WEB EXCLUSIVE: The Road to Windows

Chaos Manor has a Web site at last, built with Windows 95, no less. Read all about it in "A Web Site for Chaos Manor" in the July issue of BYTE. The exercise has put Jerry in a reflective mood. Why Windows? Read on.

Jerry Pournelle

I've finally done it: I've put Chaos Manor on the Web. I settled on Windows 95 (Win 95) and the version of Word in Office 97 to construct Web sites for myself and Mrs. Pournelle. You can read about it -- and the difficulties I had -- in the regular column.

One of my colleagues asked why I didn't just use a Mac to build the Web pages in the first place. There are several possible answers to that. In my case, however, it's simple: I'll have to do the site maintenance and adjustment, and while we have several Macs, one very high-end indeed, most of my work is done with Windows machines. I'm simply more comfortable using Windows and Windows applications, particularly for something as unfamiliar as Web-site maintenance.

That led to a discussion of why I went down the Wintel path in the first place. It's a long story.

My first machine was Ezekial (I know, I know, I spelled it wrong, but that was his name), Zeke for short, a Z80 that ran a disk-file program called F-DOS or some such. I quickly converted to CP/M, which soon became the standard microcomputer OS; I'll preen and say that several writers have said that was largely due to my influence. Whatever the truth of that, CP/M dominated until IBM in troduced the PC.

In those days, there were a number of small computers, but the major competition was between S-100-bus Z80 systems running CP/M and the Apple II. The Apple machines were hands-down better at graphics than any CP/M system. They had color. They had better screen resolutions. What they didn't have was a decent character-based text editor, and thus were nearly useless for word processing. I recall when Joe Haldeman bought an Apple II and I found to my horror that it used a single font, I believe all uppercase letters, and changed the color of letters that were supposed to be capitalized.

Meanwhile, CP/M systems had VDM boards of 16 rows by 64 columns each. The font was 9 pixels wide by 11 high, meaning that descenders (e.g., the letters g and q) looked good on-screen. VDM boards were fast. A double-spaced typewritten manuscript page was typically 25 60-column lines, so it wasn't at all hard to get used to writing with a computer, and I did. So far as we know, Zeke, now on display in t he Smithsonian in Washington, was the first computer anyone ever used to write an entire novel.

The Apple II had a number of interesting features, but it never was very good as a writing machine. Meanwhile, IBM introduced the PC. CP/M went away to emerge again as DR DOS (and that's a long story worth telling one day; CP/M very nearly became the standard IBM DOS, which would have left Microsoft as a small software company specializing in BASIC). Word processing programs like WordStar migrated from CP/M to DOS. Tony Pietsch wrote a character-based program called WRITE. It was so good that Larry Niven and I used it until Symantec brought out Q&A, a word processor and database suite containing what in my judgment is still the best character-based word processor ever written.

Then Apple brought out the Mac. Unfortunately, it was preceded by the Lisa, which may have been the slowest word processing machine ever sold. The Lisa used a bit-mapped screen, and a PageDown could take as long as 40 seconds. Savi ng files took minutes. When the Mac came out, it was faster than the Lisa, but not by much. Moreover, Steve Jobs announced that he was adamant: the 128-KB Mac was a finished product, it would never have a hard drive, and it would never need more memory. It incorporated a text-editing program that one could use to write books -- a number of people did -- but it was a pretty painful process compared to what Niven and I were doing with our CP/M machines. (I didn't retire Zeke and convert from S-100 to a PC until well after the AT was released.)

Over the years the Mac improved, but so did the PC; and if there was ever a time when a Mac was enough better than a PC for writing books, I must have missed it. The Mac certainly was better at WYSIWYG.

WYSIWYG was once one of the great controversies in the world of small computers, but I suspect that some of you have no idea what I'm talking about. WYSIWYG means what you see is what you get (i.e., what you see on-screen is what you'll see when you print). Thi s doesn't sound like a big deal now, but it was in those days. A character-based editor, like Q&A Write, used a single-size fixed-space font on-screen, but it could print in dozens of fonts and font sizes. For those who needed complex document layouts, this could be a nightmare. For those people, a Mac was the obvious solution.

