A Contrary View: Infrastructure Will Slow Internet Speed Improvements

By: Steve Outing

Robert Crandall thinks he's got a better idea than the thousands of publishers and software developers investing their electronic future on the Internet. It's not that the Internet isn't a great and wonderful technological development, it's just that the promised bandwidth improvements being touted as a solution to slow-loading World Wide Web pages by the cable and telecommunications companies are going to be a lot more difficult to achieve -- and take much longer to execute and cost more -- than we're being led to believe, in Crandall's view.

Crandall is president of PEN Industries Inc. in Hawaii, a research and development firm that has been developing what it calls the "Personal Electronic Newspaper," a hand-held, portable display unit "that receives data 120 faster than the fastest modem" and does not rely on the Internet. He faxed me an artist's sketch of the device, and it looks like the simple portable digital display unit envisioned by Roger Fidler, who for years has been touting the concept of an enhanced electronic version of the daily newspaper.

The vision is "to equip every household with the paper-less media at a nominal cost, saving trees and still keep the newspaper agencies in control of free press, maintaining their revenues and answering questions concerning the feared, yet inevitable natural evolution from the printing press to electronic media," says Crandall. The technology will allow digital delivery of "any product that a publisher desires to design." Consumers will do a broadband download via cable TV lines to get data into the unit, and PEN Industries will manage the data (but not the content) for publishers.

Crandall estimates he's six months away from production of the notebook-size devices, which weigh about 5 pounds and are 2 inches thick. (PEN will have a Web site operational in a couple weeks, if you want to learn more about what the company is doing.)

Stay off the Internet

While the PEN concept is certainly interesting, I was most struck by Crandall's comments about why he's ignoring the Internet in development of this product. "I have purposely stayed off of the Internet and while this may seem like a radical or even blasphemous stance, I did so for very good reason," he says.

Crandall has spent "the last 20 years of my life living around and in the telecommunications industry," including working as a cable splicer for 10 years earlier in his career. "I have a different viewpoint that not even the telco executives themselves understand," he says. "I know how things actually work out in the field. Many 'experts' have never even seen the inside of a splice pedestal. All of their hype is based on theoretical perfect-world scenarios." But guess what, he says, the real telecom world is far from perfect.

Crandall explains that in the U.S., the central trunking cables and long-distance cables are now fiber optic, capable of handling 2.2 gigabits per second. The newest digital switches, which point information into the direction that causes it to reach its intended recipient, are much slower, but still fast enough (655 megabits per second for the new ATM switches) to move data around the country pretty quickly.

The problems begin when data gets past the nearest switching center and heads toward residential and business subscribers, since it must be carried over conventional copper twisted pair wiring. Since it's too costly to convert all of this wiring to connect directly to the fast network, the existing twisted pair network must be worked over to accommodate fast data transmission.

"Here's the part that the telcos don't want you to know," says Crandall. "In order for current technology to utilize copper twisted pair for data, splicers must physically go into every splice case in that run, and cut out all the branches of the tree. (When built, cables are branched off to go down streets, just like a tree, but the trunk continues down the main street, and the same pairs are branched off again.) All of these must be removed for a straight shot. The higher the transfer rate, the more perfect these pairs must be. You have to have one pair for sending the signal, and another one for the return signal."

All the bad or corroded splice connections in the twisted pair wiring must be replaced or the connection will be too noisy for data transmission. This is very costly for the telcos, which is why ISDN lines (128 Kbps) have a high price; the consumer is also paying for the upgrade of the pairs to make the connection between his home or office and the closest switching station clean enough for data transfer. (Analog voice calls are much more forgiving of line noise.)

