A pocket-sized breakthrough.

This week, yet another miniaturized MS-DOS laptop--the Poquet--makes its formal debut at a San Francisco press bash. The new 8.5" x 4.25", onepound Poquet isn't the smallest mini-machine around (the casio Boss and sharp Wizard seem to be the winners in the elf-class category), and its design is probably too sober and functional to capture a big piece of the executive toy market. Nevertheless, we're optimistic about the Poquet's long-term prospects--and, perhaps more importantly, we're beginning to think that Poquet represents a genuine breakthrough in hardware design that could openup interesting new opportunities for the software world.

Clearly, the Bmall-scale market is hot. Last year, sales of laptops, portables, and hand-helds reportedly topped 700,000 units. Poquet alone (with financial backing from Fujitsu) plans to move at least 100,000 machines in 1990. This market has developed, moreover, despite products that suffer from all kinds of serious design compromises--unreadable screens, inadequate battery life, bank-breaking prices. only a Bulgarian weight lifter would call some of these machines truly portable; others have such tiny keyboards that typing even a one-page letter is an overwhelmingly frustrating task.

By comparison, Poquet's technology and design are pretty much on target. The screen is a legible 80 columns by 25 lines, the keyboard is a trifle cramped but still adequate for sustained touch-typing, the battery life (on two penlight cells, no less) is an astonishing 80-100 hours, and the base price is a realistic $1,995. We can think of a few ways the Poquet could get better, of course, but the shortcomings of the rollout version seem to be fairly trivial.

Which brings us to the software question: Why does this machine's potential success matter to software developers?

The fact is, the Poquet seems to be the only small-scale computer backed up by a true software strategy. We get a strong sense that Poquet's founders--much more than other small-scale hardware vendors--understand the key role that software plays in determining whether a new machine flies or flops. For the moment, Poquet's software evangelism effort has focused on lining up a small but carefully chosen suite of applications--Lotus 1-2-3, Wordperfect, XyWrite, ACT!, Agenda, AlphaWorks, and a few others--that legitimize the Poquet as a mobile desktop, strong on compatibility with corporate standards and well suited for peripatetic writers and salespeople.

But the Poquet people seem to understand that they have built more than just a scaled-down desktop PC. Poquet Computer hasn't said much in public about its long-range plans, but ultimately we expect that the Poquet (and other small machines) will become a primary platform for a variety of applications that are currently not part of the personal computing mainstream.

Poquet-sized machines, for instance, may turn out to be the ideal delivery medium for interactive databases and expert systems (a personalized calorie counter, perhaps, or on-line documentation for airplane maintenance, trouble-shooting tips for telephone repairmen, or estimating programs for sales reps). With the proper peripherals--bar code readers, plug-in jacks for scientific instruments, diagnostic ports--these machines also become excellent tools for remote data collection and on-site analysis. And it's even possible that small computers like the Poquet will eventually replace the pocket calculator, just as calculators displaced the classic slide rule. The crystal ball is full of possibilities like these.

Admittedly, there are a few obstacles that are bound to prevent Poquet and would-be developers from creating an instant flood of new software. Most importantly, Poquet machines don't have disk drives: software is delivered on ROM cards that take perhaps six weeks to manufacture and cost $50 each to produce. There are advantages to Poquet's choice of media: like videogame cartridges, Poquet cards can't be pirated, and users tend to get a little more speed and available memory. But adding $50 to the cost of raw media isn't a pocket-change issue for any publisher, no matter how rich. In the end, somebody--the customer or the developer--has to absorb the higher manufacturing costs of ROM-based software. (The Poquet version of 1-2-3 carries a list price of $570, which--if enough buyers bite--suggests that at least some ROM-based products may command a premium in the marketplace.)

The twin issues of price and manufacturing turnaround time may turn out to be a serious drag on software innovation for the Poquet market. We don't see how Poquet can expect to attract either the mass-market, consumer-oriented publishers or the developers of interesting utilities and desk accessories, two categories that generate much of the energy and excitement that a new platform needs. Unless Poquet and Fujitsu, which manufactures the ROM cards, get their silicon costs down, it's going to be tough for Poquet to persuade small developers to get really creative about exploring higher-risk market opportunities.

But Poquet understands this problem quite clearly--and the company in fact is already working on ways to get software manufacturing costs into a reasonable ballpark, if not all the way down to traditional floppy disk levels. once that manufacturing problem is under control, we suspect Poquet and Poquet-like machines could turn out to be one of the most interesting new platforms that the industry has seen since the introduction of the Macintosh.