Most of us, though, didn't need WYSIWYG. I certainly didn't. What I needed was a word processor that could keep up with my typing. It also had to save quickly; neither PCs nor Macs were all that reliable in those days, and we all learned to save early and often. In my case, to this day I tend to save at the end of each paragraph. Early Macs, which used the same processor for computing, screen management, and disk control operations, were notoriously slow at saving. Despite improvements, they were pretty slow at screen operations.

Mac keyboards were all right, but they didn't have any provision for third-party improvements, and the standard Mac keyboard never was as good as the Hall-effect keyboard I used on Zeke. Eventually, Northgate brought out keyboards that would work on both Macs and PCs (I'm writing this on one of the "Pournelle-configuration" Northgates, with the Backspace key where it is on an IBM Selectric), but that wasn't for a long time.

There were other problems. I always had Macs here, and periodically I would try writing with one, but I was never tempted to change over. Understand, I was still doing most of my work in DOS and writing with a character-based text editor; and nothing the Mac had would touch that for speed and ease of use. Moreover, the PC-100 and its counterpart the NEC PC-201A were truly portable machines. My NEC with a Traveling Software "Ultra" ROM replacement gave me 8 rows of 60 columns on a battery-operated laptop, and since the output was ASCII, I could transfer those files to my PC. Then Zenith brought out a line of laptop PCs you could use anywhere.

The Mac, meantime, had no portable, and when they finally announced one (with gr eat hoopla at Universal City, about 2 miles from my house), it was a lead-acid battery monster with no off switch. I think they sold hundreds of them.

The point is that the Mac had plenty of opportunities to capture my attention, and moreover when I needed WYSIWYG, I used a Mac like everyone else. Excel and PowerPoint existed only on the Mac -- I recall that when Microsoft and IBM were still united in promoting OS/2 as the OS of the future, they did so on charts made in PowerPoint on the Mac -- and MacInTax was so much the best tax-preparation program that I once said it would be worth buying a Mac if you never used it for anything else.

What the Mac never did have was a superior word processing application. Early versions of Word, on the Mac or PC, didn't interest me at all. They were slow, cluttered, and clunky, and compared to Q&A Write, PC Write, XyWrite, or WordStar, they weren't even tempting. Early versions of Windows were no better. Screen refresh rates were hideously slow.

About thi s time, Mac word processing got better, and if I ever was tempted to discard DOS and Wintel and go to a Mac, it was in the early Windows days. Unfortunately, if I changed over, Niven would have to; and while Mac-based word processing programs were now good enough, they weren't as good as Q&A Write for text creation. It would be hard to explain to Niven why I was making him learn a new system.

It was during this era that Tom Clancy and some other well-known writers chose Macs as their writing machines.

Meanwhile, Windows got better. ATI Technologies started the trend with graphics-accelerator video boards; they were followed by a host of others, and now screen operations of bit-mapped text in Windows are as fast as we got used to with character-based text in DOS. The text looks good. New versions of Word incorporated the "Pournelle feature": white text on a blue screen, which cuts down the total light coming out of the system and I find easier to read.

Most important, new versions of Word inc orporated two major features. The first they'd always had: you could put your entire novel in one file with Word. Q&A Write and most of the other DOS character-based editors were fast because they worked entirely in memory, meaning that you couldn't get more than a few chapters into a single file, making global operations like search and replace and spelling checking fairly tedious.

The second decisive feature was document comparison. Up to that time, when Niven and I worked independently on a novel, we'd have to arrange who "owned" which chapters. "You can work on anything except chapters 11 and 13." That sort of thing. Now we just work on what we feel like and merge versions when we get together.

So, a couple of years ago, we converted to Word for Windows as our standard word processing system, and at the moment we use Word 7 (the one that's in Office 95). We're looking at Word 97, but so far I haven't found any features in it that would make it useful to Niven. It is good for Web-site creation, and, getting back to how this ramble started, that is another good reason for converting those files from the Mac to PC and maintaining my site with Wintel rather than the Mac.

