WEB EXCLUSIVE: Of Digital Wallets and Digital Video

Jerry prepares to put Digital Equipment's Millicent to a real test-while David Em reports on the progress of the graphics revolution at Chaos Manor. And in "From the Workbench," in the January issue of BYTE, Jerry builds his first AMD K6-based system.

Jerry Pournelle

My trip to the Bay Area was to attend a Digital Equipment conference on Millicent. I touched on Millicent in a previous column as a new technology to watch. Now, not long after you read this, you'll be able to try it.

With any luck, John Dvorak and I will have Discontinui ty.com going in January. It will feature a weekly debate/discussion between Dvorak and me on some aspect of the computer revolution. Part of that discussion will include our separate 100-point-scale ratings of software new and old, along with our justification for the rating each of us assigns. We'll also have a CD-ROM Pick of the Week, selected reader letters with r eplies, game discussions, and staff reports. All this will be additional material not covered in our columns, and if this reminds you of a pair of Chicago movie critics, I can't say the resemblance is accidental.

Some of you will recall that John and I experimented with this before. That was a test, but it was a lot of work as a labor of love. We hope this time to do it for real. The difference is Millicent.

Digital's Millicent is a system for charging between three-hundredths of a cent and $50 for access to information on the Web. The transaction costs are small fractions of a cent-hence the name Mi llicent. Unlike other Web-cash systems, Millicent uses tokens that are the equivalent of cash. Each transaction cancels the current token and creates a new one as change. The upshot is that Millicent requires no third-party notification at transaction time; the entire transaction is handled between two systems already in communication. Thus, it's fast and well-nigh painless. A big server like a Digital Alpha ought to be able to handle hundreds of transactions a second.

Millicent allows fixed prices, like a cent or two per page, or subscriptions. Subscriptions can be for seconds, minutes, hours, days, or longer, and can cost a few cents or a few dollars. We plan to charge a dime to subscribe for a week or until we put up another debate. We don't figure there's much price resistance at that level, and by making the subscription last at least a week but automatically renewed until we put up new material, we can be sure we provide the reader's money's worth.

Our previous experiment worked in the sense

The key to the Millicent system is getting a few hundred thousand people to start using it. The process begins when a user downloads-free, of course-the Millicent "wallet" software and installs it as a kind of plug-in to the browser. The next step is to charge the wallet with a few dollars obtained from a broker. As I write this, we're not sure who the U.S. broker will be, although Digital guarantees there will be at least one before 1998. There are already two under contract in Europe.

You can pay the broker in many ways, such a s by a 900 telephone call (as an introductory offer, you get, say, $2.50 worth of broker scrip and your phone number is billed $2.00), existing electronic cash, or through traditional credit-card transactions. Once that is done, the rest happens invisibly. You go to the Discontinuity.com site, you see a subscription offer you like, your wallet makes an offer and tenders a scrip token, the server site accepts the scrip and sends you back change, and voilá!, you have access to the page you want. With luck, this will take less than a second all around.

The Millicent security system sees to it that you can't counterfeit scrip or use it more than once. Later on, the server operator tenders the collected scrip to the broker, collects money less the broker's discount, takes a fee, and sends the rest to the content providers, in this case, Dvorak and me.

We'll see, of course.

There are a lot of uncertainties. Will people pay a small amount for instant access to Web content when there's so muc h free stuff out there? If they will, and focus-group research indicates there are a lot of people willing to pay if the content is both timely and good, can the transaction be made both painless and simple? And, of course, there's a chicken-and-egg problem, just as there was in the early days of credit cards: until there's something you can buy with a credit card (or Millicent), you have no reason to have either, and until there are a lot of people with credit cards (or Millicent), vendors have no reason to make arrangements to accept them as payment.

I don't know how long it will take to break that cycle. Perhaps not long at all. Digital intends to supply both the user and the vendor software free of charge (they'll make their money from licensing software to the broker and, I suspect, from increased sales of big servers to handle Millicent Web sites). This could catch on quick.

Anyway, Dvorak and I hope to be among the first to try Millicent on a large scale; look for Discontinuity.com after th e first of the year. And maybe, just maybe, this is the start of a new phase in the computer revolution.

My desk is piled with new stuff and more keeps coming in, much of it having to do with Web management and creation. I'm trying to keep up and still finish a novel, but it hasn't been easy.

Fortunately, my associates have a lot of important things to say this month. First, there is the next installment of David Em's graphics report, which I've been promising.

