Building a better blogtrap.
The 18th century French Gastronome
Jean
Anthelme Brillat-Savarin wrote: Tell me what you eat, and I will tell you
what you are. There is a great deal of truth in that,
but I'd like to embrace, adapt and extend this wisdom. Tell me who you read and I'll tell you who I am.
For the last three months or so, I've been putting a mostly daily offering
called Take Outs. This really doesn't serve a great purpose of aggregation like
the Boing Boing blog or the Scoblizer nor am I anybody that has Rock Star
status. I am just a humble novice Tenzo attempting to express the three virtues
through the six tastes.
Something that I think I've learned from that practice is that while
the Take Out model works, it puts too much emphasis on self and not enough and
not enough on what I really want it for. Here are my six main problems.
- Finding blogs I am interested in is still too hard. Tools like
Feedster and Technorati help, but they are owner subscribed. Unlike the
root-and-node structure of Usenet, there is no root feed to subscribe to.
- Commenting sucks, but communication requires a feedback channel.
One of the biggest thing I've gotten some static for is that Take Out
presents a number of posts along with the comments on the SQLJunkies site.
That turns an otherwise passive channel (from the viewpoint of the original
poster) into an active one: she or he has to care enough to follow my
trackbacks, then cull through the list to find their post just to read my
comments. I'm not sure that's a reasonable thing to ask if I'm wanting the
poster to engage with me on the topic. Of course, if I did it the other way,
I'd have to some engine that kept track of where that post was so I could go
view (or fetch) the stream of comments some how. That's another problem:
comments aren't feeds.
- Who are you and how expert are you? I've noticed that my list of
of 1,500 blogs-to-read has really boiled down maybe 100 that I consistently
find content that I'm interested in and whose authors I've taken the time to
learn something about. How did that reduction work? First, the author had
enough information on their blogsite that I was able to make a mental
connection with them. Second, they've said enough stuff and/or have been
pointed to by others that I trust to know that I feel I can have a high
degree of confidence in what they have to say.
- Blogroll Yin, Blogroll Yang: I've built most of my blogroll from
the blogrolls of the people that I read just as much as I have the people
who admit to reading me. Why? Part of it goes back to if the idea that if I
feel you have something to say, you read the blogs of people who have
something to say. There is different sameness in why I read the blogs of the
people who read me. This Yin and Yang is problem because it limits my reach
since its a static-build process.
- Judging posts by their headlines. Sorry, but most nights there
are about 2,200 posts waiting for me to slog through. I don't have time or
mind to really read them all, so like the chicken, I peck at those that look
interesting and ignore the rest. The only mechanism I have right now to peck
at is the headline and the poster. Anything by Michael Rys, Dana Epp or Dare
Obasanjo gets immediately read. Put the words XML or Yukon in and you'll
catch my eye. I wonder how many really good posts I miss per night just
because the headline didn't fire for me?
- Developing a shared Ontology seems a fools mission. When I say
Yukon, 99.99% of the time, I mean SQL Server 2005. When a lot of other
people in the States say Yukon, they mean the SUV. I pity the SUV driving
Western Canadian DBAs. But without a shared ontology, sharing information is
very complicated. However, getting to an agreed Ontology for everything
makes Grand Unified Theory seem simple.
So, do I have a solution in mind? After all, I've been very critical of those
who complain but offer no ideas. I do, and its been brewing for a while. Time to
blow some of the gas off of it. Here's my wish list for a written-medium,
peer-to-peer knowledge sharing system:
- Such a system must function with some centralized root. Distribution to
other nodes is fine, but point is to have one go-to or one gone-to source.
This makes the first problem a bit easier. It doesn't solve the
self-subscribe problem, but it does give people additional motivation to
register.
- Bloggers would register feeds that ascribe to a specific theme. For
example, a Central root might be set up for Yukon, and any posts feed to the
that root must be about Yukon AKA SQL Server 2005. That root should
channelize content within its major aspects (SQL, XML, CLR, etc.) and each
aspect should be its feed.
- It should be as easy as possible for a post on my feed that I've said
fits in that channel to feed into that channel.
- Comments on posts are posted back to the central root and feed back out.
- Membership to the central node is required. You must post enough
information about yourself to establish your relative level of expertise to
the rest of the subscriber community. That's why NNTP failed -- it became to
easy to hide behind an anonymous shield.
- Membership is by invitation of an other member only. Yes, I know that
sounds like I'm advocating an elitist system. I'm really not, but thank the
Spammers for making this a requirement.
- First-source posters earn karma based on the number of other posts that
refer to their posts, some rating assigned by the referring poster as
modified by the karma eared by the referring poster. Higher karma posts earn
higher reading priority. In sense, its a juried Oracle system.
- All members of the community are expected to maintain their own blogs.
The central root provides a dynamic OPML that any aggregation client can
fetch on startup. That way, I'm more or less guaranteed to see stuff I'm
interested in.
- In addition to karma scoring, I want my aggregator to score all 2,200
posts based on their content relative to content that I've already said I
was interested in some statistical manner.
- The Ontology is defined, owned and managed by the central root.
Last night, sharing these ideas with Scoble, it seemed to me that Channel 9
would make a great field lab for seeing if such a system had legs.
But for now, I must go wash the rice from the sand.