On Jun 17, 2010, at 7:55 PM, Ian Dees wrote: > Do we really have the mapping capacity (in person-hours and server hardware) > to handle 300M people a day providing feedback?
that's kind of the point lets say you were a very large firm and you wanted to scale OSM to that level, how would you do it? Because I can't figure out a way that doesn't mean 'taking over' the servers, or forking. But the sad thing is it applies for much lower numbers too - if you wanted to scale to 1 million / day editors let's say. > > I mean we could point people to openstreetbugs, but it's already full up > because people are more interested in mapping the areas they are interested > in, not necessarily fixing bugs in other areas. > > Also, after several tries at editors by our community, I think the problem of > editing map data online in a community fashion is simply a hard thing to > learn. It's a big leap for someone to go from hitting the "Report a problem" > button to even marking a POI let alone putting their first way down. If the > number of problem reporters is 300M and the number of mappers fixing the > problem is 300K, we'll quickly burn our mappers out. > > On the other hand, maybe that's a good problem to have :) > > On Thu, Jun 17, 2010 at 8:30 PM, SteveC <st...@asklater.com> wrote: > Well let me take that back a bit - actually even doing some very simple > cleanup of the interface and having a feedback mechanism *at all* would be a > good first step, as people jumped on my recent OGD post in the comments: > > http://opengeodata.org/the-importance-of-timing-to-feedback > > > > On Jun 17, 2010, at 7:27 PM, SteveC wrote: > > I think you're concentrating on tiles, but that's not really the bottleneck > > I would jump on first. > > > > The conversation goes like this: > > > > "steve we have 300 million people a day look at our site and we would like > > to send their edits and feedback to OSM" > > > > Really it's the API we're talking about. Tiles are just a CDN problem. > > > > > > On Jun 17, 2010, at 7:18 PM, Frederik Ramm wrote: > >> Steve, > >> > >>> They would like to link to us directly but don't think a) we can > >>> handle the load and b) don't think it would be a good user experience > >>> to dump people on to osm.org, what with the site design. > >> > >> To paraphrase (not specifically Wolfram, but the unnamed other megacorps > >> you're chatting with): > >> > >> 1. they'd like to link to us directly but our infrastrucutre is too weak; > >> > >> 2. they would not want to give us a shitload of money to improve our > >> infrastructure, but could imagine hosting something; > >> > >> 3. there is fear that the community would view this negatively. > >> > >> To which I say, I don't think the community has anything against someone > >> doing a glorified maps.cloudmade.com; if they have really fast servers and > >> maybe even a CDN, can do lots of styles and make the tiles and services > >> available under a free-for-all policy. That would be great, and would - if > >> given sufficient long-term promise by whoever it is - allow us to reduce > >> our tile serving to an experimental capacity, freeing up resources for the > >> core database which obviously we must keep operating ourselves. > >> > >> But there is a logical problem here and that has nothing to do with us at > >> all. You say that many would like to link to OSM directly if only OSM had > >> sufficient resources. Now assume that some big guy with many enemies, say > >> Google, or Microsoft, were to offer super-fat tile serving for OSM as I > >> outlined above. We would then scale back our own tile ops to a minimum, > >> and their server would be the main OSM tile server, and whenever you go to > >> www.osm.org your browser says "connecting to osmtile.google.com" or some > >> such. > >> > >> I think that the community would be less of a problem - I don't think many > >> would care if our tiles came from MS or Google or so as long as they were > >> unrestricted and the data remained free. But all those other big guys, of > >> whom you say that they would like to link to us - would *they* want to > >> send their users to get tiles from Google, MS or someone else? Or would > >> the "we'd like to link to you but your infrastructure cannot take the load > >> and anyway your front page is ugly" then be replaced with "we'd like to > >> link to you but you must understand that the 'sponsored by XYZ' on the > >> shiny front page is a problem"? > >> > >> Of course things would be even worse if the big sponsor wanted to put the > >> tiles or service under a non-open license (e.g. a license with a > >> "noncommercial" component"). That, I think, would reduce overall > >> usefulness rather than improving it. Any funded tile serving would have to > >> be more open than what we can currently offer, not less. > >> > >> Bye > >> Frederik > >> > >> -- > >> Frederik Ramm ## eMail frede...@remote.org ## N49°00'09" E008°23'33" > >> > > > > Steve > > > > stevecoast.com > > > > Steve > > stevecoast.com > > > _______________________________________________ > talk mailing list > talk@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk > Steve stevecoast.com _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
