Internet Outlook with Richard Wiggins
|Vol. 1 No. 1||June 23, 1997|
Personalization, "Push," and the Parable of the Late Frost
By Richard Wiggins
Later that afternoon the weather was sunny and it seemed warm, so I decided to double check the weather forecast online to make sure we really faced a freeze. So I pulled up the Weather Channel site (www.weather.com) and keyed in "Lansing." To my surprise, not only did the page not say anything about a frost warning, it didn't even forecast a low temperature for that evening!
I was astonished. Here was that trusted weather information source we all watch on cable TV -- the one that warns us about everything from tornadoes to a high mold spore index -- with no information about the impending frost! So I decided to check another highly-promoted online weather source, the weather pages at www.cnn.com.
After dealing with a particularly poor application of a pick list to find my city, I still was astonished to see nothing about frost. Gee, can't gardeners and farmers rely on CNN's much-touted weather page either?
So, I tried yet another weather source: the Weather Underground. Founded at the University of Michigan under a Federal grant to provide weather information to K-12 schools, the Weather Underground is now a commercial entity, and a popular source. So I went to www.wunderground.com. Their first "forecast" screen didn't alert me to the frost, either, but rather concentrated on current conditions.
Hmm. I scrolled on down the page, and finally I saw a notice that frost was on the way. Finally, one of the major weather sites on the Web had come through, and had confirmed what I'd heard on the radio that noon!
I also checked out an ancient weather source that relies on the same National Weather Service feed -- a Gopher-based, text-only service that we set up at my Day Job at Michigan State way back in 1992. It included the frost warning as expected. It was text-only, it wasn't flashy, and it was -- horrors! -- delivered via outmoded technology. But it told me the relevant news up front.
This got me to thinking about the state of the Internet. For so many years we've been hearing about ways in which the Net will learn about us and our individual needs, and will respond to us with the information that's important when we need it. The Net will deliver "hyperlocal" content on a personalized basis. Where's my hyperlocal weather report?
We first heard this promise under the rubric of "intelligent agents." You'd interact with your agent, which would probably be a piece of software installed on your PC, and tell it the kinds of information you're interested in. It'd periodically poll pertinent information sources on the Net, and alert you to new information of interest -- a dramatic change in a stock price, news about a company you are interested in, the announcement of the Pentium II chip -- or an impending frost in your area.
There have been numerous attempts to build these sorts of agents. In practice, what we have so far is pretty limited. An example is Smart Bookmarks. This tool is supposed to help you monitor Web sites of interest. For careful tracking of sites that are directly on point, the tool can be useful. But if you try to use it to monitor all sites that have information of some interest to you, you'll probably find that it alerts you to cosmetic changes, or changes to irrelevant content. The same sort of thing can happen with Web-server-side services such as www.netmind.com.
There are promising developments in agent technology, such as the work at Firefly, but all of these tools face daunting challenges. Most of the information published on the Net is totally unstructured and chaotic. Although HTML comes from a rich background of the SGML community, in practice very little use is made of META tags and other mechanisms for providing useful structure for a document. An agent can only be so wise in parsing a document and trying to understand questions like "Is this document what Rich was looking for?" or "This document has changed since the last time Rich saw it. Is the change significant?"
While we can expect agents to become smarter, it'll may be a long time before we can rely on a single agent to tap us on the shoulder and tell us what we need to know when we need to know it. In the meantime, individual sources of information will continue to offer filtering and personalization services. But these have drawbacks:
What About Push?
But what about "push" technology? Won't that solve the problem? In fact, the Weather Channel page does update its page every half hour or so. (Technically, they use the "Refresh" tag, so it's actually client pull, not server push.) And you can personalize your choice of what city they should monitor; the choice is stored in a cookie. If you wanted to dedicate a PC to be your local weather monitor, the Channel will oblige.
But, if you really want a "push" version of the Weather Channel, your best bet is to turn on your TV. The satellite loops are much smoother on the TV. Think about it: one advantage of the Web version of the Weather Channel is its interactivity: you don't have to wait through 30 minutes of mold spore index to find out the wind conditions in your city.
Anyhow, what comes out from the web feed is only as complete as what they've chosen to give you. The "push" version of the data is the same page that didn't have the frost alert we needed. All "push" does is give you an automatic refresh of the page. Nice, but why so much hoopla?
So what will I rely on for local weather information? If I want pretty pictures and satellite loops, I'll still use the Weather Channel. If I'm really concerned about local conditions, I'll listen to the noon news on the radio. If that's not convenient, I'll often call the local National Weather Service telephone line. It's a free call, and they do include frost warnings. Or, I'll rely on that old Internet Gopher feed.
Comments are welcome
Produced by Richard Wiggins and
Created: June 20, 1997
Revised: June 23, 1997