“Web 3.0 will ultimately be seen as web applications which are pieced together.”
- Eric Schmidt
In the wake of Google’s AppEngine announcement, I want to remind you of this video I’ve posted before: Eric Schmidt talking about Google’s vision for the Web 3.0 computing era.
To me, AppEngine seems like the most logical evolution of Google Gadgets! (This is what everyone thought Eric was talking about back in that video.)
The Real Story
However, every story written about Google’s newly launched AppEngine has incorrectly correlated this new service to Amazon’s EC2 and S3 services.
I think these people are way wrong and that if the Silicon Valley echo chamber wants to make up a competitor for AppEngine, its proper correlate (by a whisker) is Facebook’s F8 platform. If you must cram this new service into a pigeon hole, think of AppEngine as the Facebook Platform for the grown-up web.
Why isn’t AppEngine like EC2 & S3? Constraints, constraints, constraints.
- Google is going the run-time environment route, not the scalable, “put anything you want in a box and we’ll scale it” route that EC2 provides. Case in point: we could run the BricaBox Platform on EC2 by tailoring our own environment (the LAMP stack) and booting it up on Amazon’s servers. We could not get BricaBox running on AppEngine without re-writing in Python, ditching functionality which needed outside libraries or languages, or relational databases.
- Google is not trying to provide pure utility here, they are trying to provide utility tethered to their infrastructure. While EC2’s initial investor pitch was “we have this scale and want to monetize unused portions of it,” many smart people called them out for what they were really doing: creating a new business based on the concept of utility computing, which had nothing at all to do with their core infrastructure at Amazon.com (i.e. they weren’t using EC2 or S3 for their needs).
The point about tying it into their infrastructure, however, is an important one. Google is clearly looking to have as much of these apps tie into existing pieces of Google’s infrastructure: everything from the authentication systems (Google Accounts only!) to code libraries (most open source or API accessible code from Google is written in Python and ready to run in their environment) is based around this infrastructure and will be based around this going forward
So, why is AppEngine more like the Facebook Platform 3.0?
- AppEngine is designed for lightweight apps
- AppEngine apps are wrapped around Google’s existing userbase and communication infrastructure (Gmail), creating a more organic Social Ecosystem for Google
- AppEngine is way more inter-operable with the rest of the web than Facebook apps, though still built around a proprietary stack (Google’s runtime env is the equivalent of FBML)
- The first AppEngine apps will be huge! The next ones will have just as much trouble emerging from the murky waters of the web as any other apps
You've found Nate Westheimer's blog. Nate wears many hats. He's the Entrepreneur in Residence at 
4 responses so far ↓
1 Kareem // Apr 8, 2008 at 10:22 am
Hmm, interesting viewpoint but I disagree. The key benefit of AppEngine is commoditizing scalability. I imagine AppEngine will be a boon for Facebook developers and actually lower the barriers to entry significantly. Even with services like EC2 & S3 it was very hard to build scalable FB applications until now…
2 Wayne Mulligan // Apr 8, 2008 at 1:19 pm
Interesting comparison to Facebook Nate, but I think you’re a bit off the mark with respect to EC2/S3 and what Google’s real aim is here.
EC2 is not inherently scalable, only after plugging in services like RightScale can anyone hope to have a reliable, redundant and scalable version of their app up on AWS. Google reduces that need by tethering all of their services together.
Also, they don’t tie the apps built on their platform directly into Google’s API’s and tool sets - they’re options that you can use but you can also choose to have your app running on your own domain with no linking to any Google services at all.
Also, Python is the “first” language they’re going to support - not the “only” language. They’ll be launching more languages soon, so I’m sitting tight until PHP is up there. But I have been playing with my beta account and it’s pretty cool for such an early stage product.
And I think the real goal here isn’t to compete with Facebook, but rather become to the web what Microsoft was to PC’s — the platform that all software is built upon. While Microsoft did that so they could sell more software, Google wants to do it to sell more ads. An ancillary benefit is that it’ll probably make it dramatically easier to identify and integrate acquisition candidates - GoogleFund? haha.
But I think that the comparison to Facebook is still interesting and will definitely be interesting to see how these apps tie into OpenSocial and OpenSocial partners networks. Because as of now, Google doesn’t really have any viral marketing mechanisms built in the way Facebook did.
-Wayne
3 Nate Westheimer // Apr 8, 2008 at 4:15 pm
Good points fellas. I guess my point is not what it is TODAY but what it will shape up to be in 10 years.
In 10 years, EC2 will be the best way to run any kind of application on it.
In 10 years, AppEngine will be a dev platform especially useful for applications tied into the Google Stack.
4 Q dub // Apr 9, 2008 at 4:57 pm
I don’t quite get it…can a Facebook app not sit on GAE or AWS? If so, they’re not really competing. GAE will be most detrimental to traditional webhosts who offer LAMP and MS Server stacks - serving small/medium sized companies who don’t want to invest in expertise to scale.
Leave a Comment