The Mac's future is bound up with the success of incorporating Next Software's OpenStep development system into Mac architecture -- and OpenStep runs on Wintel systems, too, so even that isn't going to be a Mac exclusive. Apple has been saved by outsiders in the past. VisiCalc saved the Apple II. Desktop publishing, DeBabelizer, and other graphics programs saved the Mac.

OpenStep won't itself do the job, but it may lead someone to develop the killer application that will once again be "compelling" (a buzzword that seems in universal use in the small computer universe). I wish the Mac well, but I'm going to do my Web site on the PC, and I'm converting Roberta's reading-instruction program to Visual Basic for Windows.

I am reliably told that Scott McNealy of Sun Microsystems reports a dramatic increase in producti vity in his company: he ordered PowerPoint removed from every single machine in the corporation. It seems that far too many employees were spending far too much time making fancy presentations instead of thinking about the job.

I can well understand that.

The concept of the briefing complete with charts was largely developed during the years I spent in the aerospace industry. In those days, a staff engineer generated a written report and abstracted it into a briefing. If you'd done it right, anyone who had read (and understood) your report could take your charts and give your briefing to people too busy (or too full of themselves) to read the report. Sometimes that worked, sometimes it didn't, but most of it took place in the context of military staff work. The notion was that a staff would investigate and report on the details and present the alternatives so that a commander could come to a decision. (In the business world, this sort of software is called decision support, and in large companies has its own subculture of gnomes who feed statistics to it.)

This often worked, but there could be problems, particularly if the staff had a stake in the outcome. A good commander understood that the staff had a view and would often request briefings from staffers known to have a different preference. This initiated a briefing battle, and that would sometimes generate a competition for the best presentation. A good commander understood that quite often the presentation got slicker because the ideas were poor, and periodically there would come down directives forbidding fancy presentations.

Of course, there were poor commanders who could be influenced by presentation whizbangs. I never made a systematic study, but it was my impression that you found more of that kind of leader in industry than in the military. In any event, after a few years, I noticed staffers were spending more time generating briefing aids than in refining ideas.

Then came small computers. Now it's possible to make fancy p resentations with no content at all. When we first got Software Publishing's ASAP Presenter program (it's now called Active Presenter; see http://www.spco.com ), Eric Pobirs, the Chaos Manor intern, tested the presentation generator by making up a briefing on how to skin a cat, complete with digressions on advancement possibilities in cat skinning, career effect of learning cat skinning, related activities such as mole skinning, and suchlike.

It was an excellent briefing and a good illustration of the power of the program; I gather there are companies using it as a teaching aid on presentation construction. It also makes my point, that it's now possible to make a powerful presentation about nothing at all.

David Em is our graphics associate at Chaos Manor. He's a well-known artist who was among the first to use computers -- his book The Art of David Em is worth looking through if you can find it -- and his work has been exhibited in museums of modern art in Madrid, Paris, New York, and elsewhere. We've been assembling state-of-the-art desktop workstations; David does the tests. Here's his report for this month:

"First some notes on the Mac.

"I've been focusing on 2-D applications on the Mac (while there are some very good 3-D Mac programs, the board manufacturers aren't writing Mac drivers, so NT remains the 3-D desktop platform of choice at present). Programs like Fractal Design Painter and Adobe Photoshop are now virtually indistinguishable on the Mac and PC, effectively cutting the Mac's unique-tool domain advantage, but one program that exists only on the Mac (no firm NT plans are in place, according to the company) is Live Picture.

"Live Picture is a high-end image-enhancement and compositing program that translates bit-mapped files like TIFF into its own proprietary format called IVUE . Once IVUE-ed, changes to the image are based on a resolution-independent mathematical overlay that allows for incredibly fast picture updates with very small RAM requirements (tens of megabytes versus hundreds). Anyone who has waited while an image-processing program grinds away altering an image will appreciate this. It's more than just a time-saver, too. Because changes happen in near real time, you think faster and are willing to try things you wouldn't even consider under normal circumstances.

"I've had some good-size images (10 to 15 MB each) on the disk for a while that I wanted to do some subtle warping to. However, I never got to them because it was too much of a hassle, even with a dual Pentium Pro-based system and lots of RAM. (David is using the Compaq Professional Workstation 5000 with dual 200-MHz Pentium Pro processors and Elsa Gloria-L graphics boards; see "Two Heads Are Better Than One" in the May BYTE.) I opened them in Live Picture, and 30 minutes later I was done, including the file reconversions back to TIFF.

