Of COM Ports and Digital Frogs

Jerry attends an education conference and then learns a thing or two about communications software in Windows 95

Jerry Pournelle

It has been a busy month. First off, I was the keynote speaker at the Association for Media and Technology in Education in Canada, which met this year at the University of Guelph. The city of Guelph is about an hour from Toronto, just far enough that it hasn't lost the feel of a university town in a rural setting; it reminded me of Iowa City in the 1950s. AMTEC is one of the older organizations promoting technology in education.

A major issue in education technology is distance learning. Studies by the Danish Ministry of Education conclude that the critical cost factor is how to make low-paid people--such as students--do the work formerly done by high-paid people. Danish and other studies also indicate that the general result of applying high technology to education is to increase educational quality, but at increased costs; it's rare when high-tech education saves money. That's a big disappointment in this era of falling education budgets.

Guelph is the major center of Canadian veterinary education. Lifelearn V., a private company in a joint venture with the university, has developed one way to both increase education quality and save money. They've got the first really practical applications for CD-I (CD Interactive) I've seen.

Lifelearn uses CD-I for multiple reasons. First, it's easy to use, and it requires no computer experience. Second, they can give away the CD-I box, which outputs NTSC video into a TV set, as part of the course. Finally, since many parts of the course materials feature real-time demonstrations, they want interactive full-motion video, which CD-I supplies nicely.

Contin uing education is important in many professions. Sometimes it does some good, but, alas, all too often continuing education workshops degenerate into a series of Mickey Mouse sessions at which you get your ticket punched while vacationing on Maui or a Phoenix golf course. Some of those refresher workshops may be valuable, but Lifelearn offers an alternative. For less money, you can get the Interactive Multimedia Self-Study Modules prepared by veterinary experts accredited in both Canada and the U.S. Because it's on CD-I and audiotape, everyone in the clinic can take the machine home and go through the course materials. Course content varies from canine dental surgery through cardiology to dairy farm health management.

The Lifelearn CD-I system impressed me a lot. I'm certain that soon enough this kind of thing will be available--from one source or another--for dozens of professions. Meanwhile, if you're a veterinarian, you should know about the Interactive Multimedia Self-Study Modules.

O f course, there's an awful lot of pure hype about educational software. One (very badly produced) video I have spends half an hour telling about its problem-solving approach to education. Principals wax eloquent on how this launches high school students into lifetime learning. Other teachers tell us that the kids just love this stuff because it's not a boring book. Then we're informed that "problem solving is a very unique process." You can re-create electronically just what the student did to solve the problem.

What they're selling is authoring materials. The value of the course will depend entirely on the teacher choosing the right problems to solve. While this may be valuable, it's hardly new.

This is all of a piece with the new education fad that says it's not important what kids learn. "We teach them how to think, not what to think." That sounds wonderful until you ask the next question: What is it they are to think about? And must they discard 2000 years of history--largely a hist ory of problem solving?

Long years ago when I was a student, there was an education fad called general semantics. By studying the science of meaning, we were going to solve all human problems. Like all education fads, this one contained some valuable (if not always original) insights. One of these was that humans are time binders: they don't have to learn by making the same mistakes their ancestors did. We don't have to discover all facts for ourselves.

It's clear that learning facts without understanding isn't much of an education, and students are highly motivated to play games rather than study facts. But the weakness of the problem-solving approach to education is that it's no use solving problems unless they are related to the real world; and while the ability to think things through is valuable, sometimes what you need is to be told how someone else did it.

We tend to learn to do what we've already done. Every sports coach understands this. Left to themselves, students generally won 't stumble onto proper technique. Take fencing as an example. Hand a class of beginners weapons and protective equipment, and in a week, they'll have "problem-solved" their way to so many bad habits they may never be any good.

I've recently seen essays criticizing the hypertext concept as undeliverable hype. Now it's true that despite a decade of work and some financing from Autodesk, Ted Nelson and his associates didn't finish Xanadu; but that's not the main problem with hypertext.

