I'd like to point out Cinelerra even though I don't use it, yet.
In a similar vein though, I am a big user of Ardour, Jack and jammin to produce music. The tools are appropriate for the task and, whilst not perfect, didn't cost me money and allows me to be able to focus on my projects. Since I don't pay for the application my only investment is the time to learn it, the same reason people stick within a certain commercial platform. The difference is the Ardour project allows any financial contribution I make to be in preference of features I'd like added, improving the efficiency of my workflow. Becoming productive in complex software is the biggest factor in using it and the only incentive to change is when one type of software can do things the others can't.
I think the emphasis of these questions does not apply appropriately. It should be 'What is the current State of the Art in Video production in linux" and the answer is it hasn't caught up to the state of the art in audio production under linux.
Now before the criticisms begin, I find Ardour architecturally superior to commercial audio tools because of the underlying jackd infrastructure, not because of its feature set. I have watched the developments in the audio production space over the last decade produce change radically as they became more stable. Nothing interesting is happening in the commercial audio production space, it's all happening in Linux. As infrastructure advancements similar to jackd becomes more common in video editing the application space their will undergo a similar change - just not yet.
Any investment in time to produce an A/V product requires yielding value on a previous time investment in a skillset. When I invest that skillset in proprietary software my knowledge investment can be rendered useless overnight quite easily however, open source tools provide me with a way to protect my knowledge investment because the software has it's own intrinsic rights.
Value on knowledge investment is the value proposition of open source. You may have to put up with some bugs however, tolerating them means not incurring static initialization costs from learning over and over and that results in a permanent knowledge base, the basis for radically inventive ideas.