Wikipedia:Featured picture candidates/Internet Map

Internet Map
I think this is a cool picture and is ascetically pleasing. Used on internet and computer. Created by User:Matt_Britt Support I like it, although I feel a better caption to explain the photograph is in order. TomStar81 (Talk) 05:42, 3 December 2006 (UTC) Support image:Internet_map_4096.png, while it provides very little technical information it does give a view of the internet not otherwise possible. HighInBC (Need help? Ask me) 21:21, 3 December 2006 (UTC) . I'll promote the smaller version. There's a link to the big version in the image description page, and we don't want the big version crashing people's computers when they open it from the main page... :) Raven4x4x 05:34, 8 December 2006 (UTC)
 * Big BIG version. (thumbnail size: 9 MiB... this is why we should not promote this version)
 * Nominate and support. - Ravedave (help name my baby) 04:38, 30 November 2006 (UTC) *Comment a larger version is available Image:Internet map 4096.png.--Andrew c 05:12, 30 November 2006 (UTC)
 * To anyone just joining this discussion, please read below to find out why using this big version is a bad idea. -- mattb
 * Comment - I'm going to remain neutral on this one, but I feel obliged to point out that the rendering isn't perfect... There are a few glitches that I'd like to get around to fixing sometime; mostly manifesting themselves in the form of routes that are seemingly disconnected from the rest of the network. -- mattb
 * This seems like a serious issue, since I'm guessing the data was gathered by crawling across links that should connect all the sub-networks? Debivort 20:24, 30 November 2006 (UTC)
 * Theoretically. It is possible for there to be some isolated nodes because of the way the data was generated, but it's very unlikely. The isolated nodes that exist on this image are mostly just a rendering glitch, not so much a problem with the data. -- mattb
 * Comment I'm going to go all Swiss on this one too, although I like the hi-res version a lot. It would be more interesting if some of the paths or nodes were labeled to give a sense of what is being illustrated. --Bridgecross 14:34, 30 November 2006 (UTC)
 * Consider that this map has over 40k nodes on it, and then you might understand why I didn't label them. Labelling the nodes would produce an unreadable mess. It might be neat to identify a few major backbones and label them, but that's difficult and has limited accuracy due to the way BGP works and the way the data was collected. -- mattb
 * one way this could be addressed would be to show a zoomed-in inset in the black peripheral space that shows labeled nodes. Something like the pair of images seen at except with the bottom set labeled. Debivort 20:30, 30 November 2006 (UTC)
 * I'm going to oppose it, I'm afraid, even if no-one else wants to! It's pretty, but I've no idea what I'm meant to learn from it. Of course, I'm happy to change my vote later if someone can convince me it contains some information. Stephen Turner (Talk) 15:38, 30 November 2006 (UTC)
 * You're not meant to learn much from it. It's just a clever way of visualizing how the routes on the internet interconnect. I could make a scaled down 'educational version' with a lot of nodes labeled, but it would look far less interesting. -- mattb
 * Matt, I'm afraid you aren't defending your image sufficiently. One should be able to learn a lot from all FPs, and yours could be one of them! How the routes connect is of profound importance in understanding the topology of the internet and how it operates. It is popular to know that actors belong to a small-world "6 degress from Kevin Bacon" network, but many fewer people know the internet has a very similar power-law structure. I think we could easily address Stephen Turner's concerns with a nice caption. Debivort 20:24, 30 November 2006 (UTC)
 * If I was trying to defend my picture, I wouldn't have pointed out its major flaw; something you may or may not have noticed on your own. Again, I hope to redo this with the flaw (blackened routes) fixed. Perhaps that would be a better time for a FPC discussion. As for Stephen Turner's objection, my ears are wide open to any suggestion you may have as far as captioning or labeling goes. What sort of label would you consider sufficiently informative yet not obtrusive to the aesthetic of the visualization? -- mattb
 * As above: one way this could be addressed would be to show a zoomed-in inset in the black peripheral space that shows labeled nodes. Something like the pair of images seen at except with the bottom set labeled. Debivort 20:30, 30 November 2006 (UTC)
 * Well, that's doable. I'll keep it in mind for the future. -- mattb