The big problem is the hypertext concept itself. For example, there's Nelson's book, which you can start reading anywhere you like and read the chapters in any order. That's only a book, of course. His ultimate vision was Xanadu, computers connected on-line to give you all knowledge as hypertext, so that you could read everything in the world in any order you liked: the universe of knowledge without any imposed structure.

It's attractive. We've all had the experience of going to an encyclopedia to look for one thing and emerging hours later. We often learn something that way, too; but I suspect the ones who learn the most are those who came to the encyclopedia with an intellectual framework into which they could put their new knowledge. Unorganized facts aren't science, they're merely anecdotes; it takes structured theory to turn anecdotes into data.

We don't have Xanadu yet, but we do have hypertext CD-ROMs. Most have little or no structure. You can peel off facts in any order you like. These may be useful to experts well grounded in the subject matter, but in the hands of beginners, they're more likely to be tools for amusement rather than for learning. The same is true of unstructured problem-solving education. It may generate enthusiasm, but all too often, it's the enthusiasm of the beginning fencer handed weapons and a mask.

Lifelearn's educational approach is successful because they're building on a solid foundation. They don't teach the basics of veterinary medi cine, nor are they concerned with a general education in problem solving. What they do is show already competent people new developments in their field, along with practical techniques they can use.

Lifelearn has a large staff and a big budget. Digital Frog International has neither.

I don't know how many frogs have been slaughtered to provide subjects for dissection in high school labs, but Digital Frog's "frog-friendly software" may help to change that. The Digital Frog is a CD-ROM developed by students on a shoestring; their entire capital investment, including a Power Mac, was under $10,000.

They used a high-quality 35mm macro camera to take pictures of each stage of the dissection of a frog by an expert. Shots were made from many angles, and the whole thing was synchronized with a lecture. The pictures were digitized by turning the rolls into Photo CDs. They used an inexpensive JVC camcorder to capture images for QuickTime movies of frog activities, such as a frog catching a fly.

Then they added QuickTime animations, drawings, and diagrams, with excellent narration. The result is far more instructive--at least to me--than dissecting a frog, and there's no formaldehyde smell. The Digital Frog won the "best of show" award at AMTEC, and rightly so.

If you teach high school biology, you'll definitely want the Digital Frog. It's an excellent example of what new technology and ingenuity can do for education.

My latest trip was to Microsoft for another dog and pony show about Windows 95. I've been using W95 on my main system for about three months now, going through a dozen "builds" as Microsoft fixes reported bugs. I have to say I like it; in particular, I like the user interface better than those of either Windows or OS/2. More important, though, it works.

There are some anomalies. I'll get to one of them in a moment; but the important thing is that I've had far less trouble getting used to W95 than I did Windows itself. Longtime readers will remem ber many columns in which I was screaming in frustration. That hasn't happened with W95.

One anomaly involves QEMM. W95 installs from a setup program, and it doesn't seem to matter whether you're installing over DOS, Windows, or an earlier W95. In each case, you get a warning that you're running QEMM, and you should disable it until the installation is finished, or else W95 may not identify all your hardware correctly. I suspect that mostly means that QEMM loads some device drivers into high memory and W95 isn't sure it will find them all; in any event, I have ignored that message in the past with no ill effects.

This last time, though, I decided to heed the message. I canceled the installation, removed all references to QEMM from my CONFIG.SYS and AUTOEXEC.BAT files, and put in DOS HIMEM.SYS and EMM386.EXE. Then, just for good measure, I exited W95 with the option to boot up in DOS and ran the DOS MEMMAKER.EXE program, answering "yes" to the question about running programs that need expanded me mory.