The telcos tout ADSL as their answer to fast Internet access using existing twisted pair copper wiring. But "ADSL is a myth," says Crandall. "You've heard the saying, 'a chain is only as strong as its weakest link.' Well, no matter how fancy and fast a 'box' is from a manufacturer, it is still only as fast as the cable is capable of. That cable is the main problem. When a splice case or pedestal is run over by a car or (experiences) some other damage, it is left for a future date, if it doesn't put anyone out of service. The splice connections begin to weather and corrode. They will most of the time pass voice OK, but data, never in a million years.

"When a troubleshooter goes out into the field and finds a bad pair, he will either switch to a vacant pair, or splice in a section from an unused group, if no spares are available," he says. "Line assignors rarely make note of these changes, and so have no idea what the state of the plant is in. This practice has been going on for decades."

Have you ever seen one of those green telephone company canisters sitting by the road, perhaps with wires hanging out of it? That piece of equipment is exposed to the elements, and often can become corroded due to rain or snow. "Have you ever noticed that when it rains, your telephone gets noiser? The reason for this is that your pair that you talk on is getting wet. Somewhere it has been exposed to the elements. These lines will NOT pass data," Crandall says.

What about cable modems?

Companies like @Home, of course, are touting cable modems as the way to increase bandwidth to the consumer. Since the telcos' ADSL technology is the cable industry's primary looming competitor, cable executives might be heartened by Crandall's comments. But he has some pessimistic words for the cable industry, as well, as it struggles to provide Internet access by adapting its existing plants.

Cable modems promise 10 megabit-per-second transfer speeds, which sounds great. But there are two reasons why cable modems aren't in general use yet, Crandall says.

1) "In order for cable modems to function, the bandwidth of the system must be increased. When multiple users try and use these modems, the size of the pipe gets smaller and generally slows things down again."

2) "The forward side of the network works just fine, but the reverse side, which is the side that you would need to request any information from a server, doesn't work. The problem is that the reverse side is subject to what is called 'ingress' -- the admission of outside radio frequency into the reverse side of the amplifiers, which makes the signal-to-noise ratio too high," Crandall says. Too much noise will cause the digital signal to drop completely. "This is the nationwide problem that cable companies are having with making cable modems an accepted reality."

The problem is caused primarily, he believes, by bad splices out in the field, which will have to be replaced in order to have a cable network capable of handling data cleanly without data-disrupting noise. Bad splices are the entry point for noise. Crandall says he helped splice and troubleshoot a fiber/coax hybrid cable network on the island of Oahu. Many installers made mistakes in splicing, out of either carelessness or lack of adequate training. Also, "management made us splice in the rain. There are so many wet connections and amplifiers that are full of water in this system, that if (company executives) knew about it they would have a heart attack," Crandall says. "None of these components will last very long."

If Crandall is correct in his assessment of cable and telco efforts at providing fast Internet access via cable modems or ADSL over twisted pair, Internet users and publishers may need to be patient while waiting for more bandwidth. To be sure, telco and cable executives will paint a rosier picture. Given the delays in ventures like @Home in meeting self-imposed deadlines for delivery of fast consumer data services, I'm inclined to believe that Crandall knows what he's talking about.

Contact: Robert Crandall, rdsee@aloha.net

76% of PCs linked to Web by year 2000

Market research firm International Data Corp. says that within four years, 76% of installed PCs in the U.S. (and more than half worldwide) will be on the World Wide Web. In a report released earlier this week, IDC also estimates that by the end of this year, there will be 30 million worldwide connections to the Web. This is another indication of why publishers should continue to invest in Internet publishing, despite lack of immediate pay-off early in the game.

Steve

CENTER>Previous day's column | Next day's column | Archive of columns
Presented 3 days a week by Steve Outing, Planetary News LLC.
Made possible by Editor & Publisher magazine.
Got a tip? Let me know about it

If you have a newsworthy item about the newspaper new media business, please send me a note.

This column is written by Steve Outing exclusively for Editor & Publisher Interactive. Tips, letters and feedback can be sent to Steve at steve@planetarynews.com

The views expressed in the above column do not necessarily represent the views of the Editor & Publisher company.