Chaos Manor Graphics Report by David Em

We've been threatening to get serious about digital video since the beginning of the year. The plan was to have a small video lab running before summer was over, and since September is usually the hottest month here in Southern California, technically we just made it. This is to say we've gotten as far as dipping our collective big toe into a very, very deep pool.

We decided to set up low-, middle-, and high-end video environments. On the low end, we put a TurboTV board from i xMicro ( http://www.ixmicro.com/ ) in our Power Mac 9500. For the midrange, we chose the miroVideo DC30 Professional card (the miro Digital Video Group was recently acquired by Pinnacle Systems; http://www.pinnaclesys.com/ ) running in a Dell XPS Pro P200 system. On the high end, we went with TrueVision's ( http://www.truevision.com/ ) Targa 2000 RTX bundled in an Intergraph TD-200.

Starting Small

The only video system we've had no problem getting going was TurboTV. This is a PCI card that delivers video-input capability from a camera, VCR, TV antenna, or cable TV. I popped the card in an open slot, installed the drivers off the CD, and it worked.

The card is positioned as a video viewing system. Thus, it does not have video or audio outputs, meaning you can't record back to tape with it. It can display video in a window with up to 640- by 480-pixel screen resolution. For anyone interested in videoconferencing, it will probably work just fine.

As it happened, I had a need to view some ancient 3/4-inch tapes I'd made back in 1978 off an Evans and Sutherland frame buffer (a $65,000 device now known as a $49 graphics card). They had been recorded under less-than-ideal circumstances. I used TurboTV's brightness, color, and contrast controls to adjust the incoming video signal and was rewarded with corrected video that looked great. I split the signal from the 3/4-inch playback deck between a VHS recording dec k and our Compaq QVision 210 21-inch monitor, and was able to display and record the material I was after with no problem.

We have a couple of ixMicro's Twin Turbo graphics cards on our Mac, and we've had good results from them. The addition of TurboTV to its family of products is welcome. It works in both Macs and PCs, and I've seen it for sale at Fry's Electronics for $89. If you need to view video or TV on your home or business machine, TurboTV is an inexpensive no-hassle solution.

Upping the Ante

I saw the miroVideo DC30 card at the 1996 fall Comdex and was impressed. The DC30 captures and outputs good-quality video and audio in a variety of formats, uses a single PCI slot, and comes bundled with Adobe's Premiere 4.2 editing software, all for under a thousand bucks; truly a great deal.

It turned out that miro was waiting for a new driver model from Microsoft before it could release Windows NT 4.0 drivers, so we had to use Windows 95, even though we would have preferred NT. As fa te would have it, the only Windows 95 system we had available at the time was a dual-processor system-and the miro Win 95 drivers didn't like dual processors. Although they advertised Mac compatibility, those drivers weren't yet available, even in beta.

How time goes by. The driver model miro needs for NT support still doesn't exist, says Microsoft (a fact miro must be gnashing its teeth about, since it must have cost the company a few million dollars in sales by now), but we now have a Mac beta as well as a single-processor Win 95 system to try it on, so stay tuned.

I Can't See You

Finally, our Intergraph TD-200 cum TrueVision Targa 2000 RTX. The Targa is a very-high-end (it came out at $10,995) video I/O solution capable of rendering many effects (e.g., fades in real time) that most boards cannot. We had a devil of a time figuring out how to get it going, because it comes with virtually no documentation. What it does have is not only unclear, but in some cases incorrect-inexcusable in a board of this class.

Troubleshooting information is scarce, too. The video-capture software included with the board is unnecessarily complex and poorly laid out, and built-in help is virtually nonexistent. To their credit, the TrueVision technical-support department was helpful, friendly, and knowledgeable, but even they admitted that the current manuals were subpar. If you plan to purchase a desktop video system on this level, get it from a dealer who will guarantee integration.

Our Targa 2000 RTX came with an optional external breakout box. The breakout box is a hefty item, weighing a good 5 pounds, with three big rows of video and audio connectors. It connects to the 2000 RTX via a thick cable with at least 60 tiny fragile pins on each connector.

The box allows all manner of analog video inputs and outputs: composite (e.g., low-price camcorders), S-Video (the "SVHS" connector used by Hi-8 and S-VHS camcorders), and component (Betacam and other expensive high-end analog cameras). All thes e inputs are echoed on a bank of pass-through connectors, which can be tremendously useful in a production environment.

