Memefacture

I haven’t taken this apart yet, but Memefacture looks like it’s doing some interesting buzz_like_things.

My first comment is that it’s not painfully obvious how he’s doing automatic trend reporting… might be handy to document this and let folks critique.

11 Responses

  1. I’ve sort of considered doing this, but to be quite honest the ranking algorithm itself has been changing a bit recently. In addition what the actual components of the algorithm -mean- is somewhat difficult, although not impossible to describe.

    For example, what is the name of the unit consisting of a date / number of weblog entries ^ 2?

    And why even square it? Its kindve intellectually sticky, in a way it is what it is. Beyond that I feel somewhat protective of the technology… but that’s a different issue.

  2. Two more reason not to publish the algorithms you

  3. AND ANOTHER THING, I like (yet don’t understand how you quantify it) the distinction between sites which are “popular” and sites which are “influential”.

    In my word of understanding how information is flowing through these webs of people, I use the term credibility to define sites which are influential…

  4. WOO LORD I’M DRUNK. I WILL NEVER BET ON THE RAIDERS AGAIN.

    THAT IS A LIE.

    p.s. Seriously, I’m a veritable Deuce here, less alcohol???

  5. BZZT WARNING SIGNAL/NOISE RATIO DECREASING ALARMINGLY.

    It’s an summary of weblog data.. since I’m personally too lazy to check on this myself, are weblogs added to categories by hand or automatically somehow?

  6. “Popular” is a simple factor of age of the URL and number of weblogs that reference it.

    “Influential” indicates that a site was the first in the database to mention a URL (or URLs plural) which then became popular.

    The “Popular” links change hour to hour, the “Influental” links are pretty solid, although they could change at any time based on a site’s performance. It turns out some site are particularly good at this. I did not write the algorithms aiming at any site (ie Metafilter) it just happened that way.

  7. I think it would be interesting to apply the algoritms to different sets of weblogs. First, to individual weblogs and, second, to the meta-weblogs (ie: metafilter, daypop, etc).

    I don’t think the metalogs are influential, they’re simply relaying data already present in the links between weblogs. I realize the ability to relay this data could be considered “influential” since folks use these logs to find buzz_worthy data, but isn’t the truly influential weblog the one which first (or consistently) reports a tasty tidbit?

  8. Spacelegoman 14 years ago

    I’d argue the opposite. Most people read a particular sub-set of blogs which pertain to their interests, ie I’m in the music industry so I read music industry news sites, you code and like technology so you read coding/gadget blogs. It’s unlikely that our blog-reading habits will significantly overlap. Only a more general metalog might bring a particularly interesting meme to both of our attentions.

    Now, it’s true that within the category of say coding blogs, certain bloggers will have more credibility/influence in that specific community. But usually it takes a metalog (or an off-topic post) to spread a meme beyond its intended core audience. Which to me is the definition of influence – to bring someone’s attention to something they otherwise would not have considered.

  9. Completely agree — a fine point.

    Still, the influencial power a meta-log has relative to a individual-log is very different. If algorithms were developed to track influence/popularity, running those against “Rands in Repose” and “Daypop” and the comparing the results, I would, wouldn’t be that interesting.

  10. Spacelegoman 14 years ago

    This ties in with a book called The Tipping Point that was sort of fashionable last year. The book itself is ok, although somewhat short and underdeveloped. Still worth reading though, esp for the bibliography. Anyway, one thing the author posits is that in a typical network of influence there are three different roles: initial arbiters of what is useful, good, or cool, who he clumsily dubs “mavens” although tastemakers is the term I prefer; “connectors” who transmit this information to other influential people in the network; and “salespeople” who are good at spreading the word to the masses.

    In our discussion, I think a credible individual blog would have a role similar to a tastemaker, while metalogs are connectors… thus, they couldn’t really be analysed in the same way since their functions are different. Meanwhile, someone like Drudge will be all three – he’ll post a story, but then often follow it up over several news cycles with further links or additional information, essentially arguing for and “selling” its relevance.

    For argument’s sake could we say that weblog meme “influence” occurs when someone views your blog, and then changes their blog because of this? ie, I post an article about my cat, and then someone else links to it or argues a point I make or gets inspired and posts an article about their cat…. that is how we could parse apart influence and popularity. Influence in this context causes effects; it’s not passive.

    Anyway, it’s late so I’ll think about this tomorrow.

    Link to the book:

    http://www.amazon.com/exec/obidos/tg/detail/-/0316346624/qid=1045208815/sr=8-1/ref=sr_8_1/103-5005360-5753418?v=glance&s=books&n=507846