 * I uploaded a new version of the big map (not thumbnailed here because MediaWiki won't resize it). The invisible routes are fixed, and there's a zoombox type of thing to show the detail in a small portion of the map. If you like this sort of thing, I'll make a smaller version. Let me know. -- mattb
 *  weak Strong Support big map weak oppose - I would eagerly support a version in which the disconnected sub-networks were attached. Debivort 20:24, 30 November 2006 (UTC) I'll support a version in which the zoomed in region is farther from the center so that the "zoom swath" is less obtrusive, and would suggest 1) it be made slightly more transparent and 2) the assymetric glowing dropshadow on the blowup part be removed, and 3) a zoom in region be chosen that has a bit more node diversity, like a more highly connected hub in addition to low-degree nodes. Forgive my pickiness, but it's very close to perfect. PS - I had thought it was illustrating the web, but with the IPs, I'm guessing it's the physical internet? Debivort 02:29, 1 December 2006 (UTC)
 * I'll tweak things some tomorrow and try to address your concerns. I'd like to leave in the drop shadow in some form because otherwise there is no definition between the zoomed patch and the rest of the image (black background on both). I'll try to pick a region that's a little more dense, but as you zoom in a lot of the definition is removed since I'm not rendering edges in the zoomed area that aren't connected to any nodes visible in that area. This makes it much more obvious that the paths ARE actually connected to nodes, and not just random lines criss-crossing everywhere. As to your question, this shows some of the routes between various IPs in class C space. What that actually means in physical terms is a little fluid; especially for some of the larger routers and back bones. A route detected via the data collection method may or may not be totally representative of the physical connections between routers (many routers have multiple interfaces and therefore multiple IPs). The reasonably sized data set should overcome some of these difficulties, but in a roundabout way I merely mean to say that this is one way of visualizing routes on the internet. There really is no perfect way to show all the possible routes a packet can take since these change constantly (even, for example, from packet to packet within one stateful connection like TCP).
 * I'll also ask you to consider letting me make a scaled down version of the big image. While it's superior from a resolution point of view, it's so large that MediaWiki will not generate thumbnails, so it's sadly not usable in articles (browsers would have to download the full size ~10 MiB image). -- mattb
 * Okay, look at it again... I've uploaded yet another (big) version. -- mattb
 * This addressed all my concerns - it's great! I just hope the scaled-dwon version will retain the legibility of the enlarged IPs - maybe the target of the zoom in region will need to be shrunk a little bit? Debivort 06:29, 1 December 2006 (UTC)
 * Support - it looks a lot like an artificial brain, something long predicted. Great colors, too. Rklawton 20:26, 30 November 2006 (UTC)
 * Support Weak Oppose per debivort, this could change if the disconnected lines were fixed. I'd also like to see 1) some legend on the image. 2) further explanation (what the 'white puffballs' might be, why a line's classification might be unknown...). 3) Labeling some major points might or might not be a good idea too. I think this has a ton of potential, but isn't there yet. Most things either fixed or rescinded. There is enough potential content in this image to make an article. -Spyforthemoon 22:30, 30 November 2006 (UTC)
 * I don't like legends on images. It makes it necessary to translate the image text in order to use it on another WP and it can just as easily be done through a caption. The puffballs may mean different things; I would have to be very careful about assigning them any particular significance. As for "unknowns"; those are either reserved private network IPs that are somehow being used for internet routing (yes, this actually happens) or IPs that ARIN didn't have much (or any) information on. -- mattb
 * Support per quality of picture. Sharkface217 23:40, 30 November 2006 (UTC)
 * Support per above. Is it possible to make a svg version since it's generated from point to point connection and nothing but a huge bunch of lines anyways? -- antilived T 10:55, 1 December 2006 (UTC)
 * SVG is conceivable, but I'd have to write the software to render it myself. I'll keep that in mind for the future. -- mattb
 * Comment - Allright, I've uploaded a new version of the scaled down copy. Fortunately MediaWiki will thumbnail this one. I've tried to preserve the zoom effect while keeping it reasonably unobtrusive. -- mattb
