The OpenDesktop.org website/community has many flaws that can only be fixed by fundamental revision.
OpenDesktop.org (and its specialized spin offs) combination of social networking and free/libre content publication has become an invaluable element of the FOSS ecosystem (especially for KDE) however the site also has a number of flaws that can not be fixed without creating something that is fundamentally different. A successor site could learn from the following mistakes
Encouraging use of forums as documentationEdit
This is a growing problem in the community. There is an increasing number of key software projects that do not provide adequate documentation and instead rely on user discussion to fix every ones' problems. OpenDesktop.org's Knowledge base feature encourages publishers of software to use this flawed model. A site made to replace OpenDesktop should instead encourage developers and users to collaborate on proper manuals to minimize.
- In the opinion of this writer the ideal method would be to give every published piece of software a namespace in a wiki similar to wikibooks.org
Make the successor more child friendlyEdit
It is frequently observed in the FOSS community that the next logical aim for us is to get FOSS operating systems in to schools. However no desktop environment that integrates the Open Collaboration Service of OpenDesktop.org could be accepted into schools as there is a number of wallpapers, login themes, etc that are clearly meant to be sexually arousing. A succeeding site would need to have some form of moderation over such material.
Categorise (applicable) content by stabilityEdit
This is mostly a major issue for KDE users but it probably affects others. There is no system within OpenDesktop.org that draws a distinction between experimental code and software that is ready for day-to-day use. This becomes a particular issue when content is downloaded via KDE's "Get Hot New Stuff" framework. Some things work perfectly and others are very incomplete, non-functional and in can even cause desktop subsystems to crash (particularly plasmoids). A simple solution to this problem is to categorise software by its level of stability.
- It will be very difficult to make a fair system of categorising in this manner. Content should be labelled as "unknown" by default. The higher levels of stability should be marked as having been tested in multiple environments. For example has it been verified to work on OS's other than Linux? Does it need a particular version of a library? Does it make assumptions about the bit-order use by the CPU?