Yes, it’s unfortunate and unacceptable to figure out how the system behaves based only on the observations and experimenting with it. The key solutions like allowing to destroy anybody else’s changes to content units (lessons etc.) should be stated explicitly when the feature is still on the phase of requirement definition and design. Proposed design decisions should of been discussed publicly on the forum and only after reaching a consensus they should go to implementation phase.
There is enough smart and experienced people around here who would see and point out the possible design flaws and who who will speak up against implementing such a design.
I don’t think that authorization of changes by the original author is a realistic solution. After all we are here to study languages and not to be heavily involved in system/data maintenance. I can’t imagine the author that spends a lot of time comparing several proposed “improved” versions of his/her content. I think it will be enough if anybody is just allowed to create his own modified version with the name reflecting new ownership. Even if each modified version is “shared”, anybody else can pick it for the use “as is” or for further modification with the still another name.
You can find some interesting discussion (see the link below) related to the same topics that we have here. Please join us there!