 * Oppose I don't see what is nice about this image. What determines the length of a given line segment? What do the different colors mean? Which "part" of the internet is depicted? Is this a map of where the data is actually flowing or where the data _could_ flow potentially? Ultimately the issue is that it isn't clear what is to be learned from this image, and also it looks like a big fireworks display.Spebudmak 19:13, 1 December 2006 (UTC) Oops, I didn't see the summary on the actual image page in which these issues are addressed. Sorry. I'm changing my vote to Neutral. Spebudmak 19:18, 1 December 2006 (UTC)
 * Weak support. Although it would take forever to render, svg is really the ideal format for this. And it seems that other than the nice aesthetic effects, it would be just as easy to make an svg out of this as drawing the lines render-time. Still a great image though. Good enough for FPC, though not front-page worthy IMO --⁪froth T C  21:43, 3 December 2006 (UTC)
 * I'm rather baffled by statements like this one. What use is featuring an image if it can't be used in any articles (or really, on any Wikipedia page)? At 9 MiB, the 4096x4096 image is too large to be thumbnailed, and an SVG version would most likely crash the renderer MediaWiki uses. I appreciate wanting the best quality possible, but you have to make some sort of compromise when dealing with a massive data set (with 40k nodes and 71k edges, this image represents less than a third of the data available and less than one percent of all the class-C networks). I feel I should reiterate that it is a bad idea to promote the 4096x4096 version to featured status. It will not be thumbnailed, so whatever pages it is used on will require the browser to download all ~9 MiB of the full image. Please consider the 1280x1280 resolution image, since it can be resized for other pages. -- mattb
 * [[Image:Symbol support vote.svg|15px]] Support Big version. That's pretty cool, any idea what the large spiky white balls are? 1995?!! imagine what it must be like now! --Fir0002 00:31, 4 December 2006 (UTC)
 * No, early 2005. Ravedave mis-captioned the image. Plus that isn't the whole internet; it's somewhere between a fifth to a half a percent of all the class C networks. -- mattb
 * Support big version- Does a nice job illustrating the basic architecture of the internet. --Lewk_of_S e rthic contrib talk 01:49, 4 December 2006 (UTC)
 * Support the big version - Good job at illustrating its subject, looks like featured quality. Hello32020 02:46, 4 December 2006 (UTC)
 * Support Small version - Oppose' Large version - even the thumbnail takes forever to load. PPGMD 05:40, 4 December 2006 (UTC)
 * Oppose both. This image is either too simple (only half a percent?) or too complicated (all this to illustrate the principle?) to explain the internet. --Janke | Talk 14:49, 5 December 2006 (UTC)
 * Oh come on, at least pick one objection and go with it. Which is it? Too simple or too complicated? If I rendered, say, one percent of the class C networks at these resolutions you'd see nothing but an indecipherable mass of color. How is that useful or neat to look at? I've said from the outset that this is nothing but a unique rendering of how routes interconnect and therefore how data can make it from node A to nodes BCDEFGH. Vote how you will, but don't expect this image to be more than it was intended to be. You cannot glean how the internet works in any detail from a simple image. The intention was merely to provide a rather abstract and interesting visualization of some internet routes, not as a replacement for an explanation that would take a fairly hefty textbook to elaborate. -- mattb
 * I stand by both objections: the image is too complicated just to explain the principle, and too simple to give an idea of the immensity of the web. Here's a third: If it is just an "abstract and interesting visualization", as you say, then I oppose because of the low enc. Sorry, but that's how I see it. --Janke | Talk 23:38, 6 December 2006 (UTC)
 * Well, I can respect your opinion. -- mattb
 * Support smaller version - Aesthetically pleasing, Good job of illustrating the internet. Advanced 19:26, 5 December 2006 (UTC)
 * Comment I created a GIF version from the original at Image:Internet_map_4096.GIF. The file size is 2.87 MB and the only difference I can see is the zoom box at the bottom. Could this be fixed? Mahahahaneapneap 19:16, 6 December 2006 (UTC)
 * MediaWiki won't thumbnail that either; it seems to be a resolution cap, not a file size cap. So the gif version poses the same usability issue as the large png version. -- mattb
 * Support - Fascinating picture, well done in my opinion. -- Chrislk02 (Chris Kreider) 13:17, 7 December 2006 (UTC)
 * Comment - PNGs larger than 12.5MPx won't thumbnail, bug 3771. ed g2s &bull; talk 23:14, 7 December 2006 (UTC)