The result wasn't good. Not only did I end up with DOS windows that were about 100 KB smaller--far too small to run most games--but my expanded memory had vanished as well. I rebooted. That automatically brings the machine back up in W95. When I ran the setup program again to finish my upgrade, I was told that I'd interrupted it last time and was warned there might be trouble; but there wasn't any difficulty, except that I got messages that EMM386 couldn't load, and my DOS windows remained tiny. I put up with that for about 5 minutes before I overwrote the CONFIG.SYS and AUTOEXEC.BAT files with my older versions containing QEMM. When I rebooted and let QEMM do its thing, I had no problems. My DOS windows are 590 KB, and expanded memory works again.

I've been using Franklin Quest's Ascend PIM (personal information manager) for several years now. Telemagic is a far better contact manager, but it's designed for a much larger operation than mine. While there are many good thing s about Arabesque's Ecco, Ascend is good enough.

I've just installed version 5.0, and the upgrade illuminates a problem with W95 communications.

Despite the improvements in Procomm Plus for Windows, I unrepentantly use Procomm 2. I'm used to it. It runs on my Gateway HandBook (a 286) and does fine in a DOS window; but it has quirks. After I switched to W95, I had an annoying glitch. At first, Procomm couldn't find the modem. When I hit Escape and dialed again, lo!, all was well.

Naturally I blamed that on W95. Then I found that Ascend 4.0 worked just fine in Windows 3.11 and W95, but version 5.0 wouldn't dial in W95. Instead, I got a Windows message that some other device had the COM port. Franklin Quest had no advice--surprisingly, they have never tested Ascend with W95--but they told me that Ascend 4.0 had its own dialer, while version 5.0 uses the Microsoft Dialer built into Windows.

You access the controls for the Microsoft Dialer through the Telephony button on the control pan el. For reasons having to do with cable connections, I've used COM1 for the mouse and COM2 for the modem since DOS days. I had no problems with SideKick, Desqview, or any version of Windows; but with W95, no matter that I told Telephony to use COM2, Ascend 5.0 would report that the communications device was in use by another program. Finally, in exasperation, I shut down the machine, plugged the mouse into COM2, and connected the modem to COM1. Then I told both Procomm 2 and Telephony what I'd done. That fixed it. Ascend 5.0 dials just fine now.

Now I have discovered that if you give Procomm 2 an initialization string, it must have Control-M at the end, else it waits for a Return. I lost the Control-M while installing W95 (my fault I'm sure); it was never a problem with W95 itself. My apologies to Microsoft: they've been trying to fix that bug since I reported it.

Although that fixed the problem--Ascend 5.0 dials just fine now--alas, it has not fixed the "must access it twice" problem with Proco mm 2, which remains as an annoyance. So it goes.

OS/2 Warp Connect is nifty, and it really is an improvement over standard Warp. In theory, it's still only Warp 3.0 with connectivity; in practice, they've incorporated some bug fixes and made installation simpler by adding more device drivers.

OS/2 is still harder to install than it ought to be. Every time I say that, I get letters from readers who bought one or another flavor of Warp and had absolutely no problems with the installation, and others who think it was easier to install than Windows ever was, so your mileage may vary. Once installed, OS/2 Warp Connect is pretty solid. Unlike W95, which still contains some 16-bit code, OS/2 is all 32-bit. With only a few windows open, there's little difference in speed between OS/2 Warp Connect and the test versions of W95; but if you keep a lot of windows open and do a lot of multitasking, the difference can be dramatic.

Using the IBM Pentium ValuePoint, I've managed to get thr ee simultaneous communications programs--two using 9600-bps modems, and one using a serial port--as well as a print job to run in OS/2. The printing was pretty slow, but the communications tasks worked without losing data. I haven't tried that with W95, but I don't need to. Just keeping a number of windows open (and doing nothing) will noticeably slow down W95.

The big new feature of OS/2 Warp Connect is built-in peer-to-peer networking capability. OS/2 Warp Connect supports IBM LAN Server 2, 3, and 4, and the LAN Server on AIX and AS/400. You can connect to Windows for Workgroups, Windows NT Server, W95, and the Microsoft LAN Manager, as well as all versions of Novell NetWare. The feature set is comparable to W4WG, with cut and paste across the network. I mildly prefer the W95 user interface, but the Warp interface is good enough.

