Conversation

Notices

1. Embed this notice
   HD (hareldan@mapstodon.space)'s status on Friday, 03-Jan-2025 23:50:58 JST HD

   So I went looking at some stuff at Overture, and realized they also have
   POIs, from Facebook.

   All of a sudden, FourSquare's errors pale in comparison.

   • Embed this notice
     Nemo_bis 🌈 (nemobis@mamot.fr)'s status on Friday, 03-Jan-2025 23:50:53 JST Nemo_bis 🌈

     in reply to

     • Rihards Olups
     • Ilya Zverev

     @richlv Not there are limits to this. For example, marginally *worse* TIGER imports in #OpenStreetMap in 2007 ended up causing *higher* quality in the affected USA localities, because they attracted more contributors.

     "Information Seeding and Knowledge Production in Online Communities: Evidence from OpenStreetMap"
     doi:10.1287/mnsc.2020.3764

     @zverik @hareldan

   • Embed this notice
     Rihards Olups (richlv@mastodon.social)'s status on Friday, 03-Jan-2025 23:50:54 JST Rihards Olups

     in reply to

     • Ilya Zverev

     @zverik @hareldan
     That reminds me about the arguments against bad mass imports in OSM - "OSM is not a data dumping ground".

     It does make the dataset seem impressive at first, but then ends up being bad, unmaintainable or both.

   • Embed this notice
     Ilya Zverev (zverik@en.osm.town)'s status on Friday, 03-Jan-2025 23:50:55 JST Ilya Zverev

     in reply to

     @hareldan 1.5 years in, I see Overture as a geodata graveyard, not a practical data source. Everybody still uses OSM.

   • Embed this notice
     HD (hareldan@mapstodon.space)'s status on Friday, 03-Jan-2025 23:50:56 JST HD

     in reply to

     It's from all those stupid Facebook check-ins people do. Here's 'Heathrow Airport', at JFK airport, because someone, over 6 years ago, didn't know how to work with Facebook.

     Why do we even need the resources to host these, let alone maintain and display. Who benefits from LITERAL GARBAGE in these datasets?

     https://www.facebook.com/pages/London%20Heathrow%20Airport/254807908554621/

   • Embed this notice
     Nemo_bis 🌈 (nemobis@mamot.fr)'s status on Saturday, 04-Jan-2025 00:07:28 JST Nemo_bis 🌈

     in reply to

     • Rihards Olups

     @richlv If "importing known mistakes is beneficial and should be allowed" fits your definition of "not a data dumping ground", yes.

   • Embed this notice
     Rihards Olups (richlv@mastodon.social)'s status on Saturday, 04-Jan-2025 00:07:29 JST Rihards Olups

     in reply to

     • Nemo_bis 🌈
     • Ilya Zverev

     @nemobis @zverik @hareldan
     That seems to support "OSM is not a data dumping ground" position, right?

   • Embed this notice
     Rihards Olups (richlv@mastodon.social)'s status on Saturday, 04-Jan-2025 00:17:16 JST Rihards Olups

     in reply to

     • Nemo_bis 🌈

     @nemobis But "worse" in that context was _less_ data, not "more crappy data". That is an extremely important difference.

   • Embed this notice
     Nemo_bis 🌈 (nemobis@mamot.fr)'s status on Saturday, 04-Jan-2025 00:17:16 JST Nemo_bis 🌈

     in reply to

     • Rihards Olups

     @richlv It is indeed!

     The main problem with the Overture data is probably not so much how incorrect it is, but just how hopelessly redundant and difficult to deduplicate/normalise it is. There might be a way to extract some small subset of it to seed OSM in chronically undermapped areas, but I'm not aware of anyone having figured out how to do that.