The 2000 RTX does both NTSC (the TV standard for North America and Japan) and PAL (the standard for most of Europe) video, though only one at a time; on the audio side, it offers RCA and balanced XLR stereo I/O, another reason it's the professional's choice.

TrueVision does not yet offer NT drivers for Alpha. This is a pity, because programs such as In:Sync's Speed Razor editor are nothing short of awesome on Alpha.

Back to our struggle: after much discussion with TrueVision's technical-support folks, we came to the conclusion that our board was defective (it's amazing how many defective boards have come our way of late; we've had one from virtually every major board manufacturer this year).

So we're not there yet, but we're a lot closer. Meantime, we're building up a library of some of the most interesting editing, effects, and compositing applications in the desktop video firmament. Once we're up and running, we'll finally be able to start reporting on the rapidly evolving state of the art of digital video.

One final but very interesting desktop video fact: Because the Mac can run multiple monitors and display cards at varying resolutions and color depths-and NT 4.0 can't-the Mac is superior to the PC when it comes to video.

Under NT, you cannot have one great VGA card such as the Number Nine Revolution 3D or the Matrox Millennium II for your application and a specialized one just for displaying TV. In our case, we use TrueVision's built-in (and rather slow) VGA as our sole display, or we use a good VGA card for NT and view our video on a regular TV monitor. Neither solution is satisfactory.

Until NT 5.0 comes out next year (hopefully), score a big win for Apple in the video platform wars.

The Road to Multimedia

The other 400-pound gorilla we've decided to wrestle to the ground is desktop multimedia, and on this front, there's a lot of g ood news.

I've migrated most of our sound, video-editing, and 2-D graphics applications over to Larnu, our dual 200-MHz Pentium Pro Compaq Professional Workstation 5000. Larnu's multiprocessing horsepower is great when it comes to crunching through the oceans of data multimedia typically requires, and it's still our best all-around machine.

Compaq has recently gone upscale to Pentium IIs with their new Workstation 5100 and 6000 series. Besides being faster, the Pentium II's MMX capabilities are starting to make a significant performance difference in MMX-enabled programs such as Adobe Photoshop and Macromedia Director. Compaq's Workstation 8000 (which can hold up to four Pentium Pro/200 chips and 3, yes 3, GB of RAM) competes directly with the high-end Intergraph offerings. Compaq, like everyone else, chose the Pentium Pro for machines with more than two processors, because the Pentium II motherboard architecture is limited to two processors, at least for now.

Larnu's two Elsa Gloria-L cards provide a 3200- by 1200-pixel workspace currently spread out over two matching 21-inch Intergraph 21sd107 monitors, which are sharp displays. When it comes to multimedia, I often find myself using two or three programs at a time, and every inch of screen real estate is worth its weight in diamonds.

An alternative to multiple monitors is Intergraph's spectacular 28-inch-diagonal InterView 28hd96, which displays up to 1920 by 1080 pixels in full color. It isn't cheap, but it's fantastic for high-resolution demonstrations for small groups of people (up to five), and the HDTV-like wide format is positively immersive when it comes to simulations.

I've boiled down my multimedia toolkit to a few packages I consider indispensable: Adobe ( http://www.adobe.com/ ) Photoshop 4 for 2-D, Equilibrium's ( http://www.equilibrium.com/ ) DeBabelizer Pro for file conversions, Adobe Premiere 4.2 for general video editing, Sonic Foundry's ( http://www.sfoundry.com/ ) Sound Forge for audio, and, most important, Macromedia's ( http://www.macromedia.com/ ) Director 6 Multimedia Studio. Director 6 is the latest incarnation of Macromedia's flagship product. It has been steadily improving over the years and has the largest installed base (hundreds of thous ands, at least) of any comparable product.

Coming up I'll take a close-up look at authoring with both Director 6 and a new Macromedia Web-creation tool called Flash 2 that streams full-screen Simpsons -quality animations in near real time. Take a look at Macromedia's Web site for pointers to "Flashed" sites; I think you'll be impressed.

Eric Pobirs, the Chaos Manor intern, has also been busy this month. In particular, he has been looking into developments in streaming multimedia. Here's a short report:

"I examined the Advanced Streaming Format (ASF) information on the Microsoft site. (ASF is a file format that is supposed to create a standard for streaming media-synchronized audio, video, and multimedia. It will be used on the Internet and intranets, and replace separate data types such as AVI and WAV.) I don't believe there will be a mystery codec problem. That is, a future point where the user has a high probability of being stopped from downloading a media file, because he or s he lacks the proper playback software.

