I didn't want to believe it, but it's true: No good deed goes unpunished.

Almost a year ago, I wrote a really nice review of Expert-Ease, a program written by an old college professor of mine. Although rather steeply priced, Expert-Ease was -- and is -- very good software for developing "expert systems." Expert systems, a branch of artificial intelligence software, enable computers to emulate the way experts solve specific problems.

Well, despite the punny name, Expert-Ease fizzled like cheap antacid. Now Human Edge Software (of Mind Prober Fame) has acquired the marketing rights to the program.

Similarly, I penned a rave review of Lightyear Inc.'s Lightyear -- an expert system-like program designed to help managers understand their own decision-making priorities. Despite the miserable name and the $495 price tag, I thought for sure this product would be a winner. Indeed, Lightyear got glowing reviews industrywide.

Imagine my (embarrassed) astonishment upon discovering that Lightyear has sold less than 5,000 copies of its program and is struggling to survive. The hot speculation in software circles now is that the company soon will be acquired by Ashton-Tate, the makers of DBase II and Framework.

Think about that for a moment. Both companies, Lightyear in particular, got widespread and favorable media coverage. Let's focus on Lightyear. Figure that there are at least 3 million personal computers out there capable of running Lightyear -- an innovative product by any measure and one with no discernible competition. Even if only one-third of the personal computer owners might have been interested in Lightyear, the program sold to less than 1 percent of its potential audience.

So much for market penetration . . . so much for the power of the press. But the market has spoken. Both products were -- commercially -- unqualified failures.

Why?

Were the prices too high? Were the products too early? Were the marketing plans too ambitious? Could expert systems fundamentally be products without markets? Are these even the right questions in the current messy state of the software market?

Well, the products aren't dead, but maybe this is the time for a preemptive autopsy.

Although I like the Expert-Ease/Lightyear expert systems genre, I've reluctantly come to the conclusion that they do a superb job of providing a solution for problems that don't yet exist. These programs really are tools that allow people to create their own expert systems. I thought there was a market for tools. In reality, it appears that personal computer users are less interested in the tools than in the expert systems themselves. That's particularly thorny because the demand for expert systems has yet to be established.

To put it another way, people don't want hammer, nails and wood -- they want the finished house. Or to pick a more appropriate analogy, PC users don't want to build their own Lotus 1-2-3; they want Lotus 1-2-3.

Speaking with the benefit of hindsight, the Lightyear/Expert-Ease folks made an incredible marketing blunder. Instead of selling these wonderful tools in the mass market and expecting people to appreciate their power, they should have started off by focusing on specific applications. They should have found a few specific companies and a few specific industries -- such as law firms or marketing departments in consumer goods companies -- and seeded them with the software while providing hand-holding support. The results might have been powerful successful examples of how these tools could be profitably applied. Those successes could have generated word-of-mouth popularity for the programs. Remember, word of mouth, not advertising, is how VisiCalc, WordStar and Lotus climbed to super software status.

So what's going to happen to these kinds of programs in the interim? I think we will see them tailored to interact with other programs. The reason Ashton-Tate is interested in Lightyear is that the program could be applied as an intermediary agent that would enable the user to interrogate his data base. In other words, an intelligent front end that can be programmed to search out the desired information in a way far easier than the existing DBase II currently permits.

Admittedly, those are only interim steps.

Call it arrogance, clairvoyance or stubborn stupidity but, ultimately, I think expert system tools will become popular as people appreciate the importance of being able to codify intelligence and expert advice to solve problems. Don't ask me when that might be. I've already been wrong on that.