Posts from SIGCSE: Marissa Mayer’s 9 lessons from Google

Posted on March 15, 2008


Marissa Mayer is giving the keynote at SIGCSE. In her talk “Innovation, design and simplicity”, she’s talking about 9 lessons from Google. She’s given this talk before (Stanford podcasts)  but the insights for computer science education and the Q&A are new.

Here’s my notes:

1. Creativity loves constraint

Sometimes best ideas come from constraint. Eg minimalist Google homepage. Simplistic statement Sergey Brin – “we didn’t have a webmaster and I don’t do html”.

User study: “I’m waiting for the rest of it”.

Focus on search, this constraint has opened up “whole world of possibilities

2. Problems that matter and ideas that are simple to grasp

Want to grow through word of mouth, based on simple presentation of ideas that are easy to grasp. Simple interface but complex system (also massive scale and speed). Key is that users don’t need to know that.

Search is just getting started

Today we’re doing simple things, should also

  • overhead view
  • noun and verb
  • “how much does it cost for an exhaust system”
  • trying to better understand cognitive processes and language

Also have to have answers to questions, so focusing on material for answers. two strategies: Content and universal searches.

Each of these has huge impact. eg impact on indexing as they start to look at movie content.
3. Ideas come from everywhere

Need to be prepared for ideas. Some top down (eg google maps), some bottom up . Institutional support for bottom up: eg she has open office hours (4-5 everyday).

Story of question of “How are we going to make money from email?”. Paul had seen an opportunity to take Adsense and connect to abstraction of email: Adsense. Same concept is working widely: benefit from opportunistic insight (mixed with skill set to do something with it.

Similarly, Google book search (insight: non-destructive scanning)

How can we grow search? Increase accessibility to internet – WiFi routers

4. You’re brilliant. We’re hiring.

Core to innovation at Google is giving really smart people great resources and freedom.

5. Innovation, not instant perfection.

Key is launching early and often. Two schools of programmers: castle builders, and nightly builders. In innovation, Apple is castle builders. Doesn’t work in software as can’t spend 2-5 years. As soon as have something, launch it – get feedback from users as soon as possible. Eg Google news went through 64 iterations.

Acceptance that won’t get it right. Once have the creativity, then strong role for science. Don’t guess whether people prefer different colours – release in parallel and test.

6. Users, Users, Users. (aka Users, not money)

Focus on what users can use, money will come. Search feedback before started advertising consitent “don’t put banner ads on”. Which was difficult as that was the plan. So they used

Advertisements often as good as search.

Most users want clean webpage, but some want more – customisation etc.

7. A license to pursue dreams

20% time: you can do want ever you want. And, people can choose how they use it eg once a week, every fifth month etc. In 6 month window 50% of new products came from 20% time. Why? Impassioned development. eg streetview, google sky (and note benefits back to core business).

Beautiful accidents

Transit came from group of engineers who “wanted to do the right thing for the planet”.

User generated content, beyond “funny videos” eg: discovery of roman villa, Santiago’s transit map.

They are explicitly supporting such activity eg: My maps; Sketch-up 3D models.

Map of user generated content.

Several 20% projects humanitarian projects, eg Darfur project in collaboration with Halocaust Museum.

8. Share everything you can

be proud of what you have done and share with others:

Story of development of Google in your language…Sweedish chef… This is what it would like if muppets ran the place: Bork bork bork.

9. Ideas for computer science education

Insights from Google engineers: What do you wish you had?

a) applications integration

b) understanding statistics and messy data. Science

c) large projects and legacy code:

80% of all developers are working on someone else’s code

d) scale. In college, do as many hundreds of tests, works. But scale might mean millions of queries a second.

scale crushingly large

e) Resilience and robustness. Fail fast, scale fast

f) Working in teams. None of these projects done by people working alone.


Q: Ethical dilemmas: China

A: Slogan: Don’t be evil. Huge debates inside company. In case of China balance of engagement and estrangement. Working at the boundary so they carefully listen to ethical advisory boards.

Q: Two versions of webpage, how can they tell which is better.

A: Split testing. 1/1000 users might get new experience. Measure search volumes, click throughs, long click metrics etc. So measuring abstract.

Q: Why does she use “engineers” to describe people in computer company.

A: She’s using to distinguish marketing led versus technology led. Google is decidedly technology led. So why Engineer and not Computer Scientist?

Q: Google’s plans to support Computer Science Education

A: A lot of outreach. Awareness and activity.

Q: Usability. Have they thought about publishing their huge user studies?

A: Much competitively sensitive, but they do try and be as open as they can , 4-5 papers at SIGCHI

Q: Some products confuse language and culture. eg looking for news in Spanish, assumes you live in Spain.

A: Complex problem. User, language, user preferences and unwillingness to switch default (only 2% will switch), so offer in language/culture in location of IP query source. One size fits all solution not right and

Q: back to Engineer question. Why not Computer Scientists? Would be an enormous help if the role models were described by Google as computer scientists.

A: Very good idea. Exposure to computing, course to draw people in, exciting careers.

Posted in: sigcse