As a marketing-oriented person, there's one thing about the early Borland I remember most: value. Almost every product they sold was priced at around fifty dollars (including lots of Turbo add-ons). Almost everything they sold was pretty meaty and useful. In 1984, I was a few years out of college, gainfully employed, but certainly far from flush. Even so, because Borland's products were all so cool (and so inexpensive), I'd pretty much decided that I'd collect their software. Whenever they released a new product, I'd buy it. I figured I'd always have some use for it in my library of software goodies.

While the Borland of that time wasn't just Philipe Kahn, the gregarious former Frenchman turned entrepreneur, it was still a small company. And more relevant to this discussion, its products weren't monolithic monsters, like we see so often today. Instead, they were small, slick, sleek products that accomplished just what they set out to -- and not a bit more.

Unfortunately for me as a consumer, Borland's value formula only lasted a few years. After a few years, their products moved into the hundreds of dollars and collecting was no longer feasible or fun.

In today's world of huge products like Netscape Navigator, Microsoft Office, and Lotus Notes, it's often difficult to see how smaller, single-purpose products like Turbo Pascal could survive. But it was much, much worse no earlier than five years ago. In the years before the Web, if you wanted to get your product out to people, you had to have distribution in actual, brick-and-mortar stores or you had to do direct mail with real paper mailers. Very few people in the 80's and early 90's were on the Internet. Some folks were on services like AOL and Compuserve, but there wasn't a concentration of marketing potential.

If you wanted to sell software, you had to spend serious money physically reaching customers. Even if you wanted to provide upgrades, you had to send out floppies. Today, it's hard to imagine not providing a simple bug-fix upgrade, downloadable for free. But in 1990, if you wanted to send five thousand users a free upgrade, you need to budget tens of thousands of dollars to make it happen.

Let's zoom back to today. If you could combine the charm, low-cost, innovation, and value of small, tight products like Turbo Pascal with the incredible distribution advantages of the Internet, you'd be able to see some wonderful little products.

Ah, but you can. Especially if you're building for Windows CE.

Windows CE is ideal for special-purpose, sleek products. We've seen many of these show up on the Palm Computing platform and we're seeing a nice trend toward great add-ons for Windows CE as well. The characteristics of these small applications are pretty simple. They're easy to use. They do one thing well. They're small. They're inexpensive. And they're well worth their cost.

The thing is, that with an 8MB RAM footprint on some machines, it's possible to make "big-ass" applications for Windows CE as well as the sleek wonders. From time to time, there's a good reason to use a larger, more cumbersome application (especially for special-purpose corporate applications). The challenge is to avoid the temptation of wallowing in the RAM whenever possible. If you're a programmer, I strongly advise you to design tight, design fast, design sleek. Users will love you and we'll write rave reviews.