4 ms·

> It also contains opening hours, addresses, etc., which is all in OSM. I don't understand why people always try to start anew when the data is right there. It

> It also contains opening hours, addresses, etc., which is all in OSM. I don't understand why people always try to start anew when the data is right there. It's a while since I played with OSM data for my own prototype, but I found most of the entries were 3 or 4 years out of date, and didn't include anything except a name and GPS co-ordinates. Some didn't even include a name. There was only one entry in my city with opening hours, and that entry didn't differentiate between days of the week. Some entries contained typos, and there was no consistency with naming of chains (eg one person calls it Dome, another Dome Coffee, another The Dome, and another Dome Coffee Fremantle). In short, OSM was a really bad dataset (at that time) and I found it better to start from scratch.

lucb1e 10y ago
Quality varies a lot, mostly from country to country (at least in western Europe, perhaps on other continents it's more regional). Whatever the quality though, it's a better start than nothing. Why would user data contributed to your application be any better than the user data contributed through OSM? It's a lot easier to edit an invalid property than have a completely blank map and start from scratch. Also, if projects like these would use OSM (some do, but not all), the quality gets better and it improves for everyone instead of for your specific project.
- SyneRyder 10y ago
  > Why would user data contributed to your application be any better than the user data contributed through OSM? I populated my data manually. I visited stores personally, talked to staff & store owners, measured GPS co-ordinates & WiFi speeds & counted power outlets in person. "Do Things That Don't Scale." Even as one person, I was able to create a dataset for my city that was 2.5x larger than OSM had, with better accuracy and more comprehensive attribute data. I got the idea after reading about the Google Maps Ground Truth team - turned out their secret sauce was an army of paid staff that could be sent out on demand to specific locations that needed to be verified in person.
  - JadeNB 10y ago
    > I got the idea after reading about the Google Maps Ground Truth team - turned out their secret sauce was an army of paid staff that could be sent out on demand to specific locations that needed to be verified in person. This seems to be the point, though. I don't think that your parent (https://news.ycombinator.com/item?id=13714467 https://news.ycombinator.com/item?id=13714467) is advising not doing any work yourself, but rather combining your work with existing OSM data rather than starting from scratch. Even if you took a simple algorithm, whereby you preferred any data you had for a location over any OSM data for the same location, surely you would wind up with a richer dataset? To put it another way, surely GMGT is useful only in combination with heavy use of existing data sets, not in place of them?
    - SyneRyder 10y ago
      > Even if you took a simple algorithm, whereby you preferred any data you had for a location over any OSM data for the same location, surely you would wind up with a richer dataset? That gives a messy, inaccurate dataset. It ends up including cafes that were closed years ago, places that aren't even cafes, or places several blocks from where they actually are. As a user, I don't think you want to walk out to a new cafe only to find that it doesn't exist. If you have to review the entire OSM dataset anyway, you may as well create your own. I did make a few OpenStreetMap contributions years ago, but I see they've since been 'updated' to incorrect information by someone who doesn't even live in my city. It still has a lot of inaccurate and missing information in my area. At least for what I wanted to do (and for my city), OSM was a really bad dataset.
      - lucb1e 10y ago
        > If you have to review the entire OSM dataset anyway, you may as well create your own. I don't know what terribly mapped place that is - probably one with lots of people where it made economical sense for Google to have a very high map quality, resulting in nobody using OSM - but I cannot imagine that that statement is true anywhere in the world. I've only used OSM extensively in the Netherlands, Germany and Finland and the data is very good. The occasional time I go into Belgium, data is much less good, but the roads are there and that's usually all I'm looking for. And where Google Maps is fine in the Netherlands and Germany, OSM was clearly better in Finland. > I see they've since been 'updated' to incorrect information by someone who doesn't even live in my city. Might be vandalism, though I have yet to see the first case in my five years of mapping. I've seen incorrect edits, but usually it's just clueless people doing something wrong, and most of the time it's caught by people checking up on new users. Is there a local OSM community that you can talk to, like for your country? Because invalid edits are not quite appreciated and I bet someone will jump on it when you mention it.

lucb1e 10y ago

Quality varies a lot, mostly from country to country (at least in western Europe, perhaps on other continents it's more regional). Whatever the quality though, it's a better start than nothing. Why would user data contributed to your application be any better than the user data contributed through OSM? It's a lot easier to edit an invalid property than have a completely blank map and start from scratch. Also, if projects like these would use OSM (some do, but not all), the quality gets better and it improves for everyone instead of for your specific project.

SyneRyder 10y ago
> Why would user data contributed to your application be any better than the user data contributed through OSM? I populated my data manually. I visited stores personally, talked to staff & store owners, measured GPS co-ordinates & WiFi speeds & counted power outlets in person. "Do Things That Don't Scale." Even as one person, I was able to create a dataset for my city that was 2.5x larger than OSM had, with better accuracy and more comprehensive attribute data. I got the idea after reading about the Google Maps Ground Truth team - turned out their secret sauce was an army of paid staff that could be sent out on demand to specific locations that needed to be verified in person.
- JadeNB 10y ago
  > I got the idea after reading about the Google Maps Ground Truth team - turned out their secret sauce was an army of paid staff that could be sent out on demand to specific locations that needed to be verified in person. This seems to be the point, though. I don't think that your parent (https://news.ycombinator.com/item?id=13714467 https://news.ycombinator.com/item?id=13714467) is advising not doing any work yourself, but rather combining your work with existing OSM data rather than starting from scratch. Even if you took a simple algorithm, whereby you preferred any data you had for a location over any OSM data for the same location, surely you would wind up with a richer dataset? To put it another way, surely GMGT is useful only in combination with heavy use of existing data sets, not in place of them?
  - SyneRyder 10y ago
    > Even if you took a simple algorithm, whereby you preferred any data you had for a location over any OSM data for the same location, surely you would wind up with a richer dataset? That gives a messy, inaccurate dataset. It ends up including cafes that were closed years ago, places that aren't even cafes, or places several blocks from where they actually are. As a user, I don't think you want to walk out to a new cafe only to find that it doesn't exist. If you have to review the entire OSM dataset anyway, you may as well create your own. I did make a few OpenStreetMap contributions years ago, but I see they've since been 'updated' to incorrect information by someone who doesn't even live in my city. It still has a lot of inaccurate and missing information in my area. At least for what I wanted to do (and for my city), OSM was a really bad dataset.
    - lucb1e 10y ago
      > If you have to review the entire OSM dataset anyway, you may as well create your own. I don't know what terribly mapped place that is - probably one with lots of people where it made economical sense for Google to have a very high map quality, resulting in nobody using OSM - but I cannot imagine that that statement is true anywhere in the world. I've only used OSM extensively in the Netherlands, Germany and Finland and the data is very good. The occasional time I go into Belgium, data is much less good, but the roads are there and that's usually all I'm looking for. And where Google Maps is fine in the Netherlands and Germany, OSM was clearly better in Finland. > I see they've since been 'updated' to incorrect information by someone who doesn't even live in my city. Might be vandalism, though I have yet to see the first case in my five years of mapping. I've seen incorrect edits, but usually it's just clueless people doing something wrong, and most of the time it's caught by people checking up on new users. Is there a local OSM community that you can talk to, like for your country? Because invalid edits are not quite appreciated and I bet someone will jump on it when you mention it.