Information Home

A friend passed on a link to a whitepaper by Tommaso Toffoli about the concept of an information home. I really like the way he turns the computer into part of a "cultural package", rather than just focusing on technology in isolation.

I've felt this kind of itch -- for an extended brain -- before. It was one of my motivations for making diamond wiki. But by thinking of this as a personal tool, rather than a tool for solving individual problems, it makes the concept feel much more stable and useful. It ties together things like Google's desktop search, buddy lists, mp3 playlists, bookmarks, blogs, etc. All of those can basically just be seen as skins for presenting the underlying information, and verbs for acting upon it. They each have their own way of storing the information, organizing it, searching it, and presenting it.

It would be fun to try to tie all this stuff together more. In a sense, it's trying to meet some of the same goals as the semantic web, but on a personal level instead of worldwide. The idea of trying to solve this for the entire world seems pretty hopeless to me at this point in our technological development.

I can see several pieces forming. Yes, it's true I'm diving into implementation way too fast, which is precisely the thing the Information Home concept is supposed to avoid. But hey, I'm an engineer, and that's what I do.

• You want a way of representing knowledge, like the contents of a buddy list. For now I'll just assume that subject-verb-object statements, plus narrative text, plus meta-statements, will be a good match for this. Use something like sniki, but open up the database so it's accessible to other tools.
• You need somewhere to store that knowledge. It needs to be globally accessible.
• You want standards for how knowledge should be represented. This allows sharing. Might also need to have standards for translating from one form to another, if a single standard turns out to be unworkable. Consider what would be required to share the entire Froogle database, but don't get stuck on the difficult technical problems involved. Consider integrating with mail stored on an IMAP server as simply being a kind of sharing. Solving this will mostly be an evolutionary matter of trying things and seeing what sticks.
• You want some generic visualization tools, for presenting things as lists, graphs, hyperlinks, etc. You might use a web browser as the front-end for this, but someone has to convert the tangled relationships in the knowledgebase into some kind of sensible presentation. That requires a piece that isn't common yet (at least I'm not familiar with it). Think of Tufte, or naked objects. Visualization tools should be able to piggy back off of the structure-transformation capabilities involved in sharing incompatible structures. Reminds me of the recent thread on the ll1 mailing list, about having analogs to the unix text-processing tools, but for visual data instead (unfortunately no archives available).
• You want tools that can take actions based on the knowledgebase. Sometimes it's not enough to just have the data itself and some way of visualizing it (e.g. looking at your schedule for the next month). Sometimes you need actions to be taken (e.g. notify you when an item on your calendar is about to happen), or some specialized domain-specific logic (e.g. schedule a meeting time that's compatible with five people's schedules). As much as possible, tools should work by inserting back into the knowledgebase -- that way one tool can watch your schedule and insert a notification event, but it doesn't have to actually notify you, the human, since that ability is also used by phone calls, IM messages, etc.

I know that lots of people have been thinking about this kind of stuff before, and that I'm not really saying very much new here. But I'm excited by it.