Esther Dyson on the Portal Wars

Esther Dyson, technology visionary and chairman of EDventure Holdings, shares her views on mobile Internet access, the latest version of the portal wars, and why defining data is so important these days.

Esther Dyson [EDventure Holdings] | POSTED: 06.08.03 @14:47

AlwaysOn: One of our popular themes on AlwaysOn is 'The Wireless Access Boom.' What do you think about it? Dyson: To me, that's 'freedom at last.' What's been so great about the Internet is that you just get up on the Internet and offer something to customers or readers or whoever. In telephony, you've had to go through the carriers and operators, but now you're getting more and more of this Internet-style of operations in telephony, with things like VoIP [voice over IP] and Wi-Fi. Suddenly people are no longer at the mercy of some incumbent local monopoly carrier, whether it's the consumer or somebody trying to reach the consumer with a third-party service.

AlwaysOn: What about Web content -- has that whole sector just come down to advertising?

Dyson: Well, you have AOL and Yahoo, which are trying to look different, but in many ways they're converging on a lot of the same offerings, though they're coming from different places. AOL has its existing customer base that it needs to retain and get to trade up, and Yahoo trying to build such a paying customer base.

Because AOL has always had a subscription business, I would say they tend to be a little more an agent of the consumer, whereas Yahoo's background is selling the consumer's attention to advertisers and marketing partners. Both of them are trying to get revenues from both business models. They're trying to figure out how to leverage broadband. We had a panel on that at the conference, with some of the leaders in that area, Jonathan Miller from AOL, Dan Rosensweig from Yahoo, and then to keep them both honest, Shelby Bonnie from CNet, who's got a very different model, much more vertical. His ads tend to be very relevant to the content, and higher-margin. AOL and Yahoo have to figure out how to be better at targeting. And at the conference we also had Sergey Brin from Google, which recently announced its new advertising network based on AdWords. It's really brilliant.

AlwaysOn: Yes. I interviewed Eric Schmidt for AlwaysOn, and at the time he wasn't sure whether this was going to be a marginal money-maker or a huge money-maker.

Dyson: I think it's huge, because they're using the page basically as a query. They're saying 'This is the page you're on, this must be what you're interested in -- and unless you're on the page by mistake, it is what you're interested in. Now, if you're looking at Roman history or something, I'm not sure what you'd want to buy. Maybe books about Roman history? I went to this site that Google's working with called How Stuff Works. I was looking at how an espresso machine works, and they had two ads for espresso machines -- really handy! Then I went to something for scuba-diving, and they didn't have an advertisement for that category so I got a regular ad -- for a Visa card. So I'm thinking, 'No thank you.' The ability to sell ads tightly targeted to what the consumer is looking at, that is really going to change things. The notion of getting all these irrelevant ads is going to start offending people, and you'll go back to the way the world used to be: When you ask someone where they had their hair done, you don't want them to start talking about their bank.

AlwaysOn: Isn't that funny, how first the Web just made everything accessible, and now, going forward it's going to niche everything out.

Dyson: Well, yes and no. Thank God, there are still people with eclectic interests. But, yes fragmentation is one of the issues. The world is going to change.

AlwaysOn: At PC Forum, you were also talking about what's beyond relational databases. How does that play out?

Dyson: Well, here the issue is that the world is much more complicated than what you're going to find in your typical database. First of all, things move around and there's a lot of new and interesting technology being developed to define data better and deal with things like physical locations -- RFID tags are going to be a big thing. Retailers are extremely interested in them, because now they make it cost-effective to track things in much smaller amounts than just truckloads. There's a standard for PCs and there's sort of a standard for relational databases, but when you start talking about things that measure temperatures, or meters and sensors and vision devices and all this stuff, there's not much of a standard. Maria Martinez runs a company called Embrace Networks, she's building a kind of layer that helps broker between these different devices and lets them communicate with legacy databases and applications.

There's an overarching theme here. If you want to communicate between interfaces in order to use Web services, and you want to build out this wonderful connected, rule-based Internet, you need to do a good job of defining your data. Or you'll find that you've got the communication network there, but you're talking French and someone else is trying to speak German. Or worse, you're both talking the same language, but like the UK and the US, you actually mean different things and your systems get screwed up. A great example of that is with a Valley company, where in their production there was just something weird, the numbers were just never right. This was weekly production numbers. What they discovered was that the guys in Taiwan were starting the workweek on Sunday in Taiwan, and they were counting that production for week X, and California would start in the workweek on Monday in California. You had about 18 hours of production in Taiwan that was being assigned to one week in California and to a different week in Taiwan. They just couldn't discover it, because they weren't being explicit enough about defining when the week began. It led to production screw-ups and all kinds of things.