"Live Picture isn't positioned as a Photoshop killer. The company has another program called Overdrive, designed to integrate its capabilities with Photoshop, and another called Live Picture XT, which lets you alter work directly within QuarkXPress publication files.

"If your work requires you to manipulate really big pictures (I'm talking about hundreds of megabytes here), Live Picture alone can justify buying a Power Mac, especially at today's prices (about a quarter of what they were less than two years ago).

"The Mac is still a great graphics design tool at a great price with a devoted customer base, but if Apple doesn't get the Next OpenStep based-OS (Rhapsody) out the door pronto, it's going to have trouble holding on to its current market share. The latest offerings from graphics industry leaders like Adobe, Macromedia, and Fractal Design are identical on the PC and Mac in look, features, and performance. In some cases (e.g., Equilibrium Technologies' new DeBabelizer Pro), the NT versions are now superior. If these players hesitate till next year to develop -- let alone ship -- products for the new OS, that will be all she wrote.

"Three-D graphics on Windows NT is hopping. NewTek announced LightWave 3D 5.5 at the National Association of Broadcasters trade show in April. This release finally features a new interface (the old one dates back to Amiga days) plus multithreading and scads of new features. It runs on every platform from Intel to Alpha to PowerPC to Silicon Graphics, so we'll be looking at them soon (at $1995, this is a hard deal to beat).

"Also, I've got the final beta version of Caligari's trueSpace3. It's a big jump up from version 2, with some super features like physical dynamics that let you assign properties to your objects like rubber and gravity (just drop a ball and watch it bounce) or paper (turn on the wind and watch it blow away). I'll have more on trueSpace3 soon. It has some very interesting features related to 3- D Web and Virtual Reality Modeling Language (VRML) design.

"The most interesting new software packages I've played with this month are blob modelers. Jim Blinn -- a well-known graphics expert I worked with at Caltech and the Jet Propulsion Laboratory (JPL) -- wrote the first blob modeler back around 1980 at JPL. We had an Evans and Sutherland Picture System 2 (a $65,000 vector display) that we could move little blob spheres around on and make them glob together. I was really excited, because it was obvious that blobs could be a path out of the geometric look that dominated computer graphics at the time. Blobs would be perfect for modeling people, monsters, all sorts of organic critters, and objects.

"I made a crude Venus de Milo figure with it, but by the time there were enough blobs up on the screen to do anything interesting, the screen was flashing so intensely it was a miracle nobody went into an epileptic seizure. For the next 10 years, it was the Japanese who got into blob modeling, fo r some reason, and blobs became metaballs.

"Now we have blobs on the PC, and they're way cool. Of particular interest is a product called MetaReyes 3.0 from REM Infografica in Madrid. MetaReyes 3.0 is a plug-in for 3D Studio Max that goes basic metaballs one better by letting you dynamically attach the blobs to splines, creating what REM Infografica calls Metamuscles. The effect is that you can effectively sketch in 3-D! I made a 3-D sketch of a brontosaurus in about 10 minutes with it, reminding me for the first time in a while that building 3-D models can be fun and intuitive instead of excruciating and aggravating.

"Incidentally, another small company I'll have more on soon is Digimation. It makes plug-ins for 3D Studio Max, too, including a very capable metaball modeler called Clay Studio. It lets you use cubes as well as spheres as blob primitives, which can be scaled in x, y, and z dimensions. The amazing thing about this package is that it costs only $195. And to round out the picture, true Space3 has a similar metaball component built into its tool set as well. All these programs are new, powerful, cheap, and really easy to use. I think they'll have a significant impact on evolving the look of 3-D graphics produced on PCs.

"One more note on the ongoing display discussion. As previously noted, there are two big factors to consider when choosing a display, one being color purity, consistency, etc., and the other focus. I tend to put more weight on the color-accuracy side of the equation, but lately I've switched camps. The reason, surprisingly enough, is not related to image quality, but to software interface design. Every program I work with -- from Photoshop to Fractal Painter to 3D Studio Max -- uses all sorts of floating toolbars and readouts of one kind or another that have tiny little updates of one kind or another that you practically need a microscope to read when you get up into the higher screen resolutions.

