These last days I have been testing, well, playing with issue crawler.  A tool amongst others @ govcom designed to help map and follow controversies on the net.

I expected to come up with some great results harvesting links from guifi.net and other related sites. However, to my surprise, the first crawler did not reveal itself as useful as I thought. guifi.net site did not even appear as a “node” (my fault for not understanding nor coming up with a great config for the crawler settings; As I already knew some of the actors that were engaged in free wifi issues I should have set iteration value to 1 instead of 2; that might have caused to discard some actors that had weak ties on-line but strong ones in the back office…). Anyway, after documenting myself a little bit on the site, things turned out to be a little bit more comprehensive and I could harvester a little more (still, a snowball analysis tentative was left somewhere between /dev/null and the never never land). For anyone who does not grasp what I am talking about, I deeply encourage them to visit issuecrawler.net. It just takes a couple of minutes to see what is about and it is worth the detour.

After playing with issue crawler, I felt that, perhaps, many researchers on the web are searching for graphical tools to visualize a vast array of data or just be able to set / label the links between significant clusters of actors. Touchgraph, for instance, using google “similar pages” may be a more suitable and fastest tool for such purpose. Although it does not enables you to label linking lines (which is always annoying).

With issue crawler you can export your crawler map as a .svg or .pdf or .png file and have that edited later, but still, it would be really great if you could have an app that will allow users to label linking lines or maybe it has and I missed it. However, Issue crawler, as a tool to help exploring or positioning some actors involved in a given issue (be it the BP unstoppable fuel drain in the gulf coast or the demand to open the EM spectrum) that has been traced over the net, is quite useful when complemented with other flows of data or when asking experts three or four websites to start harvesting.

In any case, I could not left unmentioned that guifi.net website is sort of anomalous site in that is not a “typical” site (fortunately). It has links to other sites, yes, but not in a way a blog might have linked his friends or neighbourghs. Nor in a way a big media site or a city council will do it…that left me thinking that there are sites more “suitably” designed to be harvested by issue crawler. And the less suitable sites included, the more experts  needed to help you identify both where an issue is located / based and where it is happening (sometimes you can not rely on the news; as the issue is not “noticiable”). Well…just thinking out loud. guifi.net site is both a site where people get info about but mostly “do things in it”

Another thing which surprised me is that crawls do change their shape when I consult them again. So, I guess the shape of a crawl is less informative than I first thought. Although clusters remain clusters. Again, I had not tested all the options, just played so I might and probably has missed a lot of things. I am not and I do not consider myself an expert using issue crwaler. But it is also useful to have these first impressions posted

Here’s a pict of my second tricky crawl;

from my harvested crawls

PS: I know pict resolution sucks, but I can not put anything larger than 495 px on my blog without deforming the post column and I can not resize the crawl .png without losing even more resolution either…Damn it!

RAX!