I came across a seed idea spread on the FeedBlog, wrote Kevin Burton, yesterday (using Talk Digger of course, you see the link to it in the blog post? It is the reason why linking is so important ). He pointed out an idea that Dave Winer gave for free 3 days ago on his blog.
The idea?
“Implement a search engine that accumulates all the stories pointed to by the top meme-engines over time. That way if I think of something I saw on Tailrank or Memeorandum a year ago, I just go to the universal meme search engine, type in the phrase, and get back the hits.”
Kevin was thinking about something a little bit different: a meta-memetracker that would look like Talk Digger.
I think that there is a place (at least emerging) for such a service considering the growing number of memetracker out there (TailRank, Memeorandum, Findatory, Megite, and probably others that I do not know of (I found yesterday a sort of memetracker on Rojo’s main page that is really cool)).
What would be the added value to users? The first thing is that you would have only one place to visit to get the top stories (obvious behavior for a meta-memetracker, no?).
However, I think that a more interesting phenomenon would happen too. The thing is that none of these memetrackers use the same methods/algorithms to find out what is a good story. Some seems to works with links and predefined list of good information sources selected by humans, other probably user some sort of advanced natural language processing algorithms, other a mix of these two methods and other probably use methods that I can’t think of.
All the memetrackers have one thing in common: they aggregate stories they think that are good (are they performing users profiling? It could be one next step to increase the effectiveness of these services Dave).
This said, some stories appear on all memetracker and other only on one of them. So, if one algorithm doesn’t score well for a specific story, it is not really a problem because the strength of the meta-memetracker is that it would prioritize the set of results composed by the intersections of the sets of results returned by each memetracker. That said, the meta-memetracker would return the bests of the bests stories because the error rate would be blended by the intersection of results’ sets.
It was my two pennies
(if you would like to read more about the socio-philosophical background of popularity, read that blog post wrote by Joshua Porter a couple of days ago)
Technorati: memetracker | davewiner | tailrank | memeorandum | findatory | megite | metamemetracker | popularity |
Kevin Burton
March 10, 2006 — 6:41 pm
Hey. Good post.
I agree there are a few challenges here. It might be interesting to see an implementation though.
Check out popurls.com.
I’m not too sure about these meta trackers though. They don’t have any hard work to do but I think they’re ok as long as they don’t steal too much traffic.
Right now they’re mostly a power user thing so its all good. I guess google had to worry about this LONG ago.
Kevin
Fred
March 11, 2006 — 10:12 am
Hi Kevin,
Thanks the kind words and thank for this comment.
Popurls.com have a beautiful design, it just display the web feed generated by each memetracker, non? It is a good start, but it would be even more interesting if they would do something with this aggregated data 🙂
I understand that you are not sure about these meta-memetrackers: everything is about the bandwidth, so money.
I questioned myself a great lot about that specific issue while developing Talk Digger and while checking it “growing”. I totally understand the reticence of search engine owners considering many factors: and I would have the same a their places.
However, I am driven by a vision of the future of the Web; the problem is that no apparent business model fit with that vision. You can read this post I wrote about that subject: The business model of a Semantic Web service.
Also, the thing is that meta-search-engines for example exists since ages (think about Copernic) and that Talk Digger is just a sample online version of this type of software. If I get problems in the future, I only have to create a software version of the service and all the problems would be resolved.
I also wrote a post on how the search engines works currently on the Web, and they is not that different from a meta-search engine: Search Engines are vampires that suck blood out of web pages
Finally, what a Meta-something do, is using the data generated by a web-service to create something new. It is one part of the vision of the Semantic Web. So, if we can’t do it right now, how could we in the future? It is a good thing to ask the question, see the answers and trying to resolve the problem.
Thanks to giving me the possibility to muse on the question 🙂
Take care,
Salutations,
Fred