The first notable example of the network effect came in the early days of the telephone. Some places legislated a telephone monopoly (only one company was allowed to string wires) while others allowed competing companies to operate. Almost without exception, phone service was more widely adopted (and cheaper!) in the places with only one phone company. One problem with having two or more is that you might not be able to call the person you wanted to talk to because they got service from another company, or would have to pay an exorbitant toll charge to make that call. (Nowadays we solve that problem with government mandates that require phone companies to play nice with each other.) The other problem was that putting up all that infrastructure was costly, and having competing companies meant that each one was dividing the costs among fewer subscribers. It’s the same reason we have regulated monopolies for electric service; in many places you can now choose an alternative generation provider, but getting the volts to your house is still a monopoly.

Getting back to the more direct application of the network effect, being served by the same telephone company as your friends made that company’s service more appealing. So people ended up being marketing tools for the provider they chose. In some cases the effect was strong enough that one of the companies died off. In other cases, each one managed to rope in a group of happy subscribers, but with the side effect that those subscribers were kept away from the customers of the other company. (You might be seeing some parallels with the iOS vs Android divide here.)

In the case of operating systems, the network effect is especially strong because it changes the decisions of software developers. A phone platform that has a 5% or 10% market share isn’t going to attract a lot of app developers. In turn, that means people won’t want to buy it. That leads to more developers abandoning it, and the downward spiral continues. Desktop Linux has the same problem; not enough market share to attract developers. (It also has the additional problem that the people who DO choose to use desktop Linux generally don’t want to pay for applications; software that is both free and libre is part of the appeal.)

Apple does OK in laptops because it has specific market niches where it has far more than 10% of the market, notably music makers and graphics professionals. Application availability for those niches is excellent, often exceeding what you can get for Windows. Apple doesn’t do nearly well in the mainstream business market because they don’t target it, and in fact do things that antagonize business customers like killing compatibility with older applications. Businesses often depend on some application that they can’t update for a newer OS because they don’t have the source code.

But there was no identifiable niche where Windows Phone was strong enough to allow it to survive on that basis. It needed more app support to have any appeal to mainstream customers and Microsoft wasn’t getting it, even though they offered incentives to developers to write apps for the platform. Microsoft also shot itself in the foot by going through so many generations of mobile operating systems and not maintaining any sort of backward compatibility; that history made it harder for them to attract developers and customers. Perhaps the most egregious one was the transition from Windows Phone 7 to Windows Phone 8; all the apps were incompatible and no option to upgrade your older device from 7 to 8 was offered. (Not that many Windows Phone 7 devices were sold. If Microsoft had been serious about making 8 a success they would have GIVEN a new phone to every owner of a device running 7; it would have been a modest marketing expense by their standards.

Nickelback? Network effects exist in music as well; there are musical fads and rabid fandoms. But they are not as strong as in operating systems. If you like Nickelback you may lose out on the pleasure of talking about them with your friends, but in the end the music is still the music and the choices of others don’t have any effect on what you hear. (Though sometimes people seem to care anyway. When the story came out that Milli Vanilli was a fake, many people demanded refunds for their records and CDs even though the music on them had not changed at all. I had never been a big fan, but if I had been the story would not have had much effect on my love the music. If it had come out that they were something worse like child abusers I would have been more torn.)

GE is a lot like Microsoft, just older. Having worked for a company that set out to emulate GE as closely as possible, in a different industry, the approach can work for a time. But once some key figures leave, it can fall apart quickly. And what I found working for that company was that there was quite a bit of corruption inside to goose numbers. A project would get cancelled to create the illusion of savings one quarter. Then, a few quarters later, the project gets respun because it’s necessary. The stop-and-start creates additional overhead and waste. I’m not saying that went on inside GE because I don’t know, but it sure happened in this other company that tried to emulate them. And GE’s stack-rank system is an outstanding way to bleed talent. I had a year that was so good it was mathematically impossible for me to repeat it. But rather than get a promotion and the kind of bonus a high performer gets, I got a 3% bonus, no raise, and was told I was average. Needless to say I got better opportunities very quickly, so I left and I’m doing the same thing for someone else now. That was an annual ritual–there was always a flood of talent who left right after bonus time. That was why.

Any further thoughts regarding GE?