"1)...
2) Inertia is a powerful thing in business. No matter how good product B is, if the company has been using product A for the last few years, they're not going to change unless they have absolutely no choice.
3)..."
You forgot:
4) IT departments that have the responsibility for keeping the organization's system's up and running and take the blame when they're not, are loath to bring any new possible points of failure into the mix without either; A - exhaustive testing (for which they often don't have the time or other resources) or; B - The name and address on file of the ultimately-responsible source of the potential vulnerability that they can pass on to the legal beagles if things go flooey.
And, yes -- part of that is mired in the inertia noted above: "Better the devil you kniow...!", because, with Adobe, et.al., they know the hoops to jump through, etc.
But they ALSO know who to blame and -- if necessary -- who to sue if things go decidedly pear-shaped. With OSS like, say, Red Hat Linux, companies know that the service contract that they're paying for means that, if things SHOULD blow higher than up, there's probably going to be someone at the other end of the phone line when they call, and that there's someone that they can throw under the bus when they're called on by the higher-ups: "They've been doing this for N-years, they should have caught this bug LONG before now, we did our due diligence..."etc., etc. With all-volunteer, decentralized software, who do you point at, who do you go after to recoup the expenses of cleaning up the mess? There's no one and it's probably coming out of your already-stretched resources.
Trust me; as one of those that so many of the El Reg commentariat like to refer to as the "colored pencil crowd", I've run into this with trying to get, e.g., FOSS drafting software for a property manager in our department (because BESIDES being the graphic designer, I'm ALSO the department alpha-geek) who doesn't need the considerable licensing expense and learning curves of full-fat AutoCAD or ArcGIS to take an existing map and say "Put a driveway here. Put a hedge there." But IT won't install software if they can't absolutely nail down the provenance and chain of responsibility for it and, with a VERY limited number of IT grunts supporting the administrative systems of a city of 100,000+, they don't have the spare resources to vet and support a bunch of randomly mismatched software installations all across the system. So there WILL be limited customization allowed, and those WILL be from a VERY limited number of VERY well-established companies.
It used to be said that "Nobody ever got fired for buying IBM," because companies trusted that they knew who to get on the blower to get problems fixed in a timely manner. Today, that can probably be extended to "Nobody ever got fired for licensing [Microsoft | Adobe | Oracle | RedHat | ...]."
With all-volunteer FOSS...? Not so much. But don't get me wrong; I LIKE and use FOSS where possible at work and at home and I wish nothing but the best for the Inkscape developers. But if they're looking for big uptake from big commercial/government establishments of their all-volunteer-all-the-time software, they probably shouldn't hold their breaths.