Firefox, arguably the most successful consumer open-source software to date, was a fork.

Standards only reach perfection the closer they get to the actual product. Without a product you don't have standards but missing some minor standards wont effect a product too much.

Forking because of minor standards issues is bad for business because you're following weaker standards that don't have too much to do with a single product that people actually use. People don't surf the Free stands Group website all day when they want to wwrite a spreadsheet although it's good initially if you need portability. It seems sometimes people are more into standards for a hobby like a collector.
Basically it's mostly better for an individual or more private company to develop and set the industry standard (It's always been that way in America and people who invent everywhere) while relying on weaker standards or community to help them comply with basic market needs. (It's always been that way to until recently with such small hidden microcircuits and software, Ugh) Fore instance, even Open Office has their prime standard of StarWriter and OO formats while retaining the weaker ODF for broadscale deployment. It's a weaker standard because it might not 'save' as well as saving it in StarWriter for instance. I wonder if there's a site to check for the 'losssiness' 1with that

Too much forking shows that the product has an identity crisis. But some forking can turn out to be great with low overhead but still it has to have muscle to be taken seriously or right away it's a headache.

This reminds me of an art project in 9th or 10th grade (in Spain.) The teacher told us to draw "freedom," so I drew a great big blue ball in the middle of a piece of white paper. The teacher told me that couldn't be freedom because freedom was about lots of opportunities and open doors ... I told her freedom was about being able to do whatever I wanted without anything getting in my way. (I was a teenager after all! :)

It seems to me that the GPL freedom is more about making sure nothing gets in its way than about leaving lots of doors open.

X.org forking from Xfree86 is another example where the branch won.

X springs to mind as a successful fork. And most distributions could be considered forks; why distributions do it so much more than discrete software projects I couldn't say.

But the general points are well taken. People do tend to work to avoid the need to fork--and it seems likely that the awareness that people *could* fork if they really wanted to keeps a lid on dysfunctional behaviour. Power corrupts. With Free software, those at the core of a project are there solely because of hard work and the respect of others; the right to fork ensures that they don't actually have any power over anyone and hence are not corrupted. Or perhaps corrupt people don't try to put themselves at the centre of free software projects because there's no power for them there. Or a bit of both. Whatever the case, the "right to fork" lurking in the background probably does a power of good even if it's never used.

"You can fork, but you can't stop others from incorporating your fork back into the main tree."

Not true, consider a project with "GPLv2 or later" license. Such as most gnu projects. Lets say they fork into GPLv2 only and GPLv3 branches. The changes in one branch CAN NOT be merged into the other as it would make them undistributable.