Hacker News Reader

Comment by SyneRyder | Hacker News Reader

> 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It's a while since I played with OSM data for my own prototype, but I found most of the entries were 3 or 4 years out of date, and didn't include anything except a name and GPS co-ordinates.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.

View on HN

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.

> 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.

> 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) 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?

> 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.

> 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.

> I don't know what terribly mapped place that is...

I mentioned above, but it's Perth, Australia. Google seems to have won the mapping wars here. Google even includes live feeds of how busy a store currently is (measured in realtime by Android phones), and shows busy/quiet trends for each store so you can plan ahead when you want to go.

The Open Street Maps Australia website appears not to have been updated since 2009.

> Even as one person, I was able to create a dataset for my city that was 2.5x larger than OSM had

Then why in the world don't you contribute it to OSM so millions of users can benefit from it? (Or maybe you did, I don't know, but you seem to be pushing a separate platform here.)

I have better things to do with my time than to work for no pay :) If someone wanted to pay me commercial rates for my time & travel costs, I'd consider that.

I did contribute some to OSM, but it looks like only a handful of edits before I just gave up on OSM altogether. Most of the OSM edits in my city (Perth, Australia) seem to be from 4 - 5 years ago and are somewhat out of date. I shelved my app, so I have no competing platform to promote.

> I have better things to do with my time than to work for no pay :)

Are you serious? If everyone thought like that, the world would be a much grimmer place with no volunteering work, let alone open source software or open projects such as OSM.

I understand that it might seem futile to contribute to a huge project (mapping the physical world) when few other people near you are helping, but please tell me you aren't serious about having something better to do than volunteering work in general.

Very serious. I have other work to do and my own customers to serve, and they pay me for my work.

If you'd like to volunteer yourself to do the work on OSM, go for it. You can do some of it remotely by researching coffee brands in my city, going to their websites & finding contact details for each of the stores. Telephone each store owner and ask them to confirm details you're not sure of. Then contact the city council for lists of local cafe businesses (probably already on their tourism website) and do the same process with those cafes.

I'll give you a head start: my two favorite cafes in the Gloria Jean's Coffees chain are both missing from OSM, if you'd like to add those.

[Edit: Actually it's missing 19 of their 20 stores here, so there's 19 POIs you could add for that cafe chain in my city. At 15 minutes per POI, it's about 5 hours work.]

Reply on news.ycombinator.com