The issue is that, basically 100% of the time, anything being released to the public has to go through a release process. If you produced it as an employee of one agency (eg the Air Force), but if was paid for by a different agency (eg DARPA), then it also needs to go through a similar release process on the sponsoring agency's side.
Typically, this has to happen for every single “release”. In other words, for the original software, again for every subsequent patch, for the documentation, for every correction or update to the documentation, etc. If you want to give a presentation about it your slides need to be released. If you want to write a paper on it that needs to be released as well.
Evey time through the release process takes two to six weeks, depending.
This works OK if you are releasing documents. Once a document is written, it usually doesn't need hundreds or thousands of updates over it's lifetime. A journal paper gets released once, then it's done. But the same process is what we currently use for software, and software is just a very different thing than a document is.
Imagine what would happen to the release cycle of a small OSS project if it took six weeks for github to get around to posting your update, and if github management rejected about half of them for reasons that were mostly inscrutible to you.
(Of course, if the updates aren't happening at all because of opaque internal procedures, that's a different story.)
I've got tons of code I've written over the years that, if I could just throw it onto github, I would. But since it would take me six weeks of pain, I don't. Some of that code I'm kind of proud of, or was at the time I wrote it, anyway. I wrote an rviz app a decade before ROS took off.
It could happen, but it would take something like the Secretary of Defense ordering it. Or a White House directive. The folks at my level, or at any level I can actually talk to, don't have the authority. The thing is, they aren't making these policies and procedures up just to be dense. There are real, hard and fast laws and department-wide policies that require the system to look the way it does.
(If there were more people filing request, I wonder if the internal procedure would be made less painful out of necessity...)
FOI is meant for information that can legally be published with reasonable effort, but not yet available.
The most effective method is to have continuous development through an open source structure so any failures where someone hard codes sensitive info in source get found quickly.
Please excuse what I suspect is a naïve question: why can't they just throw the code into a zip file, and send out the file the same way they would other documents? Digitally or on a physical thumb drive.
(The FOIA requester could then put the code on Github or similar.)