Tuesday, May 29, 2007

Google versus Microsoft

My previous post "Google gets it right" generated a bit of discussion on the relative merits of Microsoft and Google in the mapping space. I just wanted to say that I am relatively neutral on that score at the moment - I think that each has strengths relative to the other, and we are in the early stages of what will be a major battle between the two of them (and perhaps others, such as Yahoo), over the next few years. Google built up some good momentum ahead of Microsoft, first with Google Earth and then with Google Maps, and I think that their relatively recent announcement that they are indexing all KML files on the web gives them another advantage, because of their leadership in search. However, as I wrote about in GeoWorld a little while ago, Microsoft has a strong track record of being a "fast follower" and gaining market leadership after initially falling behind - as it did to Apple with Windows, and to Netscape with Internet Explorer.

The most obvious area where Microsoft has a significant technical advantage over Google is in its 3D building models and the techniques and technology it acquired from Vexcel to enable it to automatically build these from accurate photogrammetric data. I don't see how Google's current approach of allowing the community to upload models will come anywhere near the accuracy or completeness of what Microsoft is doing there. It requires a major investment to match what Microsoft is doing in this area (Microsoft say they are spending hundreds of millions of dollars on this). A related area where Microsoft has an advantage also is in the "Birds Eye" oblique imagery data that it offers, which Google doesn't. I think that Microsoft seems a little more focused on pushing Virtual Earth as an application development platform in the business world, though admittedly that's subjective - and both of them really need some work on their licensing arrangements for commercial use. As discussed in the comments on my earlier post, in some areas Microsoft has better imagery than Google (such as the village of Cropston in the UK, where I am at the moment), but the reverse is true in other cases - it's hard to draw any clear conclusion there without a lot more analysis. Other related technologies of interest that Microsoft has include SQL Server Spatial (coming next year), and Photosynth, which is one of the coolest new technologies I have seen for a while, which has a lot of potential for being integrated with Virtual Earth.

But having said all this, Google still seems to have the mindshare and momentum in the online hacker community, where people are developing lots of cool things, and in the blogosphere. KML has great momentum as a format for publishing geospatial data. But both companies clearly see the mapping / geospatial area as a major battleground and are investing significantly to try to get the upper hand, which can only be good for the industry.


Unknown said...

In relation to what was only previously acheivable in the 3D space, I'll agree that Microsoft is able to pump-out decent models with the Vexcel technologies. But, Google's steady but sure approach to producing highly detailed models is quickly becoming a favorite in the overall scheme of things.

If Google announces the use of the Stanford technologies to map buildings, then the production of detailed buildings in their case will far superceed anything that Microsoft's technology can offer in both detail and accuracy.

When you look at the context of the movements that each company has made over the last couple years - it's becoming clear who is the leader and who is the follower.

And these are the kinds of things that form an educated bias in time.

Aaron said...

Weird how some things change so quickly.....

With Street View and Mapplets, Google throws a few extra things into the mix this week.

From a consumer standpoint, the photos on Street View seem a lot more useful to the 3D models from Microsoft, and of course, being able to "walk" down the street has the wow factor to keep people coming back.

Mapplets looks pretty quick and easy to put together some lightweight "apps" from looking at the example code, so I'm sure we're going to see a million different uses (some of them actually worth using...) in the next few weeks.