By Joe Barr
Originally published in January, 1996

tres duh press 1. That portion of the computer industry which blends press releases, rumors, and misinformation and sells the resulting concoction as news.

In spite of legal threats from David Sarna, requests from IBM that I "lay off" a certain journalist, and the fact that the issue is a week late, it's finally here: the long-awaited and much anticipated "first annual examination of the computer industry's trade press" issue. I apologize for the tardiness. I would love to blame it on the holiday schedule, but in fact it's a result of my own poor planning.

What a year it's been. The trade press itself has provided some of the bigger stories. Ziff-Davis (publishers of most of the best known periodicals) changed hands and as a result has become even more intertwined in the industry itself. SoftBank, the Japanese trade- show conglomerate which produces COMDEX and similar shows around the globe, is the proud new owner. Other news from within the press was very unflattering.

One of the more respected news weeklies, ComputerWorld, was greatly embarrassed by inaccuracies in a front-page comparison of Win95 and OS/2 Warp. To their credit, they ran a front-page apology and have promised to rerun the story after getting their facts straight.

The most bizarre story involving a publication, however, seems to have been swept under the rug. Did InfoWorld leak the results of their benchmarks comparing Windows NT, Win95, and OS/2 to Microsoft (or their PR firm) hoping to snag some additional advertising bucks? That's the suggestion that was sent by E-mail to Will Zachmann two weeks prior to the publication of the results. Rick Segal of Microsoft provided the "leaked" order of finish and a suggestion as to why. InfoWorld conducted an investigation flatly denied any such leak occurred. They theorize that Segal was just lucky in guessing the order of finish of the their results.

For most of the year, however, there was only one story: Windows 95. The hype began early and still continues, although its less than spectacular debut in the marketplace has dimmed the lights slightly. Another popular theme, not entirely a separate one, and certainly not a new one since Bill Gates opined it several years ago, is that OS/2 is dead. If you got the message last January that Windows 95 was going to set the world afire and that IBM was dropping Warp, you could have skipped 90 percent of the rest of the stories for the entire year. The remaining 10 percent was focused on the Internet or on the Donald Trump of the 90's: Bill Gates his own self.

The year began with then PC Magazine editor Robin Raskin predicting that OS/2 would not survive the year. This is an annual event for PC Magazine, one of the most obvious of that fawning, servile, ethics-free trade press willing to do anything to ingratiate themselves with Mister Bill, and is usually not worthy of special mention. Bill Howard has just served his tour as flag bearer and done the honors for 1996. Others carried this tattered Gates banner prior to Raskin and it's a safe bet the drumbeat will continue for years.

What made Raskin's comments remarkable was the contradiction by reality and by the publication's efforts to minimize her embarrassment for poor timing in its subsequent issues. It seems Warp was the top selling software package in the world when she declared it dead. PC Magazine promptly removed their famous Top Ten list from the two issues (for the first time ever) immediately following Robin's attempt to put more IBM employees in the unemployment line. Warp would have occupied the top spot in both the missing lists.

A lame claim was later made that the list was removed to make room for coverage of the Pentium bug story, but that was a hot story in 1994, not 1995. I think it's obvious it was pulled to keep from making Raskin look even more like a boob.

For sheer bad luck, probably no journalist had worse than Lori Grunin. Writing in a review of SoftRAM95 for Windows Sources, she stated that "this one-trick pony works as advertised." In case you haven't read the rest of the story, it has been shown that this top-selling Win95 application does not work, period.

It has been reported that Microsoft spent two or three hundred million dollars to super-hype the launch of Windows 95. But if they had to pay for all the free advertising lavished on this latest version of Windows by the doting dolts in the cheerleader press, the bill could easily have exceeded a billion dollars.

Hype is a normal part of retail marketing, even when taken to excess as in the case of the August launch. It alone is not a damning thing for the press to have engaged in. The problem is that both the major and the minor premise they tried to drum into us were wrong. Here is all the evidence required to convict on a charge of felony incompetence: OS/2 continued to grow market share in 1995 and the most spectacular aspect of the Win95 launch is the fact that Dataquest (and other industry prognosticators) have had to nearly halve their original projections for first year sales.

Different folks read trade-related stories and publications for different reasons. Some simply want to read the ads and find the best deal on a product they've already settled on. Others are trying to find which product it is they should buy, or which they should avoid. According to DMM (the Dilbert Management Model), one of the most popular methods of IS management is MRM. Management by Reading Magazines (and then taking the pundits views as their own).

