USB implementation is slowed by Microsoft's bugs

Electronic News, Dec 8, 1997 by Jim DeTar

The list of companies lining up to introduce Universal Serial Bus peripheral chips and subsystems is lengthy and growing. The queue actually started forming way back before Fall Comdex '96 in Las Vegas. At that conference, USB demo chips were in abundance and keyboard, mouse and monitor companies were promising to implement USB by mid-1997. They have done what they said they would do. PCs are ready for it and the peripherals are available. The problem is that, for the most part, no one can use it. And Microsoft is the villain.

Under Windows 95, PC users can use USB for only a certain, limited number of functions such as a camera on a PC. The catch is that you have to use a CD-ROM to implement the USB which somewhat defeats the Plug-and-Play intentions of USB. Under USB guidelines, up to 128 peripheral devices such as mice, CD-ROM players, DVD players, cameras, etc. can be connected to a single system. At the PlugFest in October in San Francisco, a contest was held and more than 130 devices were connected to a single PC.

USB adherents--and who isn't--have to content ourselves for now with games to see how many connections can theoretically be made, much as theologians used to sit around and speculate about how many angels could dance on the head of a pin.

What happened? Why isn't this technology which seemed to hold such wonderful promise more than a year ago being implemented?

The reason is Microsoft's seeming inability to deliver a new operating system on time. When Windows 95 was brought to market, Microsoft was chided--especially by that vocal group of technologists that have never been fond of the castle that Bill Gates built on the arcane DOS OS--for being late while it did some eleventh-hour de-bugging. Consequently, Microsoft took a timid approach with its next-generation OS saying earlier this year at the 1997 Windows Hardware Conference in San Francisco (EN, April 7; April 14) that it would not set a definite date for release of its next-generation OS, at that time code-named Memphis, and since then renamed Windows 98 when it became apparent that the OS wasn't going to get anywhere near Memphis or any other city this year.

Microsoft did say at Win-HEC that it would release a beta version of Memphis by mid-year. Beta versions have been circulated and seem to work well but the final version of Microsoft 98 will have to wait until, well until '98.

Although Microsoft has probably encountered a nest of bugs that it has to squash before it can move Memphis/ Windows 98 to consumer electronics discount superstore shelves, part of the disappointment users have can be laid on the fickleness of the development community itself. When Microsoft and Intel first proposed the earlier-introduced Plug-and-Play concept, some designers pooh-poohed the idea, saying that's a nice gimmick but we don't see any significant benefits to us. Microsoft and Intel had to beat the bushes to drum up support initially, until designers realized how easy Plug-and-Play could make things and began to push for it.

Not so with USB. Having realized the full implications of Plug-and-Play, and impressed with the possibilities of connecting up to 128 peripherals, designers began flocking to the USB banner early on. And that led to USB-mania at Fall Comdex '96. Developers at that time were promising to have peripherals out by mid-'97 in order to meet the strong demand they foresaw. Quite a switch. Now, instead of Microsoft having to drag people kicking and screaming to its new technology, they were beating down the doors to get it. But Microsoft wasn't ready to play because there are other things in Windows 98 like the Win32 Driver Model; OnNow/ACPI which will make PCs more like TVs in that users won't have to wait for the system to boot; 1394 standard communication capability and AGP (advanced graphics port); as well as WDM classes for HID, still image, streaming and audio; DirectX 5.0 and ActiveMovie 2.0; full DVD (digital video disc) support, multi-monitor capability; OSR2 (an OEM service) features such as FAT32, PCI, IRQ, Cardbus and power management and a bunch of other features that have to be integrated.

So it is the fact that there was a rush by peripherals vendors to market and hype something that didn't exist yet that caused us to build our expectations up a bit too high for full implementation of USB.

Industry observers say it WILL be fully implemented by the end of the century though not bad actually for a technology introduced this decade. After all, consider how long the market has been awaiting the arrival of videoconferencing technology. Some of the younger members of the industry can truthfully say they've been hearing about it and waited a lifetime for it. In that light, waiting another year or so for USB doesn't sound so bad after all.