OS/2 Warp Connect works just fine, with one exception. In Windows and W95, if you do Ctrl-Alt-Del, you get a dialog box that gets you back to the OS, where you can ch oose to shut down individual applications or the entire system. Warp doesn't do that. If you press Ctrl-Alt-Del, the system will reboot without further ceremony. Alas, that means that if you run a particularly badly behaved application, you may find yourself unable to get back to OS/2. That happens more often with bad Windows applications in Warp, but I've had it happen with a DOS program as well, and it's a terribly frustrating experience.

One reason Microsoft held its latest dog and pony show was to impress journalists with just how many software developers are writing applications for W95; it worked. About a hundred companies, hardware and software, had booths in a miniature trade show. The booths were small, not flashy, and the emphasis was on technical demonstrations. It reminded me of the early days of the West Coast Computer Faire.

Naturally, the Microsoft Applications Group was showing the most products, including new versions of Microsoft Office; but there were many othe rs. Traveling Software was there with new versions of LapLink for Windows. You'll really like what they can do with W95. Philippe Kahn, still chairman of Borland but no longer running that company, was there demonstrating Starfish Software's SideKick for Windows.

Symantec was there, with a new version of Norton Utilities for W95. I use that, and I'd hate to live without it. They also have a new Norton Navigator (a desktop replacement) for W95. I've got it, but I don't really feel the need; I rather like the W95 interface. But if you get W95, be sure to get the appropriate Norton Utilities.

You'll also need the Windows 95 Resource Kit from Microsoft Press. It has over 1300 pages and goes into great detail on stuff you'll want to know. There's a good section on using long filenames and what happens if you transfer those files to systems that don't support long filenames. Reading that will lead you to look into long file extensions--you're no longer limited to three characters aft er the dot--and how those can be used to tell W95 things about a file. That will lead you to read the section on the Registry, a W95 trick to cut down on the sizes of INI files.

Bottom-line question for Windows users: Should you change OSes? In my judgment, yes you should. W95, Windows NT, and OS/2 Warp Connect are all significant improvements over Windows and W4WG. You'll be better off with one of those.

Deciding which one isn't so easy. If you're operating in a large corporate environment, you should probably be considering Windows NT versus OS/2 Warp Connect plus OS/2 LAN Server. You'll certainly want to consider Lotus Notes, and now that IBM is buying Lotus, you'll want to watch developments there.

For home users, the choice is a bit simpler. The less you like fooling around with your machine, the more you're going to appreciate W95. You're far more likely to have a painless upgrade going from Windows to W95 than you will when switching to Warp.

One big attracti on of OS/2 has been that it is a better DOS than Windows and, for that matter, a better DOS than DOS. That remains true, but it's not a better DOS than W95--and it's certainly not a better Windows than W95. The more you run Windows (not W95, just Windows) applications, the more you'll appreciate W95. And, of course, we don't even know what IBM plans for handling applications written for W95 itself. We do know there will be far more applications written for W95 than for OS/2.

I'm keeping both. We'll continue to run OS/2 Warp Connect, but I have to say my prime machine is already running W95. That could change. Stay tuned.

It's silly, but I'm still taking two laptops on trips. The Gateway Liberty 2000 remains my favorite for working on airplanes and in meetings, but the Zenith Z-Noteflex gets set up in my hotel room and is used for heavy-duty work there.

Part of that is Zenith's reliability. The Gateway Liberty is reliable enough, but the catch that holds the battery is next to the one that secures the removable hard drive, and I have now twice managed to unlatch that drive. The result is that the drive comes slightly loose, and you have to reboot. I've never lost any data this way, but it worries me. Of course, I can fix the problem forever with duct tape. I'm not really worried about the Liberty.

