Flickr Maps

What a conundrum. You develop a great piece of social software for sharing photos on the web, so good in fact that users spontaneously hack your tagging system to allow the easy addition of georeferencing information via the API and the visualization of these photos on a competitor’s map (via its great API). Other photo sharing sites are beginning to eye built-in georeferencing as a way to compete with you. To remain competitive, would you:

  1. Hire the most impressive of these spontaneous geohackers, and give this flourishing geotagging folksonomy an institutional backing? In-house mapping tools would make it extremely intuitive for anyone to georeference photos, but the results would still be compatible with the geotagging standard, so that the effort put into hundreds of thousands of photos is built upon, rather than supplanted.
  2. Hire the most impressive of these spontaneous geohackers, and have him build a brand new georeferencing system from scratch? This way, the technology is optimised from the get-go for georeferencing, rather than have to piggyback on tags. For example, it allows for separate privacy settings for photos and their geospatial data. Of course, the geotags on hundreds of thousands of photos on your site will be de facto deprecated, leaving those users with a geotagging “inventory” the choice of continuing with geotags or starting over with the new system…

I myself would have chosen the first option. Flickr clearly chose the latter one. Perhaps all the benefits of option 2 are not yet apparent, but what I hope will happen soon is the following:

  • Somebody offers a conversion utility that converts a user’s geotags to the new Yahoo! system. [Flickr has — see comments] Or vice versa, for that matter. (I’m surprised Flickr didn’t do this by default. Geotags are public anyway, and it would have given their new system a running start.)
  • Flickr KML Feed and Yuan CC maps are adapted to work with the new Flickr system. I want my Flickr photos in Google Earth, absent Yahoo! Earth.
  • Flickr’s georeferencing system gains features quickly to include some of the innovations being made with geotags, such as FlickrFly’s expanded geotag vocabulary.

flickrmaps.jpg

As an aside, it is interesting to see how Yahoo! and Google have differentiated themselves when it comes to georereferencing photos. Google’s solution is a tie-up between Picasa and Google Earth — both of them standalone applications. Yahoo!, in contrast, relies on two web-based services. Each solution has strong points: For Google, this allows for a far better editing feature set (Picasa) and visualization (Google Earth). For Yahoo!, the integration between Maps and Flickr is far tighter, allowing for such features as batch georeferencing and privacy levels for georeferencing.

PS: I’m traveling again, this time back to Stockholm, and then I’ll be apartment-hunting, so my access to the internet, and hence to this blog, may again be extremely spotty for a while.

8 thoughts on “Flickr Maps”

  1. Ah, the automatic importing of EXIF geodata sets the stage for extreme seamlessness. I like it a lot.

  2. First, thanks! :)

    While the geotagged format was/is good, it really had to be bought in-system to make it workable. I’ll blog more about this (yes I’m going to get blogging again now I can talk about what I’ve been doing) but we’re talking many many 64bit databases and specially written spatial search indexing libraries, just to make spatial searching do-able. The geotags just wouldn’t scale.

    In the last 16 months users have geotagged up around 450k photos, most of those using 3rd party tools, with choice of maps from different providers. These 3rd party tools don’t need to stop adding the geotags, all they need to do is add one more line to their code to also setLocation via the API. It’s upto them if they stop adding the geotags.

    In just the last 24 hours, we now have over 1 million geo located photos. There’s no way many of the third party sites could continue to suck down “geotagged” photo locations for searching locally and keep up with rate of growth.

    When the API is released (tomorrow probably) anyone can start to use it to add photos however they want, and search for photos to display however they want.

    As James mentioned above, you can turn on the option to import from EXIF. So you can use Picasa and Google Earth to “geotag” as much as you want, and then when you upload them to flickr it’ll pick ’em up. I even took a photo of me doing it :)

    To my mind the short is. Last week an easily searchable repository of over 1 million geo-located photos didn’t exist. This week it does. Never before have developers and mashup ppl had this many photos to search through based on location, and the numbers are growing all the time. To my mind, it’s a pretty big new thing we suddenly have.

  3. I’m having a little trouble with Flickr not understanding Exif GPS tags fully. It is the standard place to store them, and Flickr semi-understands the information, you can see it in the ‘more info’ part of a photo page, but for some reason they don’t translate that into a position on the map. Where as uploading to Panoramio they get it straight away and it appears in their KML feed without having to position it with an online map. So for now I’m favouring Panoramio for GPS photo sharing

  4. OK my bad, it does indeed acknowledge my Exif GPS tags, either I hadn’t turned the auto-exif reading thing or maybe it got fixed in the last few hours :-)

  5. Rev, glad to see your work get major public recognition. It’s a great start already, and I’m sure it will only improve with feedback.

    By the way, what percentage of the now greater than 1 million georeferenced photos are actually located based on the tags (e.g. ‘Puerto Rico’, or ‘Houston’) verses actually having users place them with the map?

  6. Frank, all the georeferenced photos are either from geotags, EXIF import, or the user dropping the photos onto the map.

    None of the photos were pulling in based on location tags. If you had a photo tagged Paris, that wouldn’t get pulled into the system.

    We track the zoom level that people drop the photos onto the map at. People may just go “Oh these photos are from Italy” and add them at a high level, other get added at street level.

    When using the API (and when the photos are displayed within Flickr) the accuracy can be specified. So an interesting metric would really be, how many of these are precision georeferenced and how many are general. The vast majority of them are very specifically georeferenced though.

Comments are closed.