Lies, Damn Lies, and Statistics

Megabytes, megahertz, market share–you can't have a conversation about computers without trotting out numbers. Yet the sloppiness of today's high-tech numberology is appalling. Let's try this example: how many Americans are online? Research firm IDC nails it squarely at 38 million; Mediaquest says 53 million; Intelliquest says 73 million people are online, almost double IDC's estimate.

Now let's say you're writing a Wall Street Journal article about how TV is doomed because of the Internet's rise. Which number would you pick?

See the problem? We worship statistics, measurements, and specs, but the numbers are often figments of somebody's agenda. Here are a few we could start correcting right now:

56K Modems

How is that label misleading? First, 56K should be 56 k , referring to kilo bits per second, only an eighth as much as kilo bytes , the K that we're all used to. Second, the FCC sets a 53-Kbps maximum speed, because stronger signals cause phone-line interference. Third, even 53 Kbps is theoretical; most people get 46 Kbps on a good day. And fourth, that speed is only for receiving data. All 56-Kbps modems send data at only 33 Kbps. In other words, 56K is the most fraudulent spec in the industry. Ralph Nader, where are you?

List Prices

What a wacky system: each company picks a pie-in-the-sky price for each product. That way, every catalog and store can sell it for half as much, promising us a fantastic discount.

Trouble is, nobody sells software for list. Digital Origin's EditDV goes for $899–list. But even the company's own Web site sells it at $499. Why don't we end the charade and talk about the actual prices? ( Macworld will soon do its part by publishing averaged prices from several national stores rather than list prices.)

Suggested Memory Size

When I was having trouble with Internet Explorer, a Microsoft tech rep's first suggestion was to change the suggested RAM allotment from 4 megs to 8.

In other words, software companies pressure their programmers to set memory sizes too low–for marketing purposes. We pay the price in stability.

Market Share

Macs last a lot longer than Windows PCs. Millions of people still use their nine-year-old Mac Classics and IIsi's–but how many people still use that era's 386 (Windows 2.0) machines?

So, true enough, current Mac sales aren't a big slice of the pie–but the number we hear about, sales market share, is irrelevant (especially when the calculations are as goofy as research firm PC Data's: it tallies each color of iMac as a different model, making it impossible for the iMac to maintain its best-seller status). What really counts–when calculating the potential market for software and add-ons, for example–is how many Macs are in use. That number is much higher: 16 percent, by the estimate of the Software Publishers Association (SPA). But do we ever hear that statistic? Noooo!

Software Market Share

"Windows applications accounted for 81 percent of the application software sold," says the most recent SPA report; "Macintosh sales represented just 11 percent." Scary, dude.

But the SPA's estimates are nearly useless, thanks to one little fluke: an increasing number of companies include Mac and Windows versions on the same CD. (According to the market-research service SofTrends' latest numbers, up to a third of software best-sellers are such hybrid CDs.) When you buy a hybrid CD, does the CompUSA checkout clerk ask which kind of computer you have? Of course not. All such sales are tallied as Windows sales.

Megahertz

As far as the public knows, more megahertz is better. But this year's hot new blue 300MHz Power Mac G3 is actually slower than last year's beige 300MHz Power Mac G3 (because of the effect of other components)–and that machine is faster than a 300MHz Pentium (because MHz aren't comparable among different chip families). The world has yet to invent a consistent, overall speed measurement for computers.

Version Numbers

Software version numbers are pure marketing concoctions. The upgrade from System 7 to System 7.5, for example, was a baby step compared with the leap from 7.5 to 8.0. If anything, it's the tiny increases that really count, because that's when a program goes from "4.0" buggy to "4.0.1" usable.

Of course, then there's the numbering leap from Word 6 to Word 98. That one seems about right.

http://www.davidpogue.comMacworld Mac Secrets

June 1999 page: 180

1 2 Page 1
Page 1 of 2
Shop Tech Products at Amazon