David Goldberg, eBay Inc.’s New Technical Fellow, Specializes in Ideas
By: eBay Inc. Staff
A veteran of Xerox PARC, Goldberg's early research inspired Graffiti stylus-based handwriting.
At eBay Inc., technology innovation has everything to do with progress, which explains the new arrival of a proven idea man: David Goldberg, Technical Fellow in eBay’s Buyer Experience group. For the past 22 years, David worked for Xerox’s Palo Alto Research Center (PARC), where he solved complex problems related to computer security, imaging, information filtering, user interface and data mining, and more. While at PARC, he worked on text entry for hand-held devices, and developed Unistrokes, a pen user interface that was the inspiration for the Graffiti handwriting system used on Palm Pilots.
Prior to working at PARC, David worked at Sun Microsystems designing and implementing various products in the networking and user interface groups, and was a member of the team that designed Sun Networked File System and OpenLook. He received his Ph.D in mathematics from Princeton University, has authored more than 40 publications, and is named on 33 U.S. patents.
eBayinc.com sat down with David for a casual conversation focused on his ideas, and his approach to generating new ideas. Here are his thoughts.
What have you been working on during your first few weeks with the company?
I’m initially looking at a few things. One is that we would like more items on eBay to have UPCs (Universal Product Codes). I’m also focused on improving our ranking in Google searches. These two things are related, because when our items have UPCs, they are ranked higher in the Google product pages, and that in turn helps their ranking in generic Google searches. I’m also looking at the product catalog as an avenue for working with UPCs. Creating catalog entries is currently very labor intensive. Making this process more automated is a challenging problem,
I’m also quickly coming up to speed on the many tools used here at eBay, such as Teradata, Hadoop, and more. Hadoop [an open source software framework overseen by Apache] is used regularly for tasks such as eliminating duplicate listings.
You come from Xerox PARC, which has a storied history, and was part of early graphical interfaces on personal computers, among other things. Based on your experience there, are you watching any disruptive technologies or trends that might have an impact here?
Social media and augmented reality are obviously making a big impact. One thing we worked on at PARC was electronic paper, similar to the technology found in the displays of electronic book readers. With this technology, devices don’t need to use power except when they are changing what’s being displayed, which has many applications. For example, signs at department stores represent one application. PARC wasn’t able to turn this into a successful product. But they didn’t consider signs that point you to related products at eBay. That might make the economics work out.
I think eBay applications such as Group Gifts represent the beginning of a big trend. I could see that expand, and become an app in a bigger system. As an example, what about group shopping? So, on your screen you’re shopping, and you can also see on your screen what your friend is shopping for.
Here’s another idea: Back in the old days, people would have a telephone book next to the phone. What could be easier than looking up someone’s name in the telephone book? And yet, many people couldn’t be bothered to do that. They would call 411 to find a number. The takeaway is that when one thing is just a little bit easier than something else, people will embrace it.
So what about the idea of a phone-based eBay personal shopper? You call them up and tell them what things you want. There, you have someone else doing the work. It’s an idea.
Do you see any big technology, shopping or information-consumption shifts happening?
In the future, when we look back on this time, I think many people will focus on what happened to newspapers. They are dying, and there will be new models. One way of looking at this is to consider new ways of financing news reporting.
Could eBay-sponsored online newspapers of various types employ reporters that could produce eBay-specific stories with links to eBay that would help engage people? That could be interesting.
How can search improve on eBay?
I think there is much room for user interface research. To make search better involves more than just algorithms. You have to reevaluate the paradigms through which users interact with the site. As an example, you type in “flute,” and right now you get flutes that you might find in an orchestra, fluted glasses, and a mish-mash of results. If you then pursue searching by price, that’s not going to make much sense.
Imagine an interface that comes back and shows you pictures of the different things “flute” could refer to, and makes you pick one before you proceed. That requires an extra mouse click, and interfaces generally try to minimize the number of clicks it takes to get to your goal. But in this case, perhaps this extra click early on in a search process can save you down the road. Your net experience might be better.
What's your philosophy on generating ideas?
Here’s something someone once told me that I really believe: If every idea that you have seems like a good one, then you don’t have anywhere near enough ideas, and the ones you have are probably not that good. The people with really great ideas have hundreds of them, many of which are completely ridiculous. When one of their ideas doesn’t fly, they quickly come up with another one.
The way I come up with ideas isn’t to get up in the morning and say, “Hmm, can I think of a new idea?” Instead I focus on a problem, and during the process of that, new ideas inevitably arise. For example, from a monetary point of view, the notable thing that I did at Xerox PARC was to develop a handwriting system that turned into Graffiti. The novel idea there was to require people to adapt to machines rather than vice-versa.
That idea developed gradually, as I struggled with the problem of how to do text entry on small hand-held computers. I didn’t start out to come up with a novel idea. I first tried the straightforward things, and when they didn’t work well I gradually hit on the single-stroke Unistrokes/Graffiti idea. That’s what often works for me — picking a problem, starting with the obvious possible solutions and if those don’t work, that naturally leads me to new ideas.