“Blogsnow ‘reads’ blogs, depending on what’s going on between 100 and 240 a minute. Out of this stream of information it extracts links and displays them in different views every 10 minutes.”
“In the top section you find the most current results. Those lists focus on emerging links. If a topic is widely discussed, but has been so for longer than the timeframe of the view, then it will be ignored. This keeps the view fresh. Come back in an hour and you will see.”
“In the second section you’ll find a couple of views for different target sites. I was wondering which movies are being discussed in blogs. I was curious which ebay listings generate some hype. Those views are attempts to answer these questions. Depending on what they display they may have additional features. eBay listings expire, so I keep a cached version on my server. imdb, which I use as a reference for movies, displays a neat image, so blogsnow shows you that as well. Different to the ‘now’ section here all links ever found are being counted. I will add recent topic based views once these lists become to stale to be interesting.”
There is also a section for the most popular weblog links to the New York Times, CNN, stories from Snopes, and Wikipedia.
ResourceShelf interview with Andreas Wacker, the creator of Blogsnow:
- RS: How does Blogsnow set itself apart from other blogging resources?
AW: I use Blogdex and Technorati frequently. But I looked more often than there was something new to see. Blogsnow updates every ten minutes, and it focuses in on topics that are emerging: If it the ratio between recently added and existing links balances then it will ignore those ‘old’ links. Old is a matter of a days or hours here. Blogs are quickly reacting to events, and blogsnow goes along with this. When Tenet resigned last week, it was nice to see that blogsnow was about an hour faster than technorati in picking the links up. Blogs are a realtime environment. Blogsnow is trying hard not to add too much delay into it. I am certainly reinventing the wheel somewhat with blogsnow. But to me those wheels resembled more the shape of stop signs. Which is not saying that I got rid of all corners on my wheel. There are still lots of those left.
RS: Blogsnow does what no other blog content aggregator does. It shows us the most popular links for a particular source/database? In non-technical terms, how does it work?
AW: Those views are based on where people link to. Lets take the imdb as an example which is an amazing resource. Blogsnow simply notes which blog recently added a link to any of its pages. The internet is brimful with great sites that provide allot of valuable information. Those topic based views try to combine this with the authorship of millions of bloggers. The results vary as much as the contents do.
RS: What is your criteria for picking resources to display results?
AW: A content based view like imdb or google searches must reflect what is going on. Those top 50 lists should express what is a topic that people care about. And its results must be robust. Future trends should be equally well displayed. Since blogsnow is so new, the verdict on this still open. It might be that some views may disappear in the future, if they stop returning interesting results. Topic views are based on what I am curious about. blogsnow contains a set of raw data. When there is something in it on a given subject/target site, then I try to generate a topic view for it. Sometimes it works, sometimes it doesn’t. The beauty is, that I play with this quickly. The data is already there, I just need to filter it right to get there.
RS: I understand that you ping the most popular weblog change databases, such as weblogs.com and blo.gs. Do you only grab information from the blog posts or from the blogrolls (favorites links) as well?
AW: Blogsnow only follows ping lists. This has also the advantage that blogsnow sees only blogs, that people like to make public. Allowing people to ping blogsnow directly is a topic that is high on my to do list.
RS: How much work do you have to put into creating each list? How much is manual labor?
AW: It really depends on the list. Some are easy, others take a bit more work. A couple of hours I would guess. To write the code is very manual labor, but then it just runs. Or, at least it should.
I just realized that my first inclination is to read “blogsnow” as “Blog Snow,” which is kind of cool, but confusing.