My point is simply this: stories in print have influence on buying and management decisions. This influence is the power of the trade press. When a Robin Raskin or a Bill Howard makes predictions about the future of a product, they are helping to shape its destiny. Why they choose to become players instead of observers and reporters is a moot point. The fact is that they and others do so as a matter of routine. The computer trade press as a whole has less objectivity and fewer ethics than hour-long advertorial on late-night TV. Less because at least the TV shows are required to identify themselves as being ads.

What you didn't read about is just as revealing as what you did. Several stories which would seem to warrant major coverage but which were either ignored or glossed over include: record returns of Win95 by retailers, outright rejection of the product by the United States Navy and other security-conscious government entities around the world, and the fact that Win95 runs slower on Intel's Pentium Pro than on its predecessor, the Pentium.

IBM reportedly has tried to get wider coverage on how OS/2 soars on the P6 (the new Pentium Pro), but with no success. Who can blame them for wanting to see it in print? OS/2 Warp applications have been shown to run between 30 and 121 percent faster on the new P6, much faster than any gains shown on Windows NT. The problem is that such reports would be a real embarrassment for the Wintel establishment. It seems that key components of the "all new-all 32-bit" Windows 95 are actually 16-bit. The Pentium Pro is tuned for 32-bit operations and actually runs 16-bit code slower than the P5 Pentium.

But Intel, Microsoft, and Ziff-Davis seem to have linked arms in their attempt to ignore/dampen the ashes of this story. Intel reps touring the country to talk about their new P6 to user groups have told at least one such group in Dallas not to move to the P6 unless they were running Windows NT. According to the Intel rep doing the presentation, Win95 and OS/2 have too much 16-bit code and actually run slower on the P6 than they would on a P5. While this is definitely the case with Win95, OS/2 screams on the "tuned for 32-bit" P6.

Microsoft employees and advocates online produce data about the percentages of 16-bit code in both Win95 and OS/2, hoping to blur the lines which distinguish the two products. But the percentages they present are based on lines of code in the OS, and are meaningless in terms of performance on the P6 CPU. The important percentage is that describing the amount of time each spends executing 16-bit code versus 32-bit code. The P6 doesn't lie, and Warp smokes Win95 in comparison of 32-bit apps. That's why you're having trouble finding the benchmarks in print. That's why IBM is running into difficulty getting the story told. After all, these are the same good folks who spent the first eight months of the year telling us how wonderful Win95 was going to be, they don't want to admit that the yellow stuff on their cheeks is egg yolk.

The best effort to cover up and distort the story comes from PC Magazine. A story uploaded to the Ziff-Davis Editorial forum on CompuServe is one of the most blatantly dishonest and misleading I've ever seen. Matching the accompanying spreadsheet with the text is mind boggling once you realize what was done. If you harbor even the slightest doubt about the amount of bias in this publication, I urge you to download those files for yourself and take a look.

The article describes testing done on a 150-mhz P6 and a 133-mhz P5. Nick Stam writes "testing shows a 150-MHz P6 system to be noticeably faster than a 133-MHz Pentium PC when running 32-bit applications under Microsoft Windows NT."

Hello, Nick! Good grief, man, it should also be noticeably faster when run on a 150-mhz P5. It has a nearly ten percent faster clock speed which makes everything the CPU does faster. Where is the adjustment for the difference in clock speed? This is like staging a drag race between a Corvette and a Geo and pretending the difference in performance is because of the paint job instead of the horsepower.

But Nick's not finished, he goes to claim "The P6 machine generates mixed results with 32-bit applications under Windows 95, and clearly lackluster performance for 16-bit applications running under Windows for Workgroups 3.11 and OS/2 Warp Connect."

Whoa, dude! Fifteen yards for intentionally missing the point! What Nick is hiding here is that "mixed results" means things are running slower on the P6 under Win95 than on a P5, even when running 32-bit applications. For icing on the cake, Nick added 32 meg of RAM to the NT test machine and then noted that gave NT a nice boost in performance on the P6.

What did IBM's benchmarks show for Colorworks/2 on a P6? A whopping 121 percent increase over a Pentium. Even with the RAM doubled for NT it didn't approach that sort of performance benefit. What mention of this does PC Magazine make? None.

