Companies make life on the Net unnecessarily difficult by failing to support established standards and attempting to set their own.

by Joel Snyder

I'm disappointed in Netscape and Microsoft. Why? Because they pay no attention to Internet standards. Their products aren't compatible and they don't seem to care. I just bought a new Intel-based PC and wanted to load Windows 95 on it. Naturally, an important part of this project was going to be making the PC an Internet workstation with all the right toys. Getting Windows 95 running was astonishingly easy. This was a prepackaged system from Digital Equipment Corp., a Pentium Celebris. I picked it because it reminded me of a Macintosh: All the bits I needed were integrated into the motherboard--video, audio, serial and parallel ports, IDE and SCSI disk controllers, and Ethernet. No running around adding bits and pieces to try to make a complete system. This system replaced an older PC that was running vintage 1995 software, including Netscape Navigator 1.0 and Qualcomm's Eudora e-mail package.

I was using both Netscape and Eudora because that was the package Netscape sold as the Personal Edition in 1995: everything you needed to cruise the Internet. For 1997, I went to the bookstore and picked up Netscape's new Navigator Personal Edition 1.2.

The first thing I noticed was that my beloved Eudora had disappeared. In its place, Netscape had left its own mail package. Now, this is not a software review, so I'll refrain from saying that Netscape's mail capabilities are about 10 steps down from what I can do with Eudora.

What I want to talk about is how the Netscape mail program ignores current Internet standards. It's not worth going into the technical details here, but the end result is that Netscape's software engineers slapped support for a preliminary standard (RFC 1806) into their product and left out the existing, well-supported MIME standard. This means that certain types of attachments lose information when read by a non-Netscape mailer. This is unacceptable. But let me digress for a moment so you can understand how Internet standards are set.

Standards Procedure
Internet standards are known as RFCs (requests for comments). Each RFC has a number assigned to it, and there are just over 2,000 RFCs today. In the world of the Internet, the standards are agreed upon by committees after long, educated discussions on the best way to include new technologies in the Internet in a standardized way. Most RFCs come out of committees that are part of the Internet Engineering Task Force (IETF). The committees meet occasionally in person but often are conducted via e-mail.

But not all Internet standards come out of the IETF. For example, modems are standardized by the International Telecommunications Union (ITU; http://www.itu.ch), which is part of the United Nations. If you have a standard 28.8-Kbps modem, it conforms to ITU Recommendation V.34 (and probably V.42, V.42bis, and others). That's a non-IETF standard in widespread use.

One complaint people make about non-IETF standards is that they can be difficult to acquire, which discourages people from implementing them properly. (To obtain an ITU standard, for example, you must call the ITU bookstore in Geneva and pay on the order of $20 to $200 each, or contact one of the few U.S. companies that have them available, usually at slightly higher prices.) The IETF's RFCs are different: They're free and you can download them from the Internet. To see for yourself, connect to ftp://ds.internic.net/rfcs and download the rfc-index.txt file, which lists all RFCs and tells you how to determine their status. Thus, there's no excuse for anyone writing software for Internet use that does not conform to current standards.

This is the basis of my annoyance with Netscape Navigator's new mail functionality. There's no reason for the company to have built this kind of incompatibility into its product. Doing so is either sloppy or anticompetitive. I tend to believe the former, although there are some paranoid folks out there that prefer the latter interpretation.

Fine. If Netscape's not going to play by the rules, there's another game in town: Microsoft. The company was kind enough to build its browser into my Windows 95 system, so all I had to do was click to explore the Internet.

Right. All I had to do was spend 30 minutes with Internet Explorer 3.0 to find out that the folks in Redmond, Wash., are even more careless than their counterparts in Mountain View, Calif.

Internet Explorer won't talk to my Usenet News server because the server doesn't support the "XOVER" command. Is this part of the standard for news servers? No, it's not. Is there any reason why Internet Explorer couldn't work without this command? Of course not. There are only six other newsreaders on my workstation, and all of them seem to survive quite well without it. But Explorer doesn't even give me the choice to work with my poor old news server. You don't support XOVER? Tough luck.

