The "middle way" is to have enough control to maintain compatibility but with licensing that's open enough to allow third-party innovation. This is what Microsoft achieved with Windows, and what Google is trying to achieve with its rapid release schedule for Android - to oversimplify what Google's Chris DiBona told me, if Google releases new versions at a rapid rate, the downstream suppliers can't diverge too much or they won't be able to keep up.
However, looking at the past 30 years of computer history, it may be that top-to-bottom openness is a recipe for failure.
In the 1980s, for example, I really did think Unix was going to take over the world. At the time, we had six distinct classes of machine -- home computers, PCs, workstations, minicomputers, mainframes and supercomputers -- and Unix was the only operating system that ran on all of them. The IT world was in chaos and users needed another standard besides IBM's. But nobody wanted to support 57 different operating systems, and everyone recognised that there was a huge amount of value in having compatible applications. Unix must win!
Unix failed for a number of reasons, including its notorious lack of user-friendliness and generally high prices (up to $1,000 a copy). But the major problem was that Unix failed to deliver the compatibility that users expected. Dozens of different companies produced their own versions with different names, different interfaces, and different marketing strategies. When Sun and AT&T tried to standardise the industry on a single version, System V Release 4, rivals such as DEC, HP and IBM had a hissy fit and created a rival standard, OSF/1.
I’ve recently watched Linux suffer from the same mistakes in the netbook market. When the first netbook, Asus' original Eee PC, was first unveiled in the UK in 2007, it only ran Linux, and Asus was soon followed by its main Taiwanese rivals, Acer and MSI. I said they should agree a common standard, such as Ubuntu or perhaps Kubuntu, and club together to fund and support compatible drivers. But, sadly, they saw Linux as a way of differentiating themselves and, like Apple, not being dependent on Microsoft. United, they might have stood, but divided, they fell. When Microsoft provided a slightly modified version of XP and a cut-price deal for netbooks (ULCPCs), sales took off, and Linux’s share of the market collapsed from a 100 percent to less than five percent.
When the Linux hype first grew loud in 1999, and Microsoft posted its Linux Myths attack page, many people predicted world domination. Some Linux supporters reckoned Microsoft would be dead within 10 years. It hasn't happened, and on NetMarketShare's figures, Linux's PC market share is still less than one percent even now.
Worse, the current trend is away from openness, with Microsoft and other companies trying to replicate Apple's huge successes with its tightly controlled proprietary offerings, the iPod, iPhone and iPad. Indeed, the latest signs from Apple's preview of OS X 10.7, aka Lion, are that Apple wants to exert similar control over its PC buyers by providing an App Store and making it work more like the iPad.
The message from the market is that buyers aren't as interested in freedom, openness and social benefits as they are in personal benefits. They want something that works for them, and the less effort it involves, the better. Is that sad, or not?