I very much agree with the idea that software gains power by its pervasiveness, such that users no longer have the freedom to choose their own software. As Henri says, it may be chosen by their employers, or even just by the force of the marketplace. As economists have pointed out, there's a natural tendency to monopoly in software, because once it reaches critical mass, you tend to need the same software that others are using. But note that this is only true if the software uses proprietary data formats! If the software is "open" as to its data formats and interfaces, the user is not locked in.
To me, this very argument points out that the problem is abuse of monopoly power, not the intrinsic nature of software distributed without source code. Software distributed with source code frees me from a certain kind of lock in, but this is less of an imperative if the vendor isn't trying for lock in in the first place.
I'm mindful of the wonderful old Usenet .sig quote from Doug Gwyn, which went something like this: "If UNIX didn't let you do stupid things, it wouldn't let you do smart things either." One of the problems I have with the GPL is that it tries to correct for abuses by proprietary vendors ("stupid things") by making them impossible. But this cuts off various business models that, if done properly ("smart things"), provide huge benefits to users and other developers. Not all software (or other work products, such as books) are done for the love of it, or to "scratch the creator's own itch." A lot are done to put food on the table and a roof over someone's head. And so we have to make room for models that provide a richer spectrum of incentives for the work to be done. I believe in marketplaces -- a willing seller and a willing buyer, with a range of currencies (from dollars to the esteem of peers) as the exchange -- and trust that over time, any abuses can be corrected by education and advocacy.
I'd rather leave people the freedom to do stupid things, while encouraging them to do smart things.
These issues are really worth thinking about very deeply, because the ground is changing under our feet. If we don't separate out the real goals (deep freedom of choice) from more superficial freedoms (e.g. the ability to change the source code), we might find ourselves in a world, where, for instance, source code is always redistributable but vendors have found other forms of lock-in (patents, for instance, or just monopoly market power that allows them to keep changing a data format faster than competitors can keep up.) The fundamental freedom is the freedom of choice, not the freedom to have source code. Having source code is a step towards that freedom, but not the first freedom in and of itself.