Microsoft's implementation of FTP also is nonstandard. It doesn't work with many perfectly good FTP servers. Why not? Because the Microsoft engineers weren't paying attention and didn't read the standards. I get the feeling that Microsoft's idea of testing is to connect one PC to one server and if it doesn't show a General Protection Fault, to ship the product.

Compared to the criminally poor implementation in Internet Explorer, Netscape's error is like jaywalking. I could almost forgive them, if it weren't for HTML (Hypertext Markup Language), the language in which Web pages are written. HTML 2.0 is a standard: RFC 1866. Dated November 1995, RFC 1866 describes how HTML should be written, what's legal, and what's not. If it were up to me, I'd say that writing HTML that doesn't conform to RFC 1866 should get your pages relegated to the trash bin. (The proposed specs for HTML 3.2, the newest generation of the code, can be seen at http://www.w3.org/pub/WWW/TR/PR-html32-961105.)

But fortunately for all those folks who use nonstandard multimedia animated backgrounds with nonstandard in-place frames and 12 different nonstandard fonts, it's not up to me. The reason people do such zany things with their pages is that the browser manufacturers--in other words, Netscape and Microsoft--let them.

The Pace of Change
The folks at Netscape and Microsoft are completely dissatisfied with the speed at which standards are made. They couldn't possibly wait for the market to settle down, for products to become stable, and for people to have experience with a feature before it gets incorporated into 10 million Web pages. They have to have change, and they have to have it on a weekly basis. If you're a Web developer, I'm sure it puts fun and excitement in your life.

On the other hand, if you're someone who doesn't think that changing software as often as you change sheets is a good idea, these guys are putting you through the wringer and hanging you out to dry.

Another company going down this same path is US Robotics. Its x2 technology is a new way of using phone lines to increase bandwidth between ISPs and their customers. (x2 is only useful when one of the parties has a fully digital phone line, which is pretty unusual.)

Unfortunately, x2 is an outgrowth of US Robotics' research labs and has nothing to do with national or international standards. It's nonproprietary, and US Robotics has said it is submitting the protocol to the ITU. But there is no guarantee the ITU will accept the proposal, or that it won't change it. This is a hardware equivalent of what Microsoft and Netscape have done: going ahead with a standard they like without getting external validation. And it's a lot harder to change modems than browsers.

People buying into it may not be aware of the horrible days of V.32 terbo and V.fast, the prestandard proposals for faster-than-14.4-Kbps modems. Lots of companies sold nonstandard modems in the hopes that the final standard (V.34) would match their choice. It didn't, but many people had bought the new modems without understanding that they were betting on a long shot.

It all comes down to standards. In the standards business, there's a curve that looks like a camel with two humps, with each hump representing a phase of a new technology. The first hump is the innovation period for a technology--lots of activity, but too early to set standards because things are changing too fast. The second hump is the adoption phase--lots of sales and too late to set standards.

The right time to tackle standards is between the two humps: after the innovation but before everyone buys it. But as far as HTML goes, that model is dead. We're all reluctant beta testers for the Internet because there is no low point between the two humps. There is no good time to standardize, and the main players aren't willing to wait before releasing a new version of their software.

You could argue that the reason the Internet is growing so fast is that its fever pitch makes it exciting to join, and all this builds on itself. You could also argue that the reason we're getting so many valuable new services is that businesses look at the size of the market and want to hop on as soon as possible. But that argument doesn't hold water. I'm happy to see the Internet grow fast, but it could grow just as well at a slower rate.

My gardener told me it's dangerous to feed trees too much fertilizer when they're young because they grow too fast. An overfertilized tree becomes vulnerable to disease and weather. A tree needs to grow at a slow and steady pace to be healthy.

The Internet works the same way. The rapid pace at which Netscape and Microsoft are introducing buggy, nonstandard software is hurting us all. Joel Snyder is a senior partner at Opus One, an Internet consulting company in Tucson, Ariz.

Reprinted from Internet World magazine Vol. 8 No. 2 (c) 1997 Mecklermedia Corporation. All rights reserved.


Internet.com home