[Background: Worked as a researcher at IBM and a senior computer scientist at Adobe Systems; holds nine patents; earned a PhD in applied mathematics at Cornell University.]
THE COMPUTER AS INTERMEDIARY
"I had glaucoma at birth. Doctors managed to rescue a little bit of vision in one eye. I could see a little bit with one eye until I was 14, enough to read and write. I don't see anything at all now.
This was what got me doing the research I do. Information on paper requires a human to look at it before it comes to life. If you can put information in the computer, then the computer can act as an intermediary and produce it in a form that works for you -- [Braille] marks on paper, spoken to you, whatever. I have been sort of targeting this for a long time."
EVERY GOOLGER'S A RESEACHER
"A lot of people in engineering at Google would be in research at other companies -- that's where all the PhDs usually are. This makes it difficult for those researchers to take an idea the whole way through the development process: Technology transfer becomes a big problem. But research isn't off in a corner here. It's not an ivory-tower goal. Where does innovation happen at Google? It happens everywhere, because everybody does research."
TESTING RESEARCH ON THE WHOLE WEB
"In my past experience at other companies, if I had said to a group that was about to ship a product, 'I have this idea I want to try on your code, can I?' they would have told me, 'Go away until we ship. Go write a paper or something.' The Google team told me, 'Sure, here is the code. It does this and this. If you want it to do different stuff, we'll tell you how to do it."
Normally, you build a small research prototype and then someone else takes it. So if you give researchers the option of 'Would you like to build a toy system that works on 50,000 systems?' versus 'Try your idea on the whole Web?' it's a no-brainer. Researchers would kill for this opportunity. You're seeing your innovation turn into something real. It's worth the pain involved in doing that hard work. It's not fancy. It's sweat and elbow grease. It's engineering."
FINDING RELEVANCE IN A DATA BLAST
"Let's say you have something highly interactive online, a basketball game, and the display shows you who has the ball, how many points he scored last month, how many fouls he has. That's an incredible amount of information being blasted at you. It's an obstacle for someone who cannot see. You're lost.
I try to overcome this problem with technology. Where your eyes do the work, I want the computer to do that processing and tell me who's got the ball and who is likely to score. That's the Web today. We are blasting data using fancy user-interface tricks and letting the human eye figure it out. It's difficult, but the computer to take over the cognitive load of understanding what's relevant and giving it to you.
Accessible search analyzes Web pages for complexity. It's a hugely complex formula that takes all kinds of numbers into account and then cooks up the search results you need. The change is subtle, but it takes into account the usability of the page."
ACCESSIBILITY MEANS DIFFERENT THINGS TO DIFFERENT PEOPLE
"When somebody asks if the Web is more accessible, I ask, 'More accessible or less accessible to whom?' Something that is all audio, like a podcast, is perfect for a blind person but useless for somebody who is deaf. I don't look at accessibility as a half-full or half-empty cup. I'm just happy the cup is a lot bigger."