Stratobe: Stratified Blog Engine

The software behind the main metatron.sh areas have the following design goals:

1. For people and projects: Metablog engine where tags and usernames are not separated. If i want to post something on my personal blog i tag it "maxigas" and if i want to post on a project page i tag it "projectname" and that's it. Tags are aggregated on tag.metatron.sh pages. (Implementation level: all this but it's only metatron.sh/tag not tag.metatron.sh.)
2. Privacy-aware: Tags have read and write authentication. I can make sure that only me can write "maxigas" tagged posts but i can also make sure that only my intimate friends can read my "personal" posts. Many bloggers are writing about their private life online because they want to share only with their friends, but it's not always easy to manage this from a technical point of view. On social networking sites you restrict your data flow to your friends only, but the invididuals who are my "criminal friends" don't necessary overlap with my "intimate friends": maybe i want to share my personal moments with a certain group of people but my hacks with another group. But how clumsy it is to tell your friends: "here's my blog, and here is your password for it". Since it's clumsy i'd prefer a blog engine that authenticates based on Jabber or where authentications can be bookmarked. The disadvantages are that on the one hand not everybody uses Jabber and on the other hand bookmarking doesn't seem very secure. (Implementation level: 'null)
3. Filesystem based: Administered is possible with or without a browser, online or offline. The whole data structure could be managed through SCP or GIT, so it's possible to work directly on the server through SSH, locally on the road after downloading the latest version. Web interfaces are user friendly but they also thrust back application usability 20 years into the past. Even with AJAX interfaces they are slow, clumsy and have poor keyboard support. They should be provided for beginner users and those who like them but for specialists there should be specialised tools. It's all a question of design goals: to be userfriendly or to be efficient. I am willing to learn if i can be efficient, but that shouldn't mean i exclude others, at least not from the managent of a blog engine. :j (Implementation level: it's already efficient but not userfriendly.)
4. Pet project: It's a Do-It-Yourself effort in autonomy -- producing your own tools for your own purposes. It's written in a programming language from the 70s MIT Artificial Intelligence Lab, therefore a true hacker tool. This dialect of LISP called Scheme does live coding, web apps, classic executables with GUIs, scripts and whatever you want, but it's not widely used in any of these areas. Therefore it's a challange and a learning experience to work with it. Let's see if it's Functional Programming methods are highly efficient or not. (More about this in the Scheme blog).