AlwaysOn: That's a great story.

Dyson: These are teeny little stupid data things, but they matter a lot when you start trying to represent it electronically. You have to be really explicit about what you're talking about. This is the exercise a lot of companies are going through. And things like, for example, suppose you're a retailer and you're trying to figure out 'Is Mother's Day going to be hot this year?' You look at this March versus last March. And Valentine's Day is kind of like Mother's Day, so you look at this year's Valentine's Day versus last year's Valentine's Day. And you look at how last year's Mother's Day compared with last year's Valentine's Day. Then you look at things like what happened to the overall economy in the meantime, and adjust for that. When people become more price-conscious they will start to buy, for example, cheaper things. Instead of a $30 shirt they'll buy a $25 shirt. But when they're buying for their mom or their girlfriend, they won't do that. Instead of going for, say an expensive $80 nightgown, they won't go for some cheap nightgown, but they'll go to $40 for expensive chocolates.

The transformation is different depending upon what kind of goods you're talking about and what kind of quality you're talking about. If you want to model July 4th, weather has a much bigger impact because there's lots of barbecues and outside stuff, whereas Mother's Day is weather-independent. How do you represent the data, so that you can model to make the right prediction? It's not simply taking a spreadsheet and putting in numbers. You need to define what 'expensive' means for shirts, versus what it means for gifts. And where do things cross categories?


In part one of this three-part interview, Ms. Dyson talked about her views on Web services. In part three, which will be posted soon, Ms. Dyson considers the hot-button issue of privacy and identity management in an increasingly always-on world.

+ HOME

Tell us what you think of this post using our On or Off rating system. Only your most recent vote will count.

WAY OFF								ON THE $
	[1]		[2]		[3]		[4]		[5]

Join the Discussion

NOTIFY?

I love Esther dearly, but some times her idealism gets the better of her. Specifically in her initial comment, "Suddenly people are no longer at the mercy of some incumbent local monopoly carrier, whether it's the consumer or somebody trying to reach the consumer with a third-party service.", this is pure fantasy. Just this week I wanted to get my Earthlink DSL service and my MCI local/long distance service moved to my new apartment. Well guess what? Because the lines are owned by SBC they won't authorize the necessary work since I'm not using their local, long distance or DSL service. Hence, I can't have either Earthlink nor MCI installed w/o first switching at least my local phone service to SBC, then get the Earthlink DSL installed, then try to get MCI back on w/o SBC noticing. How does VoIP help considering SBC still has to be involved? ;-) We're still hostages to the RBOCs but are under some false pretense of deregulation.

Here's a question I ask of WiFi advocates but never actually get a real answer:

What will be the services that will represent the economic ecosystem surrounding WiFi. Nope, do not trot out security and infrastructure. I mean real honest to goodness services specific to WiFi that people will pay for. Oh yah, and who gets to solve the billing issue? Yah yah, we all agree the mobile operators are SOBs but heck, at least the SOBs can send me a single bill.

The "tour d'horizon" scratches the surface of a lot of different interesting topics.
Network operators: Even with things such as VoIP and WiFi people are dependent on networks. It doesn't really matter if you are at the mercy of of some incumbend former monopoly carrier or at the mercy of a competitive voice and/or data network operator. Fact is, that all communication between network terminals (clients and servers) needs a network layer. Only private local networks without outside connection are independent of public networks (telco or ISP). But then the network would be restricted to local communications only. For unrestricted communication you will end up paying a network operator, be that a voice service provider, a data service provider or a combination of the two.
Data: One important thing that Esther points out is that teeny little stupid data things can matter a lot. The example of different languages is one that is crucial, especially in communications. On the web we can define our language preferences for content, but for voice communications there is currently no way to do it. So it is quite well possible that you are talking French and someone else is trying to speak German. Between humans, these things can be sorted out, but as soon as you deal with machines (e.g. remote voice recognition) the lack of knowing which language is used for the communication is the perfect recipe for disaster. Voice recognition is already pretty unreliable when it has to be speaker-independent. If it has to deal with another language than the one it is expecting, it is pretty much useless. There is a lot of business potential in exactly those teeny little stupid data things.
Web: One issue that has not really been covered so far is that the Web might not be the right metaphor for a lot of applications. Especially in low-bandwidth environments and when using a crippled terminal (small screen, low resolution, keypad/thumb keyboard) there are better solutions. The web is and remains primarily a publishing platform.

nice article