History Repeating Itself? Open-Source Challenging Google Map's Dominance

13 points by dhanrajm 4 months ago | 11 comments
  • n4r9 4 months ago
    Thanks for the article Dhanraj. I was wondering if you could expand on this part:

    > Transforming this data into a usable format, such as that required for routing applications, necessitates expensive data pipelines.

    What do you mean by "usable" here? I've worked with OSM data in .osm and .pbf formats for routing applications; there are a few things to look out for (e.g. badly formatted tags), and the ways/nodes graph models has a different structure to many proprietary formats, but generally speaking it's a one-off dev job to get some code up and running which directly imports data in those formats. Are you thinking more about importing the data into a postgres DB or something?

    • dhanrajm 4 months ago
      You are correct about ingesting the osm data directly to something like Valhalla. What had in my mind when I mentioned "usable" is the data after that will be the output of some sort of quality control process. As you said, osm data will have quality control issues (being a crowd sourced) and organisations will want quality checked before the data is fed to their internal systems including inserting to database or to a routing engine.
      • n4r9 4 months ago
        Cheers for the reply. By "quality control process" do you mean manually checking bridge height restrictions, recent road closures etc..., or do you mean something more automated such as filtering out badly formatted data?
        • dhanrajm 4 months ago
          its mix of both. I don't think full automation is possible at the moment.
    • karussell 4 months ago
      Sorry, but this is a strange article. Not sure what the reader should learn from it.

      > However, while OSM’s open, collaborative nature is a major strength, the quality and accuracy of its data still lag behind proprietary services like Google Maps.

      Such a general statement is wrong. For car routes maybe. But even for POIs I have found that OSM catches up fast.

      But Google Maps has definitely not the best quality for outdoor routes (like walking, hiking and biking). And this is not just my personal opinion.

      > These pipelines, which process and format the data appropriately, are typically only feasible for large companies.

      No, this is wrong. Already smaller companies can handle these pipelines.

      Furthermore there are many companies that offer OSM data as a SaaS (routing, geocoding and/or maps)

      > they are largely ineffective without access to the latest and most accurate map data.

      'largely ineffective' is misleading here. OSM is already sufficiently precise atm (ofc depending on your needs). Sure, you do not have live traffic, but there are already many (Uber-like) companies that use OSM and enhance them with their own data. And also there are a few options to buy live traffic for OSM data.

      > Therefore, currently, their best option is to utilize Google's APIs

      From my experience many customers find them too expensive especially for the ride-hailing use case OP describes.

      • 4 months ago
        • dhanrajm 4 months ago
          Geospatial tech is evolving fast with AI and cloud-native solutions. We’ve seen past disruptions reshaping the other fields. Are we at another turning point? Thoughts?
          • Chyzwar 4 months ago
            I am working on geospatial project, initially mapbox and maptiler competitor. My perspective is that there is combination factors that make geospatial tech to suck.

              - Legacy open source solution GDAL, mapzen and mapbox abandonware
  - Google semi monopoly on map data
  - Dead-end desktop apps like ArcGIS and QGIS
  - Plethora of formats/interfaces for storing and accessing map data
  - Plenty of uncoordinated efforts from government funded organizations
            My general thinking is that open-source is not suitable for geospatial tech because this is a highly specialized field with relatively small TAM. I think that there should be an antitrust case against google that should force them to contribute data to OSM.

            I am hoping that current landscape do not change in follwoing 1-2 years.

            • teruakohatu 4 months ago
              > - Google semi monopoly on map data

              I am sure it varies depending on where you are in the world but in my experience OSM plus official government data is as good as, and sometimes better, than Google.

              It would be nice to have mine Google Places and Review data but other than that I never feel I am using 2nd rate data when using OSM, in fact I am frequently amazed at the quality and depth of the data.

              That said, I do agree with your other points. A greenfield gdal alternative would be nice.

              • Chyzwar 4 months ago
                This depends on what you want to do. If you want to use OSM to find address, then it might be ok.

                If you want to create directions based on live traffic or find POI/business on map or have street view of location/route, then good luck. Google map have many datasets that just not available or too expensive for competition:

                  - Street view
  - User upload images of places
  - Pedestrian and car traffic
  - Public transport information and timetables
  - POI and business database
  - Consumer reviews
  - High quality aerial images