"Some very good monitors like the Eizo Nanao T2-20 can barely resolve this kind of information, to the point that I've switched to the Compaq 21-inch Qvision 210 for 3-D work. The Qvision 210 isn't absolutely perfect in the color department -- although it's by no means shabby, either -- and its on-screen controls are slightly counterintuitive, but its focus is very sharp, comparing favorably to the ViewSonic P815 MegaMonitor.

"I believe it's very competitively priced, as well. At some point, we must look at the new Sony and Eizo Nanao monitors, which claim to offer the best of both worlds."

David's report ends there. I can add that I've been using ViewSonic monitors of various sizes for a couple of years now, and I don't know of any brand that offers more performance for the dollar. If you're planning on working in front of a computer screen day after day, it's important to have a monitor that doesn't produce eyestrain.

I have my choice of just about every monitor made, and I consistently find myself using ViewSonic monitors. My monk's cell upstairs h as a ViewSonic 17-inch monitor, my main system sports a ViewSonic Professional Series PT810 21-inch screen, Roberta has a ViewSonic downstairs, and I keep one at the beach house where I go to write without distractions. Meanwhile, David uses the ViewSonic P815 MegaMonitor on his main system, and it's the one to compare everything else to.

Meanwhile, we've also installed the Wacom ArtZ II digitizer on Princess, the Compaq Professional Workstation 5000. This is a small pad -- the one I have measures 9 1/2 by 13 inches overall with a drawing area a bit smaller than a standard sheet of paper -- with a stylus that you use just like a pen. There's a bar on the side; push on the bottom for a left-click, the top for a right-click. It runs under both Win 95 and NT 4.0, and both it and the mouse are active at the same time.

David Em finds it refreshing to use the ArtZ II as a mouse when his hand is tired. For myself, I find a mouse much more accurate for controlling a program, but the pen works well fo r text editing, and the digitizer is a surprisingly good mouse pad. Mostly, of course, the ArtZ II is for drawing. The drawing program I use, mostly for maps and sketches of what I want for book illustrations, is Fractal Design's Expression, and I'll have a lot more to say about it in coming months.

The most interesting thing at Internet World was Digital Equipment's Millicent. This isn't a product yet, but it soon will be. Millicent is a way of doing commerce on the Internet with extremely low per-transaction costs. Ideally, the transaction costs can be a small fraction of a cent, hence the name.

It works this way.

Suppose John Dvorak and I open a Web page, which you can access for a nickel, and you want to look at it. Meanwhile, there are other products on the Internet, from access to Web pages ranging up to mail-order catalog purchases, that you'd like to buy, but you don't want to give all these people a credit-card number. Worse, in the case of information products, the transaction costs may be higher than the product cost.

Under Millicent, you go to a broker. This may be someone you don't know, but it could just as well be the Bank of America, which already has your credit-card and account numbers. The broker issues scrip, in denominations you specify. There's some minimum overall transaction size. When you deal with the broker, you get a "wallet" of scrip.

Now your wallet has magic money issued and backed by the broker. When you want to buy something, your wallet tells the broker, and scrip gets translated into vendor scrip good for only one thing, purchases by you from that vendor. You now have a nickel's worth, and you have 2 cents of vendor scrip left over.

Keeping track of costs is done automatically. Everyone has Web stores now, and when this system gets implemented, there will be more products on the Web. Millicent can do very small transactions; so far, Cybercash comes in 25-cent increments and isn't widely used. Millicent is potentially a way for authors to make money. People probably aren't ready to read novels on the Internet -- I sure don't like doing that -- but I can see myself paying a dime for access to a short story, comic book, or newsletter. Millicent makes self-publication potentially profitable.

Millicent purchases can be anonymous. You can, for instance, give a donation to a charity knowing that your contribution won't be spent in attempts to get more out of you and subscribe to a newsletter without being put on sucker lists.