What mention of the other OS/2 applications tested, none of which showed less than a 30 percent boost on a P6? None. What does PC Magazine say? "We did not test with 32-bit OS/2 applications because few mainstream titles are available." Also, "We believe the large amount of 16-bit code still present in OS/2 contributes to the reduced performance observed..."

When you look at the spreadsheet accompanying the article and see that the only head-to-head comparison made shows that OS/2 is penalized less when running a 16-bit application than Win95 is when running the same application, you no longer have to wonder about the question or degree of bias in the trade press.

Time now to move from the general to the specific, from the industry as background to the recipients of the various Best and Worst categories. Although I solicited votes in both categories, I treated them as suggestions only. The final choices in all categories are mine alone. A little background on the criteria I used in selecting them.

I included writers/editors/publications in the general, consumer, and professional trade press. Some of these appear both online and in print, but unless they appear in print I did not consider them for the awards. This put some very fine people out of the running: Charles Cooper of PC Week Online comes to mind.

I did not include publications with a specialized focus, whether for hardware, software, or application. It is worth noting, however, that publications in this category are generally more honest and forthright in their stories if for no other reason than they tell you where they are coming from.

I based my selections on knowledge of the subject, objectivity, and ability to communicate. It takes all three to stand out as one of the best, but the lack of any one of them can flag you as one of the worst. In other words, you can be a great writer and very knowledgeable, but if your objectivity is hindered by anything, whether personal preference or self-interest, you're outta here.

Given my generally low opinion of the trade press, you might think it an easy process to select the Worst examples in each category. Actually, it makes it more difficult because there are so many worthy contenders. Obviously, all these awards are based on my opinions and biases. How happy or unhappy you are with the trade press depends to a great extent on how well your specific needs are being met.

If you are a Windows 3.x user with a 4 to 8 meg machine, you should be feeling pretty good about it. You will be catered to, complimented, and made to feel good about being part of the great horde. You will also be carefully coaxed along the upgrade trail to Windows 95 or NT, and a lot of effort will be spent to make sure you don't fall off into the Mac or OS/2 camps along the way. You will also think that all the fuss about bias, dishonesty and lack of ethics in the press is way overstated.

If you are a DOS, Linux, Mac, or OS/2 user your needs will be mostly ignored. You will be ridiculed for not being "mainstream", and your product of choice often completely misrepresented by unsighted "journalists." Except for those publications aimed at your specific segment of the market, you will see the trade press as misinformed at best and as bootlicking whores of Redmond at worst.

As James Fallows, the noted author, pointed out to Rick Ayres, now the Technical Editor for PC Magazine , a couple of years ago, bias can manifest itself in ways that are not obvious to those in the mainstream. Coverage of the computer industry by the mainstream news media is spotty at best. Bloomberg Business News and the New York Times syndicate seem to be the most active of the widely carried, print variety. Bloomberg is better than the NYT, not so much because they are good but because the NYT is almost always slanted towards Redmond. Some people think that's because they distribute the Bill Gates column worldwide. Others point out that IBM has representation on their board of directors. I don't know why it is, but I do see a lot of spin on many of their stories.

The popular, consumer-oriented publications are usually better in terms of technical validity, but not by much. Spin is also the name of the game in this genre. The old established industry pundits (Dvorak, Seymour, Pournelle) have become icons rather than oracles. It's definitely a case of personalities before principles. Byte, PC World, and the long list of Ziff-Davis publications are the familiar titles. My pick of the very best from this group was ruled technically ineligible because of residency requirements, but if you want to see my idea of what all these rags should be emulating, point your favorite HTML 3.0 Web browser at PC Magazine UK and check out Guy Kewney and the rest of the cast. If your reaction is anything like mine, you'll find it awfully hard to believe that PC Magazine and PC Magazine UK are from the same organization. Oh, they do vendor bashing all right, but they do it based on products and service rather than on vendor name.

The best source of news and industry insights comes from the weeklies and monthlies aimed at the computer professional rather than at Joe Consumer. ComputerWorld, InfoWorld, PC Week, and others have a much better ratio of substance to hype, they have more knowledgeable writers who are conversant enough with the industry to describe its activities with some reliability.

One area they seem to fall short in, however, is in the replication of vendor press releases as news. Whenever you read the same person's byline a dozen or more times a week, it's a good sign that they are "reporters" who are relying on the truth and accuracy of the vendors themselves in their stories. In this industry, like most others, that is a very dangerous assumption.