Google, Innovation, and the Web April 25, 2003 - O'Reilly ETCon Craig ====================================================== Could people re-edit their e-mail addresses now please style just e-mail: tom@plasticbag.org, phil@gyford.com etc ====================================================== This is new info not products or vision a process for how we make google products Google ingredients: mission statement organize the world't information make it universally accessible and useful refs google catelogues, google on cells mission statement actually keeps google busy, yet focused do things that matter attract people who want to matter flexibility to "not be evil" relentless focus on the user resisted pop-ups, despite needing money invented inline text (pyrad style) ads google labs used to refine ideas that they like, but don't know how to implement users take part in the experiments since search engine switching cost is low, google remains focused on the user brilliant people have good ideas a creative environment helps beanbags! [that's so Xerox PARC in the 70's (we did all that with etherphone! *smile)] A process that works: ideas come from everywhere (employee, management, external people) collection is important collection in ways that users can't see touchgraph google browser [1] design for users Larry Page, the origin of "PageRank" name design early, design often shows early google splash (speed, ease,...) compile, discuss, prioritize compile: top 100 list of projects that google wants to do, prio by importance built on PARC's sparrow (like Wiki) [2] discuss: relaxed product discussion forums 5 to 10 minutes about changes to the list evaluate: graph usage (jump at new partner for wireless access in October ... Danger?) kill unused projects small teams are fast and agile communication is key Tools to organise editable webpages it looks like a form for people to sign-off before launch weekly snippets posts small details of what they've done each week takes 45 mins to read each week weblogs loads of speculation in the press: No one speculated that they might use it internally. first thing that Blogger team suggested. great ideas that people have determine great products. Design Process User studies get the really obvious bugs out of way. First user-study ever had feedback like this: "I'm waiting for the rest of it" "Is this some guy's homepage" "How many people work in this company" "Are you from the psych department." Experiment Labs allows people to experiment internally. Hiring Doesn't look for experience so much as ideas. One hiring committee that goes through every CV that comes in. Separates the hiring process from the headcount process. Keeping up the standards for hiring is more important than everything else. VERY CONSERVATIVE HIRING PROCESS. Only ever hired one or two people that were not a benefit to Google. The bad employers are the time-sinkers. Experimentation and Implementation process (News) Not practical to have anorexic news page Iterative process (three slides about Google News) Did the web change everything? Maybe not, but... We wouldn't have had Google Early and pre-web could use very bad search engines Small levels of information Wouldn't have been able to share this technology without web interface We wouldn't be able to communicate internally so well Memos don't wor as well as editable web-pages Feeback - we couldn't have had the amount of feedback Logs about how people use their product Advertising software is effective cos it's fast Ads are dropped if they're not clicked on User-focus STAY TRUE TO YOUR MISSION: never undestimate luck - Google in right place in right time... know your mission statement get people doing something *useful* Addition of Q&A at end: Q (Dweinberger) Why are you so stngy with pages? Aren't there more than 3B? A: Limits are in terms of load. They crawl as much as they can in any time frame. Limitations on far end (sites). And a lot of pages are blocked, etc. Q: What are you doing w/Applied Semantics A: Won't talk directly about plans (policy). Targeted analysis, probably to be used in advertising bysiness and technology. Q1: Semantic tech could be applied in more than ads A1: I agree. It's semantic. I can't say more than that. Q: Importance of engineers and center of process sounds good. But if you put me at the center of the process, I'm going to search for nichy things. A: Yes, we can't skew information by small set. We do have unrelated ideas, and they do get captured, but some ideas are held apart by gatekeepers. There are decisions made. For the system to work, you have to have trust that your ideas will be heard and evaluated fairly. Q: RIAA just sued college students developing MP3 internal network. Is Google not better than Napster for finding MP3s....? Are they not at risk? A: People always want a music search. And the reason we don't have it is because of these IP issues. We have the goal to make the world's info available to people. The method is not to spit in the face of IP. But if it has legitimate use, we try to make it available. We only nix content if the owner requests it or if it's abusive (stealing rank in engine) Q: Thanks for nice presentation. But what are the limits that other companies aren't as great? Are there too few smart managers, or too few smart employees? Or does this small team etc work only in internet world. A: Shrug. Enabling tech is so new, it's hard to see a history of experimentation with these techniques. The technique of small teams is quite brittle, which is why the hiring process is so key. Have to prevent bad-actors, have to be eternally vigilant. Every company has to make their own trade-offs. ========================================================= Could people re-edit their e-mail addresses now please style just e-mail: tom@plasticbag.org, phil@gyford.com etc to make it easier to mail. ======================================================== Add your email to get the text of this document after the session Trevor F Smith | tfsmith@parc.com | http://trevor.smith.name/ esinclai@pobox.com, matt@blackbeltjones.com, mgraham@well.com, phil@gyford.com, richard_gayle@excite.com dan@wde.com, jjg@jjg.net, jmay@pob.com, allen@hutchison.org, tom@plasticbag.org, tfsmith@parc.com, etcon@crystalflame.net =========================== External refs: [1] http://touchgraph.sourceforge.net/ [2] http://www2.parc.com/istl/projects/sparrow/ Google celebrates the Double Helix today ( http://www.google.com/ )