Two nights and 900 photos later, all but a few are now mapped. (As with all things Flickr, they made it easy enough to become a compulsion.) I have a few notes and a couple of suggestions.
First off, though this was perhaps an obvious move to many, I’m still thrilled that it’s here. Flickr’s implementation is nice and intuitive, and works mostly the way I’d have hoped. Mapped out photo locations is nice extra bit of metadata that will no doubt change how I view my photo archives.
Integrating Yahoo Maps was an obvious move, given the corporate overlord just happened to have a ready-to-go mapping tool on hand. But less-obviously, it’s a move that will get Flickr’s users more familiar with Yahoo’s map offering, which is a nice win for the company. The problem is, compared to Google Maps, I just don’t feel the love yet. Google is faster and feels more responsive, its keyboard shortcuts work better, and its data feels more complete. The data issue is a relative one, as Yahoo is more broadly comprehensive with at least major highways showing up for most countries, but it falls down on the fine details. (No street maps of London, for heaven’s sake.) Over time this is sure to change, but in stark contrast with Google Maps, Yahoo maps is a tad frustrating. (To be fair, they acknowlege it themselves.)
Then there’s the privacy issue. Sure, before allowing you to geotag your own photos, Flickr gives you a nice little warning that you’re about to do something that could have repercussions. But that only helps you avoid do something silly like give away your street address by posting a cluster of photos from inside your apartment. What about other people using your name as a tag on geotagged photos? There are some huge implications here that are going to make a lot of people uncomfortable. The immediate saving grace is that this is in no way realtime, but we’re bound to see some interesting discussion about this as people catch on.
The only thing that’s really bugging me though, is the way photos have been paged. A widget lives up in the top left corner that allows you to navigate through various “pages” of location-mapped photos. There are a few things going on here that might be a little subtle, and I’m going to make a few assumptions, so bear with me.
Let’s take it for granted that there are simply too many results to show on the map at any given time. This may be more true for some places than others, but in general, with 2+ million geotagged photos so far, we’ll assume it’s an issue. Representing that many photos on a map of the earth just isn’t going to work, you have to break it down into chunks.
A Flickr photostream runs in reverse chronological order, with newest photos at the top. A Flickr map’s pages work in a similar order, but the experience is not the same.
The solution they’ve hit on is to break down the photos chronologically into separate “pages”, almost exactly the same way your average photostream is broken down. But the disconnect comes with the fact that dots on a map don’t indicate chronology the same way photos in a linear stream do. It took me a lot of second-guessing to figure out quite what was going on here, and even now that I know, I still feel a sense of disorientation when I know there are many more photos in a location than are currently being shown. The load time in between when I click on the pager control and when the map actually updates isn’t helping matters either.
Recognizing the need to break photos down into more digestible chunks, I can see two possible ways of coming at this problem. The first is the way they’ve already done it. The second takes into account other inherent metadata that’s not immediately obvious, the cluster phenomenon. Popular places will contain clusters of photos, whereas more pedestrian places will have only a scattering of individual pictures here and there.
Instead of filtering by time the photos were taken, I suspect a more useful filter would be a popularity threshold where the default behaviour is to only show locations with large amounts of photos (large relative to the total number of photos within that area), but allow the user to manually change that filter. At a glance, I’d be able to tell which spots are hottest, and if I wish I can filter out those and look for the less-travelled paths. And the data that would drive this already exists, given that these number show up on the map as-is, they’re just not the main driving force behind the existing filtration method.
Here’s a pair of sequences, this first one showing a few pages of the chronological breakdown as it exists at the moment.
This other one is a mockup for how a cluster breakdown might look. The latter is just my own thinking, but I sure hope to see something similar in the future.