"The format has dedicated metadata sections to make a file self-documenting in regard to where it came from, what codec is needed to play it, and where that codec can be obtained. This helps a lot, but there is another consideration that simplifies things even more: a market shakeout. By the time ASF becomes common, most of the lesser entries in the video playback field will be naught but memories. At the moment, streaming video over the Internet is an anyone-can-play game because the rewards are still hypothetical. The best codec technology simply doesn't overcome the fact that decent video isn't going to happen dependably or otherwise on an analog connection. Two things will likely narrow the field.

"First, inexpensive dedicated browser boxes get an installed base sufficient to draw the interest of commercial sites. In all likelihood, this translates to WebTV. Its installed base is nearing 100,000. The remaining stock of first-generation units will soon be rebated down to the impulse-buy price of $99, and the new Plus model will take over the original $300 price slot, until it too is replaced by a more powerful model. (If they keep repeating this pattern, the installed base could soon be in the millions at a very low per-unit investment on Microsoft's part. Hmm...) The one element that remains truly constant for the foreseeable future is the TV display. The codec in WebTV is designed to take advantage of NTSC limitations. Full screen is a mere 640 by 480 pixels and needs only 30 fps at best. Whoever owns the standard box for TV Internet access gets to dictate the codec for NTSC over narrow pipes. (Note: Because WebTV Plus has its own tuner built in and can overlay graphics on video as well as detect links encoded in the video, this might reverse the relationship. Rather than go to a site and play an optional piece of video, the more couch-taterish WebTV user would often watch a video and jump later to the optional Web site. Liked Indy's hat? The Paramount Store will sell you one for $44.)

"Second, fat pipes get more common and video starts to really matter, leading to a shakeout and real standards. The three necessary archival formats are already in place and ready to be used, based on the speed of your connection.

"T1 or slightly better: Use the MPEG-1 format already standard for Video-CD with almost everything in existence and already put on a disk somewhere in Asia. Anyone can play at this level. Shareware programs let you master MPEG-1 on a Pentium 133 system, and newer machines commonly provide hardware encoding. Not a format you'd want to watch !ITAL!Citizen Kane on, but plenty better than we get with plain old telephone service (POTS) connections now.

"5 Mbps or better: The not quite MPEG-2 or MPEG 1.5, depending on whom you're talking to, used for digital satellite broadcast already has an immense library of professionally mastered content to draw on. Companies who produce both DSS and WebTV boxes (i.e., Sony) could combine the two so that the WebTV end could take advantage of the DSS decoding hardware. Don't forget, Bill Gates also has this Teledesic thing happening.

"10 Mbps or better: MPEG-2, of course. Specifically, the version used in digital versatile disc (DVD) players. Still getting started, but the library is growing rapidly and the chips to let folks roll their own for cheap are Real Soon Now. The combo-box concept offered above also applies here. (Another Note: Among the new facilities in version 2.0 of Windows CE is a standard Ethernet client. The next WebTV to be announced will probably make use of this for the cable modem market. By integrating DVD playback, it could also serve as a set-top box for digital pay-per-view offerings.)"

News Item: Microsoft Internet Explorer approaches 50 percent market share. Analysts are skeptical of the claim, but Eric isn't:

"I don't doubt the claim, actually. Numerous ISPs offer Explorer as their default browser for obvious reasons (Look ma! No licensing fees!). As more new u sers enter the Internet with what those ISPs provide, their numbers cumulatively account for a greater overall percentage. Sure, some of them switch to Netscape or demand it from the beginning, but most of them simply don't know the difference, since by and large there isn't any.

"Remember when Windows 95 came out? Most of the material on the Plus pack (including Explorer 1.0) was supposed to be included, but Microsoft was forced to sell it separately due to pressure from the feds. "Apparently, this isn't an issue anymore, because NT 4.0 has the equivalent built in and Windows 98 does, too, although it still appears as Plus in some of the menus. So for quite a while now, almost every new machine sold has Explorer 3.0 and soon version 4.0 included. Some also have Netscape, but many plying the low-end consumer market don't bother. I've yet to see any new system include Communicator, which might be a strong indicator of which way the wind is blowing.

"So far, Explorer 4 is getting positive reviews. C ommunicator is still stronger in certain areas such as mail, but will Joe Sixpack see enough difference to bother? One notable strength in Internet Explorer 4: According to early reports in the PC press, it may be better at running JDK 1.1 applications than Communicator. That's kind of surprising, considering the current stare-down between Sun and Microsoft. My bet's on Microsoft. Experience has made me run the other way when people get religious over a development platform."

