On Dec 30, 2012, at 9:01 PM, Russ Nelson <nel...@crynwr.com> wrote: > Serge Wroclawski writes: >> Steve suggested we need addresses. He didn't ask for a crazy huge >> import. > > Well, he kinda did. The TIGER data has addresses. The original import > didn't include them. We *could* triple the size of the data in the USA > by creating address ways alongside the TIGER ways. Eventually we will > triple it, if only by hand. > > The real question about this import (which is technically feasible), > is whether we can fix the errors with less effort than it will take to > input them by hand. Speaking as someone who has done importing and > hand editing, I think we should do the import ... of course only in > areas where there aren't already addresses.
There are some other angles to this. * One, is this: Can we expect to reasonably map the United States with addressing in a reasonable amount of time? Fred wrote an oblique blog post hoping that we will. That addressing is just like any other dataset that at first looked too hard (I don't know, footpaths or whatever). That is, it will get done quicker than you think. Fred lives in Germany and day-to-day finds a very different culture to the one I live in though. For example, mapping parties suck in the US because people drive and therefore don't go for a beer. We found this time and again. The culture has taken an age to mature here compared to Germany. The per-capita mapping is far lower than in Germany. Fred would probably say, Stallmanesque, that our ideals are more important than skipping ahead and doing an import. That even if it takes another 20 years to get it all done, we should wait since that way we would be doing it the 'right' way. For some definition of 'right'. The problem is that the anarchic-libertarian ideals that have worked so well at those German stamtisch are just not working here, or in a bunch of other places. Therefore I don't see why each country or state (i.e. Mass. and their own imports) can't have it's own solution which reflects the cultural realities there. As people have been pointing out here, it's kind of a false starting point /anyway/ since everything in the US practically started as an import anyway. So we can noodle with what "reasonable amount of time" means. To me, it means yesterday since this has already been going on far too long. * Another is, the threat of importing crappy TIGER ranges is motivating people to go look at available county data. That is fantastic. So as Ian alluded to pushing the conversation forward is itself a motivator * The most important though is to look at the realities of where we are. OSM is the third-largest crowd-sourced map behind Google and Waze. Fred will jump in and say it isn't a competition and to a degree that is true. However the world is changing. It may be soon (say 1-5 years) that there aren't just three crowd sourced maps of the world, but there are ten. Or twenty. Your computer-illiterate relative really doesn't care what map they use, they just care that it is up to date. It's arguable that they will therefore go use the map that came with their car or their phone and contribute to that. There is enough spare attention in the world to easily sustain 10 or 20 global crowd sourced maps. Skip back to Wikipedia. It took off by being both the place to contribute text (input) and also to read it (output). We've given up on the output part, that is the project thus far makes the website output (map style, user experience) useful to mappers and that's it. It's looking essentially impossible to break the core culture around how osm.org is built, maintained and operated but for the few highly-technical, competent and hard-working people who control it. Just look at the last major update, swapping one JS library for another JS library to show map tiles. It's technically elegant but it's completely irrelevant for, say, making the website easier to use, report a bug, get help or whatever. I digress. The point is that we don't have the feedback loop that wikipedia did. Our output is intermediated by the hundreds of sites and apps which use OSM. We gave up on the output side, Wikipedia did not. And, incidentally, neither has Waze or Google. The leap of faith you're asked to take is that wikipedia succeeded because it was free and open. I'm really not sure that's true. I wonder what would happen if we could rewind the clock and have Britannica or Encarta build similar websites back in 2002, or whatever. Would they have failed? I suspect not, actually. Just look at how Google and Waze have not failed. Tying this together, I posit that to be bigger than just "the worlds third-largest crowd-sourced map" we have to a) be quick b) import data where we have failings and c) fix the output stage and tie it back to the input to fix the imports as required. iD is a nice start, but potlatch has been nice enough for a long time. The issues are much plainer and in other areas, like the search box not finding my house and giving a list of 37 guesses for what I am searching for, or the inability to route. To my disgrace I posit missing leadership, but I no longer control anything. * Another angle. There is only one thing stopping large-scale adoption of OSM in the US and that is addressing. Right now everyone has to build their own crappy geocoding solutions. So people import TIGER or something else, then they use one of the various geocoders and cross their fingers. And they don't really work, compared to if they could just all use OSM and we all fixed that, hence my suggestion that we bite the bullet and pull in TIGER. The question comes up all the time "does OSM do addresses?" and the complicated answer we have today for the US is worse than a "no", and people just switch off. If we could hand wave a "yes" it would change a lot of things and create positive feedback in more people resources to fix the map. ... Anyway. Maybe I'm completely wrong. Steve _______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us