Leaving Google
blog.douweosinga.com
blog.douweosinga.com
I've also been stunned at how few Googlers I come across are working on startups part time. Those that I find are workign on something on the side are very hush-hush about it, because of non-compete stuff. (It's hard to work on a software startup that doesn't directly compete with your parent company, when your parent companies mission is to organize the world's information).
Finally, I stopped by Google Ventures this week for a "Get to know Hackers & Founders" chat, like I've had at dozens of other VC's the past few months. Google Ventures was the _only_ VC that presented me with an NDA on sign in (via kiosk). The kiosk wanted me to sign a "I will never, ever talk about what I might learn about Google inside these walls" before got a name badge or got buzzed in the door.
I walked out before my meeting with the VC. If they are that concerned about what I might learn.... I don't want to know. There's a couple hundred VC's in town that I can talk to without having to sign or decline and NDA.
So, GOOG, there's reasons you're getting the "Google is the new Evil Empire" meme attached to you.
Frankly, Microsoft has been orders of magnitude more helpful to startups at H&F than Google, and we've been meeting within spitting distance of GOOG for 3 years. The BizSpark team has been fantastic to work with, and are helpful to startups whether or not they're on the MicroSoft stack or not.
Anyway, I think your perception is basically accurate, at least in my limited experience. Google is the wrong place to go if you're interested in working on a startup on the side. Between the draconian "all your IP are belong to us", the intellectual demands of your day job, and the numerous opportunities to try out interesting side projects under the Google umbrella, the path of least resistance is usually just to do your side-projects as Google-owned 20% time.
It can be a good place to learn a whole bunch of technical, leadership, and product skills and then do a startup. Google's quite a bit more open internally than most big companies, so you're exposed to a lot more than you would be at another large corporation, and possibly even more than in a startup. There are specialists in performance, accessibility, internationalization, security, and all these other tricky corners that startups typically muddle through, and the quality of documentation on how to handle those is much better than you'll generally find on the web. You get daily experience with scaling, such that thinking about how to build scalable solutions becomes second-nature. You're often directly involved in product decisions, and can see how your choice fare in the real world. You have a wide variety of hard UX data at your fingertips; much of this is cutting edge applied psychology, and isn't available anywhere else.
I think we're starting to see a trickle of Xooglers leaving the mothership and founding their own things. I know at least 6 of them personally, plus there're posts like this blog entry. In general, I think this is a very good thing for the Valley startup ecosystem, just like how Fairchild spawned almost all of Silicon Valley in the 60s, and Xerox PARC spawned much of the rest of the modern computer industry in the 70s.
It's kinda ironic that Google is totally open for users, but is a walled garden for employees. Meanwhile, Facebook is a walled garden for users, but is relatively open about letting employees publish and open-source infrastructure. (Actually, that's perhaps an unfair comparison: Google is also really open about letting employees open-source infrastructure, the problem is that Google culture tends to result in re-use of core proprietary infrastructure, such that when it comes time to open-source the software, it's impossible to disentangle from the parts that definitely can't be open-sourced.) Meanwhile, Microsoft is now relatively open for both employees and users, but is a walled-garden for developers.
Worse still, if you go and ask if a certain idea would be a conflict, you risk showing your hand. So, anyone who's serious about trying things is likely to leave, and it will only get worse as more business areas are roped in.
but, i do think you are over reacting about the google kiosk NDA. every visitor signs it. you are not special.
I'm suggesting that founders have a choice when it comes to investors. If I'm asked to accept or decline an NDA before I have a conversation with a VC, I'd rather go to one of the other hundred VC's to talk to in town where that isn't an issue.
You can decline the NDA and nothing bad will happen. All visitors are asked to sign an NDA, which is a silly policy but easy enough to ignore.
C'mon now. Outlook Web Access and Oddpost used Ajax-like technologies for webmail long before GMail's first line had been coded.
Also, while BigTable certainly has had some influence on NoSQL, I find it difficult to credit Google with the entire movement considering how long these types of DBs have been around. Petabyte-scaled high performance DB storage? Maybe. NoSQL? Nope.
I wouldn't credit Google with as big a part with the NoSQL movement. BigTable is a big innovation but it wasn't making much of an impact on the rest of the industry for a long time. Memcache probably had a bigger impact on the NoSQL movement, for example.
When gently informed (not reminded--most of the ones I know have never touched Outlook in their lives) that Outlook Web Access was the reason XHR was invented, the response so far has invariably been "but that doesn't count, because it's Microsoft."
I somewhat chalk it up to the "cool factor" that surrounds Google, but a good part of it may just be ignorance.
This is somewhat ironic, since Maps didn't use XHR at all, and instead used an iframe transport. For that matter, "AJAX"-enabled Google Search also uses an iframe and contains no XML. Monikers are weird.
http://bytepawn.com/readings-in-distributed-systems/
Some code:
Prevayler.org for instance has been around for many years, since 2003 or so ; the lisp code for doing prevalence has been around almost as long: http://homepage.mac.com/svc/prevalence/readme.html .
Apparently Klaus Weustefeld proposed some of this in 2001. An IBM developerWorks article is from 2002: http://www.ibm.com/developerworks/library/wa-objprev/
Ten years to an "overnight success" ... sounds about right.
He linked to this, though, which I didn't know about, and seems at first glance to be very useful and neat: http://sharedspaces.googlelabs.com/
At Google, there's a culture that if something is broken, call it out so that it can be fixed. At best, you will save thousands of employees hundreds of hours. At worst, you'll learn why it's broken. Many problems are inherent in running a large company, and there is no easy fix for them. But many others have simple solutions that make everyone's lives better.
The way I see it (and this is completely my personal opinion here, not speaking for Google), he and other Xooglers who've written posts like this are being Googley. You don't work for seven years at a company without absorbing some of the culture, and that culture doesn't simply leave when you hand in your badge. Posts like this are not all that different from something you would see on internal Buzz or mailing lists, and I think the spirit of them is not to "bash" Google, but to call out things that Google could improve upon even after the Xoogler has left.
Ever watch a friend slowly slip away? They change from something you knew to something unrecognizable. You have to at least try to do something about it.
The introduction - and then pervasiveness - of unit tests was one such change that basically took over Google culture despite the ambivalence of the founders. People found they could simply move faster when they had confidence in their code, they started sharing the knowledge of how to write better tests, and eventually it simply became the accepted thing to do. (Interestingly, the testing intergrouplet failed several times before they struck on a bottom-up approached that worked. As with everything, persistence is important inside big companies too.)
Or maybe not. Oftentimes, things don't even need to bubble up to the top to have an impact. You've got a very distorted picture of large organizations if you think that Larry suddenly makes a decision and then the whole company immediately does what he says. Large companies are really collections of small independent groups, and nobody really understands everything that's going on, and often times the worst problems are best addressed at the middle-management level. I think most rank-and-file Google employees have the same goals as Larry Page: we want to launch things quickly, we want people to use our stuff, we want to have a big impact on the world, and we want the share price to increase. The problem is translating that big-picture desire into specific practices and processes that'll result in that.