The average number of founders in successful startups doesn't mean much unless it is from a random sample of startups that have both succeeded and failed.
If you looked at a graph of the total dataset, what you'd see is that almost all startups with only one founder fail. So the reason the successful startups you list mostly have two founders may not be because two is the optimum number, but rather because two is the next smallest (and therefore most common) number after one. We can't tell without additional data.
What I find interesting, is that a comparatively large number of companies on that page had only one founder, even though the odds seem stacked against them. Maybe there are substantial advantages to being alone. For one, you avoid having to design by committee.
This article could be a chapter in the book How to Lie with Statistics.
It is clearly a biased sample, listing only "successful" companies but not the much larger set of unsuccessful ones. The "exception" of Excite thrown in with 6 founders, IMO makes the author's intentions (conscious or unconscious) clear since one would expect some argument about whether Excite was really a success in the end.
Then there is the question of what constitutes a founder. This isn't always so easy to define. How many people founded IBM?
Despite Thomas Watson being listed as the "founder" of IBM, it is hard to even answer this question since IBM was formed out of the merger of three companies before it was even called IBM. (http://www.answers.com/topic/history-of-ibm)
Several responders pointed out correctly that the number of potential communications channels rises as the number of founders increases. And while that's clearly true, it is also true that more founders means more mouths to feed early in the company's development. And to run any business you need to learn to manage communication amongst your team.
Finally, one has to be careful not to confuse correlation with causation. Just because Microsoft was founded by two people, doesn't mean that was the primary reason for their success. It might also have had something to do with who the two people were after all. ;)
It's not even a list of "sucessful" companies. It's a list of well known companies. I'm sure there are plenty of successful companies piloted by only one founder or more than 2 founders. The sample isn't statistically kosher if you will. Success doesn't depend on the number of people founding the company. It depends more on the idea, the talent, the will and the execution of the idea.
We find two or three works best. One advantage of three is that you have a third person to defuse any dispute between the other two. But if the two founders have known each other long enough, like Emmett and Justin, you may not need that.
What I find interesting isn't so much that the mean is around two, but how strongly the median is two. Are the benefits of having two founders over one (soundboard off each other, keep each other motivated) and problems of lack of focus and communication overhead of 3+ founders so strong that a startup of two is more likely to succeed, or is two just the most natural size for a startup to condense at?
I guess the other possibility is that the sample in the article is skewed to make a point...
The difficulty of keeping everyone informed increases greatly when you go from 2 to three people, as you go from one communication channel to three. A small programming team can take advantage of easy communication to have better efficiency over a large team, and I imagine that the advantages are similar for all of the tasks in a startup. Because a startup is desperately trying to make itself successful before running out of money, efficiency is probably even more important and increases the chance of success disproportionate to the increase from having more people involved.
Having never started a startup, I wouldn't know, but how long does the average successful startup go without hiring more people? The number of founders pretty clearly only matters during that period. And, is that the hardest part of getting a startup off the ground? PG wrote that he generally hired people when none of the people already at the company could do the job themselves. Seems to me that if 3 people have incredibly much trouble communicating, you'd have a similar amount of trouble once there was too much work for two people and you had to hire another one.
What's different between the very earliest stages of a startup and the stage where you've hired one person?
Is it not the case that a major motivation for doing your own startup is to pursue your own idea? So the question that needs to be asked is this: what is the probability that k people will collectively come up with an idea for a startup so that each person feels they have contributed sufficiently to this idea? Maybe for groups with sizes greater than two, this is pretty unlikely.
I didn't think the link provided a good analysis of an optimal number, but it was nice to have a list of some big names next to their number of founders.
It's hard to say how many founders works the best. You just need to find what works for your situation. I'd say it's usually on the low side so you're not burdened with too many conflicting inputs at one time. When you have founders upwards of 5 or 6, it reminds me of high school clubs/organizations where everyone was a co-President, co-VP, or co-Secretary. Sometimes people just want to ride along and get a position, whether it's warranted or not.
Lets look at this from the practical point of view. Even if these numbers are valid, would they change the way you start or run your business? No! The reason is that statistics is about the average and there is nothing average about you or your business. You make your decisions based on the intimate knowledge of your own situation, goals, etc. I guess VCs have to use statistics (even bad one) because they get thousands of applications
IMO, this has to do with what a company is about: people and relations between them. That is, communication. With two persons it's easy to get to a 100% understanding. As the number of persons increases, the relations between them increases exponentially. So does complexity of those relations. Maybe two or three is ideal, but I tend to be a loner :D
This one is probably related to the well known property of human short term memory which states that we can remember about 7 plus or minus one things. This is the reason that phone numbers are seven digits for example.
Err, no and no. That was a hypothesis from the 50's that has since been invalidated, and many (most?) places in the world do not have 7-digit phone numbers. Memory is "chunky". For instance, you can easily remember phone numbers in your same area code even though they probably have 10 or 11 digits.
If you looked at a graph of the total dataset, what you'd see is that almost all startups with only one founder fail. So the reason the successful startups you list mostly have two founders may not be because two is the optimum number, but rather because two is the next smallest (and therefore most common) number after one. We can't tell without additional data.