Last: stewart butterfield
Next: bbc ican
April 25, 2003
google keynote
google, innovation and the web
Craig Silverstein, director of Technology
ingredients
a company needs some things to be successful - not universal, but these work for Google:
mission statement - organise the world's information and make it universally accessible and useful
Can directly tie products to this mission statement - catalogs.google (paper), google on phones (universal access), spell correction (useful)
Keeps the company from straying too far from the field - there's 200 years of stuff to do just with this statement.
Trying to do things that matter - this focus helps them hire people, flexibility to not be evil.
Relentless focus on the user - e.g. advertising - text ads rather than popups
Google labs - trying to find what's useful for people. USer studies just offer limited feedback, better to get it into hands of users and see how they use it.
Switching costs for search engines is very low - just type a new url (all short urls are taken now though).
Brilliant people have good ideas - hiring is the key to Google and their process.
creative environment - attracting good people is easy and hard - free lunch, dinner, massage, keeping people motivated by getting products out the door.
We use process that works:
1. ideas come from everywhere
but bad ideas come from everywhere - have to be able to filter them - but this should happen late in the process.
Touchgraph google browser
2. design for users
think about design very early in the process
advantage as company - comes from hci background
delta force of ui experts that come in in planning stage
tricky business - lots of effort to make it look simple
good user experience isn't just what a user sees, but how they see it - speed etc.
3. compile, discuss, prioritize
has a top 100 (actually 230) of projects Google wants to do, in vague priority order. makes sure keeping all parts of mission statement as followed.
project idea board (sparrow from xerox parc) - anyone can add
All ideas are discussed every week by anyone who wants to join in (product discussion forums) - 5-10 minutes about products that have been added - brainstorm - can get feedback and refine
iterative process - wireless reprioritized after watching stats (big jump in Christmas 2002 (signed new partner))
reprioritized every week - go down list and check still right priority
4. small teams are fast and agile
teams of 3 people - they own product - they do design, code, test, launch - everything! people have to feel comfortable in all the roles. they get to do different things all the time, and also get to launch their own products - very motivating. related teams help each other.
5. communication is key
100 small teams - have to communicate to function. very broad, very regular. teams need to talk to others outside their realm - not hierarchical. design process - project notes are sent to all people in the company (with a few official reviewers) - everyone hired is interested enough to plisten and comment on everything. code is reviewed before is can be used - checks it is well written and easily understood by others (+ stops reinventing the wheel). Content is more important in code than the syntax. 10 inter-teams that oversee communication - volunteer positions (over and above day job).
Tech talks - every week people talk about projects near completion. Videoed, put on web. Can get up to speed on old projects very quickly.
6. tools to organise
editable idea page
launch process checkpoint page
engineering weekly report - everyone writes a few lines about what they've done and what they plan to do. normally takes 30-40 minutes to read it - then people write comments and suggestions and feedback.
internal blogs - foster communication betwenn engineers. first implementation of blogger in google.
7. test, experiment, iterate
used to just put it up and wait for email
now do user study to iron out any big problems that were missed
if not sure how primetime it is, it goes on google labs first.
(bad employees are a time sink - conservative hiring has meant only 1 or 2 that haven't benefitted Google)
((shows three different versions of Google News))
[geeks don't understand the virtue of whitespace]
gave up on information above the fold - never enough room to put everything wanted, so added summary to top, and lets the user scroll if they want more info.
the web changes everything?
what makes this process work - culture, and technology.
Wouldn't have been possible before the web.
Early search technology was sufficient before the web (even early search engines when the web was really small).
Wouldn't be able to communicate as a company. Web allows easy communication between teams (if people use the tech properly) - easy archiving, easy search. Also need users to give feedback - both through logs and email from users. Know they have enough data to prioritise successfully.
Stay true to your mission - + luck! Great stuff that does something useful. Battlefield is in the technology - marketing doesn't give lock-in on the web - user experience does.
Q: Google opening office in NYC - how will that change collaboration?
Hardest is timezones, plus NYC is 9-5, Google often works from 12... have to use web efficiently to keep everyone in the loop. Same development in NYC as west coast.
Q: (Weinburger)
How many pages actually are there? (Google only indexes 3 billion)
crawl as much as we can in 3-4 weeks. Lots of pages they can't - robots.txt.
Q: what about Applied Semantics?
content ad company... can't comment more.
Q: do business decisions change focus and priority?
gatekeepers make decisions - get down from all ideas to 100-200. People have to trust that every idea is evaluated fairly.
Q: google is better than napster for finding mp3s? (stop right there!) what are you going to do about that?
Google doesn't have a music or video search for IPR reasons. Have to respect IPR. Tricky to reconcile with mission statement. Try not to spit in the face of copyright owners - music search has little/no legitimate value.
Q: what are the limits on a workplace culture like Google? Not enough cool people to hire, only works until so big?
Hard to know the answer - companies have to try these things out. Google technique is brittle - so hiring technique is essential. It would only take one or two bad employees to bring down the culture.
Trackbacks
TrackBack URL for this entry:
http://www.anti-mega.com/cgi-bin/mt/addmttb.cgi/39