Years ago, Dvorak and I conceived of a sort of weekly Siskel and Ebert debate. Our notion was to update frequently and charge a nickel or a dime for access to the Web page. We thought the potential would be high, leading to enough revenue that we could hire assistants to check things and chase rumors. The only problem was collection: there wasn't and still isn't a simple system to let you do "impulse access" to our page, so the only revenue source was advertising. We didn't like the notion of advertising: neither of us has a sales staff, and there's a clear possibility of a conflict of interest.

Millicent, if it appears, would let us implement the original plan. I'm looking forward to it.

I am told I unjustly maligned the Swiss shareware program known as Windows Commander, which is available over the Internet. Readers tell me it's excellent, and what I took for a threatening message when you use an unregistered copy is merely an inconvenience with no threatening content. You can use the unregistered program without fear, and if you like it, you can send a check by ordinary mail to Switzerland. Many do like it, and it's apparently improved rather frequently.

I'm still happy enough with Symantec's Commanders; that is, I keep Commander 4 available in a DOS windo w and do most of my DOS applications from that. I also have a copy of Symantec's stealth product Commander 95 for Win 95 operations, including long filenames. One of these days I'll have another look at the shareware Windows Commander. I have to say it has a number of readers who are loyal enough to have written me about how pleased they are with it.

I continue my experiments with pointing devices. My main system sports a Microsoft IntelliMouse, the mouse with the little wheel in the middle, because I like its feel, but I've disconnected the IntelliMouse software. A story goes with that.

The IntelliMouse and its associated software are excellent for nearly all Win 95 applications. Some applications, like Office 95, don't recognize the "scroll" features of the wheel, but trying it with them does no harm, and meanwhile, the wheel can be programmed as a third mouse button. I used that for "double-click," but you can make it be any one of a variety of operations through the IntelliMouse software. The IntelliMouse and its software together make the best mouse I have ever used.

However, the IntelliMouse software will break a number of DOS games, including Blizzard Entertainment's Warcraft II: Tides of Darkness. The break happens at random intervals: you might go half an hour, you might go 5 minutes, but eventually it happens. Suddenly the mouse and the cursor are only very loosely coupled. Move the mouse and the cursor moves, but not where you expected it to. Worse, the keyboard is locked out. Ctrl-Alt-Del does nothing. The only way out is hardware reset.

This can happen in games designed for Win 95 as well.

Since one of my jobs is to test games -- it's nice work if you can get it -- I decided the IntelliMouse software had to go.

Unfortunately, uninstalling that software was a near disaster. For reasons I don't understand, when the IntelliMouse software was gone and I reset the system as instructed, it came up in VGA 640- by 480-pixel mode, and nothing I could do would get a high er screen resolution!

Eventually, I went on the Internet looking for new drivers for the Orchid Fahrenheit Video 3D board that's in the system just now, and that was an adventure by itself. While Orchid offers a number of drivers for their boards, I couldn't figure out what board I had. It's not written on the board itself, and if it has any kind of software query, I couldn't invoke it. I finally found the box with the original Orchid drivers and even that doesn't tell me enough to identify this particular board. You won't have that problem. I have, it turns out, one of Orchid's early prerelease boards, and it doesn't have a designation.

Reinstalling the Orchid drivers from the original CD-ROM fixed my problem, and don't take this as any kind of indictment of the Orchid Fahrenheit Video 3D board. I could easily change back to the Matrox MGA Millennium board that came with Cyrus, and I haven't bothered because the Fahrenheit board is at least good enough for both 2-D and 3-D. Meanwhile, spelunking on the Orchid Web site finds that although the Fahrenheit Video 3D board isn't on their Web page, there is a new driver on the FTP site. In this case, there's a not very conspicuous button to take you to the FTP site; some places, you just have to guess at the FTP-site address.

What does worry me is how removing the IntelliMouse software could clobber my video drivers. I have no answer to that, and so far neither does anyone at Microsoft, although they're looking into it.

I reported my problems with IntelliMouse to the Microsoft people designing Memphis, and they promise to look into what was blowing up my games. Moreover, Billy Brackenridge, Microsoft's in-house games fanatic, is on the case, so I believe things will happen.

