Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!ucbvax!decwrl!reid From: reid@decwrl.dec.com (Brian Reid) Newsgroups: news.config Subject: Re: print this file on a PostScript printer Message-ID: <101@jove.dec.com> Date: 25 May 89 18:03:57 GMT References: <96@jove.dec.com> Organization: DEC Western Research Lines: 23 In article karl@giza.cis.ohio-state.edu (Karl Kleinpaste) writes: >Random question of curiosity: How do you convince your software to >draw the lines where it does? That is, the basic arcs are no >surprise, that's straightforward, but in this incarnation of the map, >the transcontinental lazy `S' lll-winken=>uunet surprised me. Actual algorithms are used. I'll show you the code if you want. It's basically a brute force topological search of a 2.5-dimension space. >Small critical point: The `noisiness' of the map is higher than in the >first incarnation, when only 30-odd sites were listed. I think I >preferred the more restricted version, because I could get a better >feel for what was going where. There's simply too much data now. I'm still searching for the right level of detail. What I find is that when I put the top 30 sites on the map, the information is easier to see but the sites that have always thought of themselves as "backbone" status which aren't in the top 30 don't like it and ask for more. When I include more, it muddies the chart a bit. If you have a PostScript printer that can print 11x17 paper (e.g. the DEC LPS40) and if you have an Internet connection (mail from me to you does not have to travel over a dialup link) I can mail you a PostScript file that shows every single news flow in all of the continental US. It is awesome. There are about 30,000 flow lines on it.