To recap briefly, I've posted in the past (I think those posts are lost) about Scratch's annoying take on open source software.
That is: in your NSF grant proposal state that you're going to do ongoing public source code releases; actually license the software under a permissive free software license; while not making it impossible to access the source code, disable the standard method of doing so; when asked about this respond as if confused as to why anyone would think you should be publicly releasing the source, despite the fact that the people of the USA are paying you on exactly those published terms, which you volunteered unprompted in your original proposal.
Still, it is kind of a technical point -- the software was released under a recognized free and open source license. They just made you find a work around on their forums to get to the actual source.
In the current release, they've made things much, much worse. They've given you instructions on accessing the source (good), but they've stopped using a free software license and switched to a non-commercial license (bad, bad, bad). From a licensing point of view, they've forked the project.
Briefly, there are a few problems with this. Nobody knows what a non-commercial license means in this case. Can I, a professional educational technology consultant, hand you a cd with Scratch on it? Can I train you to use it? Since it is un-free software it cannot be put in Debian, Ubuntu, Red Hat, or any other free software distribution. Can it be shipped on the XO? This license significantly restricts the distribution of Scratch to children around the world, and to what benefit?
Just to be clear, I don't have complicated or obscure desires about what Scratch (and other publicly funded educational software) should do: use a free software license and make the source code publicly available. That's it. I don't know whether MIT, the Media Lab, or the Scratch team is responsible for these problems, but the children and educators of the world deserve better treatment.