Meanwhile, I'm using the IntelliMouse but with older Win 95 generic PS/2 mouse software. I'll install the IntelliMouse software when Microsoft reports a revision that doesn't blow up games.

The removable-cartridge mass-storage situation is best described as an em barrassment of riches. We now have Iomega Zip and Jaz, three sizes of SyQuest drives, the Olympus glass disk I mentioned last month, Fujitsu magneto-optical (MO) drives in two sizes, and two sizes of Nomai drives.

Every one of these works. All the electronic systems are comparable in speed and access time, although I make no doubt that one or another have proofs of speed superiority. The cartridge costs are comparable. All of them work with SCSI-1 controllers. Iomega makes a parallel-port version of their ubiquitous Zip drive, while Olympus very cleverly allows you to connect as straight SCSI or plug into the parallel port with their SCSI-to-parallel converter built into a special cable. The SCSI drives all connect without special software, but most come with some optimizing drivers. Of course, you have to run a program off a floppy disk to connect up a parallel-port version.

The only fundamental difference here is between the Fujitsu and Olympus products, which use a glass medium (and can rea d and write to each other's disks), and the others, which are purely magnetic. Optical disks are noticeably slower than magnetic ones, but they are far more permanent; electronic cartridges will, over years to decades, lose bits until they're worthless.

Every one of the SCSI devices remap your drive letters if you boot up with a cartridge in the drive, and nearly all will bump your CD-ROM drive down a letter whether there's a cartridge in there or not. You'll just have to learn to live with that. Several readers have suggested that the remedy is to use Device Manager to map your CD-ROM drive to be R and have done with it; it will keep that letter unless you fill all the letters before R with other devices.

SyQuest makes an internal version of their 1.4-GB drive, and I'm going to install that in the new Pentium II system we're building; it will be the boot device, meaning we can boot up in alternate OSes by changing cartridges. There will always be a place for the Fujitsu DynaMO, because I use that for archiving. I use the Olympus to talk to systems with no CD-ROM drive and no SCSI. I have two Zip drives because Larry Niven and I use them for file transfer and safety copies (I also make safety copies on the Fujitsu). Having said that, I point out that all the others I have mentioned work about as well as these do.

I'll have more on this in months to come, but right now, if you're selecting a removable-cartridge-drive system, you're probably safe enough deciding on price. Do look into medium availability, and it probably wouldn't hurt to buy a few extra cartridges just in case.

---

Product Information

ArtZ II........................$389.99

Wacom Technology Corp.
Vancouver, WA
Phone:    800-922-6613
Phone:    360-750-8882
Fax:      360-750-8924
Internet: 
http://www.wacom.com


Clay Studio....................$195.00

Digimation, Inc.
St. Rose, LA
Phone:    800-854-4496
Phone:    504-468-7898
Fax:      504-468-5494
Internet: 
http://www.digimation.com


LightWave 3D 5.5.............$1,995.00

NewTek, Inc.
Topeka, KS
Phone:    800-862-7837
Phone:    913-228-8000
Fax:      913-228-8001
Internet: 
http://www.newtek.com


Live Picture...................$695.00

Live Picture, Inc.
Scotts Valley, CA
Phone:    800-724-7900
Phone:    408-438-9610
Fax:      408-438-9604
Internet: 
http://www.livepicture.com


MetaReyes 3.0..................$695.00

REM Infografica
Madrid, Spain
Phone:    + 34 1 319 4155
Fax:      + 34 1 319 4174
E-mail:   
info@infografica.com

Internet: 
http://www.infografica.com/


trueSpace3.....................$795.00

Caligari Corp.
Mountain View, CA
Phone:    415-390-9600
Fax:      415-390-9755
E-mail:   
sales@caligari.com

Internet: 
h
ttp://www.caligari.com


HotBYTEs
 - information on products covered or advertised in BYTE

---

Jerry Pournelle is a science fiction writer and BYTE's senior contributing editor. You can write to Jerry c/o BYTE, 24 Hartwell Ave., Lexington, MA 02173. Please include a self-addressed, stamped envelope and put your address on the letter as well as on the envelope. Due to the high volume of letters, Jerry cannot guarantee a personal reply. You can also contact him on the Internet or BIX at jerryp@bix.com .