com-mod-i-ty: noun: a mass produced unspecialized product
During the cocktail hour at last week’s Where 2012 conference I had a conversation that I hear a lot these days: Maps are just a commodity now. Google’s move to charge for maps has sparked an well publicized exodus to mapping alternatives; many, but not all based on OpenStreetMap. The statement always bothers me…granted there are many open source alternatives out there now, but has it really gotten that easy, that reproducible, that “unspecialized” that it can be bought and sold with the same lack of differentiation as wheat futures and pork bellies?
If true, some pretty big guys have wasted a ton of money in the space. Google Maps, the gold standard by most accounts, has been at it since 2004, employed many hundreds of very bright people, spent billions of dollars to build what they have. They have introduced new features every few months during that time. Microsoft, Nokia and others have committed resources on a similar scale. Is that now truly commoditized by a dozen engineers, a few months work and the availability of open source code?
It’s a problem of definition. To many people, “maps” are the visible manifestation that you see; the map tiles that get served in response to a query. Clearly, there are a lot of sources of map tiles out there and open source options make it pretty easy to get in that business. But many of us use “maps” as a short-hand way to describe a wide range of services that come bundled under that term including things like geo-coding, routing, local search. I argue that many of those things are not commoditized. In fact, they are pretty freaking hard to do well. About a dozen slides into his excellent presentation on whether to switch from Google Maps, Sebastian Delmont of StreetEasy defines many of these elements of the Web Map Stack. Here’s my thoughts on a few relative to their commodity status:
- Map Data: Still a killer to do well. There are only a handfull of sources. I am a big fan of open source options like OSM, but it won’t work for everyone everywhere. That’s a blog in itself (Side note: Check the comments following Foursquare’s announcement of their switch…I love the back and forth between Brazilians whose towns just disappeared from the map and OSM’ers who try to get them to join The Cause. Brazilian’s last word: “I do have a day job so I don’t have time to help improving maps.”). NOT a commodity.
- MapStyle: Cartography is an art form but it yields to smart folks and effort. I’m liking the new styles and differentiation that’s coming out. The carto should serve a purpose. For example, I like that Foursquare’s map fades the street grid to emphasize the places, which is their bread and butter. I’d agree that this has been commoditized by the ubiquity of the Google map look, but think there’s room for a lot more creativity here. Check out Stamen Designs if you want some ideas.
- Geocoding: This has been the proverbial Vietnamese Quagmire for many companies. Seems easy at first, but to do a full on, single line, high accuracy geocoder is really really hard. And then you have to do it again for each new country…they each have their own structures. To quote a knowledgeable Geo-rati “All geocoders suck. You just need to figure out which suck less”. Open source options exist but it’s a long hard slog to turn them into something really good. NOT a commodity.
- Routing: Again open source routing code is out there. It seems like the first
80% of building a routing engine isn’t too hard. But after that, you need to build the weird corner cases…the jersey jughandles, the Swindon roundabout, etc. The problem with routing is that you can do it right 99 times, but the customer will remember til he dies that time you got it wrong. NOT a commodity.
- Local Search: Local search, meaning dropping points on a map isn’t too tough for the simple version. That’s usually a circular radius around a point. Not to big a deal. I think there are more compelling search models out there that take advantage of more advanced geocoding that we’ll see coming up.
And there’s more…aerial imagery, real time data, traffic, etc., but this is long enough. My conclusion is that Maps (meaning the full stack) are not commoditized. In fact, they’re pretty tough to do well.
Why does it matter?
It matters because if you think maps = map tiles and switch to someone who can give you that, you may quickly learn that the rest of that stack didn’t come with it. And that the stuff you didn’t get is actually pretty hard to do. If you don’t need it, fine. But I have seen several companies (and some of them pretty big ones) make the leap and then find it tougher slogging than they’d thought.