The other part is the keyboard. The Z-Noteflex's keyboard is just better for typing when I'm trying to bang out text. It's not that the Liberty's keyboard is bad, just that the other one is better. And the Z-Noteflex has a built-in floppy drive, while the Liberty's floppy drive is an external attachment. This makes the Z-Noteflex heavy enough that I don't really want to put it into a briefcase.

The upshot is that I've got one of those wheeled carry-on flight bags, and when I stuff it with two computers, their power supplies, a couple of manuals, and my Ascend notebook, the thing is heavy enough to leave ruts in the tarmac. But I've never had any problem stuffing it into an overhead rack, and I need the exercise.

The Z-Noteflex has a Data Race RediCard RC-1496 data/fax modem. It works fine at 9600 bps. But it needs a special cable that plugs into the PCMCIA card on one end and connects to a phone line on the other; more than once when I've been on-line, something jarred the cable connector enough to make the system hang up.

By contrast, the Liberty has a Megahertz 14.4-Kbps PCMCIA data/fax modem with XJack. That also works just fine at 9600 bps, and the XJack connector lets you plug a normal phone cord into it. I've never had that shake loose. I've tested both modems for months now, and while I have no preference on performance, the XJack's convenience is a deciding factor. I recommend the Megahertz PCMCIA data/fax modem card.

It's easy to forget that the first A in NASA stands for aeronautics; but in fact NASA Ames does some great work. One such is a software simulation of a Boeing 737. As its developer Steve Casner put it, there's someth ing uncanny about carrying around the brains of a big modern airplane under your arm. The simulation runs on a Mac PowerBook. If you're interested in finding out more, you can E-mail him at casner@eos.arc.-nasa.gov, or write to Stephen Casner, NASA Ames, Mail Stop 262-4, Moffett Field, CA 94035.

The book of the month is Crime , edited by James Q. Wilson and Joan Petersilia (ICS Press, 1995). This will tell you more than you want to know about crime in this country. Essays are presented from nearly every rational point of view. It's not fun reading, but perhaps it's time citizens gave some heavy thought to the problem.

Two computer books of the month . The first one is Jeannette Lawrence's Introduction to Neural Networks (California Scientific Software, 1993). This isn't easy reading, but no book on neural networks is; but it is comprehensible when it talks about back propagation and the like. Neural networks are becoming increasingly important as computers get more powerful.

The second computer book of the month is by Ronny Richardson, The Ultimate Batch File Book

Next month: more on connectivity, and a whole mess of small applications.

---

WHERE TO FIND

In Ascend 5.0 ($149.95), some changes are for the better, and I expect overall it's a genuine improvement. Contact Franklin Quest Co., Salt Lake City, UT, (800) 877-1814 or (801) 975-9992; on CompuServe, go franklin.

The Digital Frog (US$170) is an excellent example of what new technology and ingenuity can do for education. Contact Digital Frog International, Puslinch, Ontario , Canada, (519) 766-1097; dfi@sentex.net.

The Interactive Multimedia Self-Study Modules (per module, US$299) for veterinary medicine impressed me a lot. Contact Lifelearn V., Inc., Guelph, Ontario, Canada, (800) 375-7994 or (519) 767-5043; rnigol@ovcnet.uoguelph.ca.

The big new feature of OS/2 Warp Connect (fullpack edition with Win-OS/2 code, $299) is built-in peer-to-peer networking capability. IBM Corp., Austin, TX, (800) 342-6672 or call your local IBM dealer; http://www.ibm.com .

The PCMCIA data/fax modem with XJack ($249) works just fine at 9600 bps, and the XJack connector lets you plug a normal phone cord into it. Contact Megahertz Corp., Salt Lake City, UT (800) 527-8677 or (801) 320-7000; http://www.xmission.com/mhz .

---

Jerry Pournelle holds a doctorate in psychology and is a science fiction writer who also earns a comfortable living writing about computers present and future. Jerry welcomes readers' comments and opinions. Send a self-addressed, stamped envelope to Jerry Pournelle, c/o BYTE, One Phoenix Mill Lane, Peterborough, NH 03458. Please 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 .