Because that is makes feel validated and comfortable. So while looking for those golden swans in the lake amongst the predictable white… the black swan sits there completely alone and stunned at everyone else’s sheepish and predicatable behavior.