Google I/O: Where Google Goes ...

Last week I attended Google I/O 2009 in San Francisco. This conference differentiates itself from other vendor conferences by the type of attendees it draws; it could be categorized as a D2D conference ("developer to developer"). There weren't many customers, only a small number of enterprise partners - the bulk of the audience seemed to be freelance developers. This may be due to the fact that, unlike other vendors, Google sells little of the company's technologies (with the exception of a couple of enterprise offerings such as Google Search, and its geospatial products). Most of its tools and applications are free and ready for programmers to use, adapt and develop. It is that model that has made Google so successful in the past and will likely keep it successful in the future.

There were roughly 4,000 attendees, which is impressive considering that the show is only in its third year and doesn't get much publicity. But then again, with the number of products Google has out there, it is bound to draw developers for one reason or another. Attending the conference, you can't help but be impressed with the reach Google has with its product line, from application hosting, to search, to Web toolkit, to geospatial products and many more... That said, there were just a few geospatial presentations and most, if not all, were focused on Google Maps and not Google Earth. Nevertheless, there was enough content for a geospatial person to find it useful and it is also a good opportunity to expand your knowledge of the Google world.

Day 1: HTML 5
The conference kicked off with a keynote speech by Vic Gundotra, vice president of engineering at Google, who presented, along with a number of collaborators, Google's vision of the future of the Web along with what excited them about it. From that presentation, I could extract (based on their comments) a few of the major trends within Google's overall vision. First of all, the Web has won; it is now the platform of choice. Gundotra mentioned that 10 years ago nobody believed that a product from a company called Keyhole (Google Maps) would be able to run purely in a browser...

Second, it is clear to Google that the next killer app will be Web-based and will leverage the new capabilities of HTML 5 (more on that below). Also, Google is fully committed to JavaScript. Considering that JavaScript performance increased a hundredfold in the last decade, Google is banking that all future apps will be developed using the browser capabilities without requiring extensions or plugins. Finally, and this has not been mentioned directly, but I got a sense that for Google, the problem with the Web right now is Microsoft's slow adoption of standards such as HTML 5.

As for HTML 5, the new standard is seen as the way to improve from Web 2.0 applications to the future. It is currently supported by the latest versions of Chrome, Firefox, Safari and Opera, but not Internet Explorer just yet. Officially it is supported, but there are a lot of shortfalls. Google has identified five "exciting" key points pushing them to go forward with that standard:

Canvas tag: This new tag provides extremely fast vector manipulation on the client side using only the browser and JavaScript. It will let developers build applications in the browser that were only possible in the past by using plugins or VML.
Video tag: This will allow the integration of video in applications, in the same way the image tag currently works.
Geolocation: Google is working with many partners to integrate, as part of HTML 5, a feature that would locate the browsing user using IP databases, Wi-Fi hotspot information, cell towers and other tools. If this becomes part of the standards, it will be possible to build location-based applications that know the approximate location of the user.
App cache database: This will let Web applications store user data and navigation information in local databases within the browser, hence enabling the applications to work in an offline mode and synch back with the central application once the machine reconnects to the Web.
Web workers: Everyone knows that too much JavaScript can freeze a browser. This new feature allows JavaScript intense processes to run in the background without affecting the user experience.

Other subjects were addressed during the keynote presentation, such as Google's app engine, a platform that lets developers run their applications on a Google hosted environment; the Google Web Toolkit (GWT), a toolkit that simplifies JavaScript development and cross browser compatibility; and the Android phone, which every attendee received as a gift to enable the development of new applications. All in all, it was a very interesting talk with lots of information and a good introduction to the rest of the show and the various tracks, nine in all.

Geospatial Stuff?
As stated previously, the bulk of the geospatial presentations were focused on Google Maps, with very little on Google Earth. It is obvious that use of Google Maps is more widespread and has a much bigger developer community. Of all the presentations on that subject, the one hosted by Pamela Fox (well-known in the maps developer community) and the one given by Susannah Raub were definitely the most interesting.

Fox's talk on performance tips for mashups provided a good overview of the Google Maps product, its limitations and how to work around them. It also very briefly introduced the new version of the Google Maps API, which Raub addressed in her presentation in much more detail. She identified the three main performance problems with mashups:

Loading the JavaScript API: This is the main reason why a version 3.0 of the API has been released. The download time for the API has been reduced threefold.
Too many markers on the map: It is well-known that above 100 markers, the performance of the map is drastically decreased. This can be circumvented by various methods such as clustering, generating heat maps or using a secondary rendering engine to generate tiles on the server containing the points.
Polylines or polygons are too complex: The problem is similar to the previous one. The solution to this can also be to use a secondary mapping engine or to use the vector capabilities of the browser (canvas tag or VML for IE). However, these two approaches also come with limitations at some point, depending on the number of nodes in the entities.

As stated above, a new version of the Google Maps API has been released (version 3.0) and was presented by Raub. This new API was built specifically to reduce download times and for mobile devices. It is absolutely not backward compatible with the previous version of the API and, at this point, does not offer the same functionality as version 2.0. The early adopters of this new API will be mobile developers wanting to reduce load time (from 20 seconds to 7 seconds) and also to streamline integration of mobile browsers (currently supported are iPhone and Chrome in part), since Google handles all the compatibility issues for the developers.

The current roadmap for the product is to complete the set of functions to offer the same level of flexibility as version 2.0 of the API. The release cycle for this API will follow Google's mantra: "Release early and often..." At this point, there are no plans to port this new approach to the Flash API. Version 2.0 will live for a while (time unspecified), but all applications will eventually have to be ported to version 3.0.

Day 2: Google Wave
On the second morning, Google introduced its new product, Google Wave. Much has already been said about Google Wave, but it is impossible to talk about this conference without mentioning it. Google Wave is presented as a collaboration and communication platform for the future. This platform is currently in its early form, but essentially the company has conceived it as a way forward for e-mail. The platform will be open source and will let developers build and add extensions to it. This approach, similar to Google Maps, should help to promote wide adoption of the product. Whether this platform will succeed is unknown, but judging from the favorable reaction of the crowd, it has the potential to revolutionize the way we work and collaborate.

Google Wave is entirely built using the Google Web Toolkit and HTML 5. Impressive demos were performed of real-time collaboration between multiple platforms across the Web. One particularly impressive demo had one presenter writing text in English on one screen and the other seeing it being translated into French in real-time on his screen. I have to say that the translation was perfect. It was using a new algorithm that promises to finally deliver on translation, something that has still eluded us despite the number of products and Web applications out there. Here are some interesting high-level features:

Real-time collaboration on documents, or WAVES as they are called
Playback of e-mail (or anything else) history
Everything is embeddable and still fully dynamic and linked to original wave
Live with mobile devices, as well

Google I/O is obviously not a conference about geospatial; it's about Google and its vision of the Web. I have to say that from what I saw, the future looks very bright and Google's place in this ecosystem is definitely at the top of the food chain. What will be highly interesting for us geo people to watch now are what new gates HTML 5 will open with regard to Google Maps features and applications.