Good point.
> Not being able to inspect the code to see what it is doing may have ramifications for the user -- perhaps not so obvious on a device that has no network, but it is where it starts.
Sure. This connects to data and privacy concerns.
> If I don't give you a binary, then I never have to give you source code.
True of the GPL, but not of the Affero GPL. Also, permissive Free Software licences like Apache impose no such requirements, but are still Free Software licences. Free Software isn't the same thing as copyleft. (Of course, if you distribute binaries but refuse to share the source, then what you are doing is no longer Free Software.)
> It is purely a consumer rights issue.
This makes for easy contrast with the Open Source movement, which is good, but I think it's a little misleading. As an analogy, the right-to-repair movement doesn't expect hardware companies to hand over their design documents, they're just seeking to end open hostility toward third-party repairs.
I fully appreciate Stallman's 4 Freedoms, including Freedom 1, the right to study software, but I'm not sure I'd quite call it a consumer rights matter. Payware proprietary software licenses prevent you from freely copying/distributing the software. Patents and design patents can do a similar thing with hardware. That's not really a consumer rights problem.
To put that another way, I can see the sense in treating Freedom 0 (freedom to run the program as you wish) and Freedom 1 (freedom to study how the program works) as consumer rights issues. I'm not convinced though that Freedom 2 (freedom to redistribute copies) and Freedom 3 (freedom to distribute copies of your modified versions) are consumer rights issues.