But Google Earth was an acquisition. So was YouTube, despite the effort they put into Google Video. So was Google Docs word processor, although the spreadsheet was home grown.
Google has, what, 20,000 engineers? That's a LOT of effort for what they have to show for it.
That criticism would hold true if Google were a pure product company. Salesforce is a great example of this: in their SIGMOD/SoCC keynote they've mentioned they only employ 200 engineers. They don't face the sort of scalability and algorithmic challenges that warrants developing custom infrastructure: they use of the shelf commercial products and open source products (Oracle RAC, Lucene).
At Google that great deal of work goes into amazing internal infrastructure that is rarely read about (I'd imagine the published papers are only the tip of an iceberg) i.e., people hacking on the kernel, on file systems, on databases, firmware etc... This infrastructure is their competitive advantage and their moat. At Salesforce, their ability to out execute competitors (via aggressive selling and marketing) on an innovative but not inventive business model.
In the case of Google, being a major contributor to the Linux kernel, they're effectively their own operating systems vendor. With BigTable, MegaStore, Spanner, their MySQL patches and more, they're their own database vendor. With Chubby, GFS, MapReduce, their own web server, etc... they're their own distributed middleware company. They're also one of the largest hardware manufactures, with themselves as their only consumer.
In terms of their core competencies (search/IR/ML), they're an incredible research organization. Not all research results in products, but they've been able to productize their research much more effectively than Microsoft has (perhaps at the cost of having a less theory and more systems focused research organization i.e., employing Rob Pike but not Leslie Lamport).
I have a feeling it's great to be a researcher or an advanced developer (working on infrastructure or algorithms) at Google: you have the capability to focus on actually building incredible pieces of software. In terms of application development, it might be slightly less appealing (although, I've been extremely impressed with Android, Chrome and Gmail as _finished products_ too).
Of course, if someone's _primary_ goal out of a job is to become an enterpreuner and found their own company, going to Google may not be the best place for this (other than for making connections and gaining credibility much like going to Stanford): building an infrastructure start-up is much more difficult than building a web application start-up. For those who are first and foremost technologists (that would be me), Google still symbolizes the "ideal" job: perform incredibly advanced development (implicitly frowned upon in most product companies) and build practical, working software that gets put to production use (implicitly frowned upon in research centers and academia). At one point one such place was AT&T Bell Labs, at a later point it was Sun. Now that place is Google.
that sounds about right! :)
I worked on Google Wonder Wheel, which had a tiny, startup-sized development team. Probably 99.5% of you have never heard of it. However, the people that do use it really like it. Enough that they complained when we temporarily disabled it for the recent visual redesign, another project I worked on. It has orders of magnitude more users than Bingo Card Creator, and has generated orders of magnitude more revenue for Google. Yet you hear about BCC all the time on HN, while maybe once every 6 months somebody might mention Wonder Wheel.
1-2) AdWords and AdSense revolutionized web advertising. Those two products generated $6.5 billion in revenue just in Q1 of 2010.
3) Android. Others have commented that this was an acquisition, but that happened in 2005, and it was a relatively small team. I'd wager that most of the current Android code was written after the acquisition, and by engineers who weren't from the original Android team.
4) Chrome.
Regarding the number of engineers, Google had 20,000 employees as of March 2010. You can make your own guess about how many of them are software engineers, but it's a lot less than 100%.
In any case, I consider Google Maps to be through and through a Google product.