The news has given us more than a little Microsoft bashing lately. Even Bob Dole has gotten into the act. Some of this is mere bile. When people talk about "Windoze" and how Windows 95 crashes every hour, I stop listening: if Windows 95 is terribly slow, the problem isn't likely to be Windows 95. If you try to run it with 12 MB of memory on a 486 without a graphics accelerator, of course it will be slow.

Even with decent hardware, if you fill your hard disk with gubbage leaving no room for a swap file and then try to kee p 20 windows open at once, things will be slow and probably crash, and what else is new?

A second complaint is that Microsoft is forcing people to install Internet Explorer 4 (IE4) if they want to sell Windows 95. Oddly enough, most of those I hear complaining say they hate Windows 95 and won't use IE4 because it's horrible. This makes me wonder why they care. Apparently, a case can be made that some Microsoft practices have been in technical violation of the consent decree signed a few years ago to close the last government involvement with Microsoft marketing, but it's not clear that the case is valid. It may be; but in any event, it's nearly irrelevant.

Microsoft's opponents were demanding that Microsoft not insist on including IE4 with Windows 95, and if it is included, to notify customers that they can remove IE4.

Big deal. IE4 is free, and most resellers will choose to include it, if only because if they don't, those that do will bellow "Free Internet software with our computer!" and u se it as a sales feature.

As to the second point, how many are there in this world who don't know you can remove IE4 and also have any informed opinion on why they ought to remove it, or what to use in its place; and who could install, say, Netscape?

There may be thousands. Big deal.

It will all be moot not long after you read this. Memphis-Windows 98-incorporates IE4 as an integral part of the OS. Eric's experiments show that it's not without flaws, but it's pretty good. Some will opt to use Netscape instead of IE4, but many more will find IE4 good enough. This can't be good news for Netscape, but I fail to see how it's illegal.

Finally, we have a more legitimate complaint: Microsoft has a virtual monopoly on small computer OSes and thus ought to be treated as a monopoly.

Assuming that to be true, the question is, what to do about it? One proposal is to break Microsoft into two companies, an OS company and an application company. This seems a rather heavy-handed thing to do. I, for one, would like to have more and more applications-Word Pad, a good scientific calculator, games, a terminal program, peer-to-peer networking, disk compression (although I don't use and don't recommend disk compression), Internet Explorer-incorporated into my OS at no extra cost to me. I'd hate to see Microsoft lose the incentive to keep doing that.

One thing does disturb me. There have for years been rumors of undocumented API calls within Windows, and I know of at least one, a capability for compacting DLL and other system files even though they are in use. Microsoft says that undocumented calls are not supported and are not guaranteed to be in future editions of the product; one uses them at peril. This is a reasonable attitude-provided that Microsoft applications don't make use of such features.

I'm always afraid of government intervention in the economy. That is not to say I don't believe in antitrust actions. One of my favorite economists is David McCord Wright. He holds that one reason Marx's analysis of the future was a failure was that he didn't see that antitrust actions could and would keep the means of production from falling into fewer and fewer hands until there was virtual monopoly and no competition. It's an analysis worth thinking about.

But believing in antitrust action is not the same as saying it ought lightly to be done. Microsoft is a big success story, and that one company alone has a major effect on our balance of payments. It's important that it stay healthy-not to mention that the Constitution presumes one has a right to get rich without the government stepping in out of sheer spite.

Still, perhaps something should be done. My suggestion is that rather than bash Microsoft, the government and industry lawyers concerned about the conjunction between Microsoft application and Microsoft OS groups concentrate on requiring publication of all API calls, supported and unsupported. They should also make Microsoft agree that no Microsoft application will make use of an unsupported call without, in a timely manner, upgrading that to a supported call and publishing the upgrade information.

I can't see how that particularly harms Microsoft. It should help other application developers, it's not hard to implement, and it keeps the government's interference to a minimum.

Another time, I'll look at other antitrust action possibilities and the problem of Microsoft Java extensions. Also in upcoming columns, how to keep Windows NT from driving you mad. Stay tuned.

---

Jerry Pournelle is a science fiction writer and BYTE's senior contributing editor. You can write to Jerry c/o BYTE, 29 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 . Visit Chaos Manor at http://home.earth-link.net/~jerryp/ .