First, a matter of definition. The crux of the question is what is meant by "vaporware." Stephan states that "[a]t best, vaporware is an honest product announcement that proves inaccurate ex post. At worst, it includes announcements that are knowingly and intentionally false. . . ." Since antitrust concerns must consider whether vaporware production is a strategy, focus should be uniquely on intent--and, in fact, most of Stephan's discussion relates to that. Inaccurate ex post announcements are not the issue. Still, in his article the one term covers both intentional and unintentional events, and this leads to confusion. In what follows, "vaporware" will mean the intentionally produced phenomenon, and the result of innovating firms' genuine prediction error will be referred to as "fumbleware." Stephan reaches the conclusion that vaporware will not be produced. Since antitrust law requires not only that vaporware increase market power, but also that it be (in the above language) vaporware and not fumbleware, "vaporware should not be of concern to antitrust authorities."
Yet I still suspect that vaporware is a part of the output portfolio of software and hardware firms. Vapor(fumble)ware, as Stephan quite correctly explains, includes both those products that do not meet delivery dates and those that arrive on time without "features or capabilities promised." Delivery date is only one in a set of characteristics that may not appear as promised by the firm. Keeping in mind this definition, one would be hard-pressed to deny the phenomenon in the computer industry, especially in software. In fact, it is so ubiquitous that all experienced software buyers expect and accept it. What would in another industry be a defect or evidence of incompetence is in the software industry a "bug." Tolerance for bugs is high--so high that legal action for remedy is rare, and the line between an "update" and a "fix" is quite fuzzy. Software without a fair number of significant bugs would be a front-page marvel; it is arguable that consumers of software have lower expectations, or subjectively attach higher discounts on claims, than those in any other industry.
If a massive amount of "what-could-be-vaporware" exists in this industry; the issue is, is it ever vaporware? I think it is quite possible, and the reasons can be seen by examining the industry.
All software firms are continually innovating. Innovation is not just the behavior of one firm among a set. Further, whether by technological failure, excessive optimism, or pure deceit, all software firms at one time or another miss delivery dates or manufacture and sell products that do not meet their preannounced promise. In such an environment, a firm's credibility is not likely to be completely shot by joining the fray. There might be temptation, therefore, to do it on purpose.
In addition to the presence of multiple innovators, another key characteristic of the software industry is its massive production of genuine fumbleware. Innovation inherently deals with uncertainty; it is logically impossible for an innovating firm to know a delivery date with certainty, because it hasn't yet solved the technological and research problems that are going to come up in the innovation. Anyone who has ever done any kind of research knows that the only certainty one can have about the time it takes to solve a problem is looking backward after the problem is solved. Since all firms in the industry are innovators, all firms face real uncertainty about their ability to meet a specified date with a promised package of product characteristics. Consumers who keep track records of firms' performances can tell what percentage of delivery dates are met, or try to form some index of how far the delivered product falls short of the promised one, but in the uncertain environment of technological development, they can never tell why the deviation took place. They cannot tell fumbleware from vaporware.
So why doesn't the phenomenon occur in Stephan's model? Two crucial elements are responsible. First, the model has one innovating firm; i.e., one potential producer of vaporware. Second, this firm either knows its true upgrade date or knows the true probability of meeting that date. This is an exceptionally large amount of information. Because of it, and because the firm is alone as an innovator, it cannot pass off its vaporware as fumbleware; its credibility is at stake with any announcement. (By contrast, the noninnovative competitor has no credibility issues at all; its reputation is not in question.) The nonemergence of vaporware in Stephan's model hinges upon these two assumptions.
Of course, a firm that produces fumbleware by the ton will not have a much better reliability reputation among consumers than one that produces vaporware, but given fairly high levels of fumbleware in its industry, a firm's reputation for honesty is not so much at stake. A damaged reputation for honesty is hard to repair; it stays damaged over time. A damaged reputation for technical accuracy is I think easier to repair; there is so much error to go around.
In summary, "intentionally false" is a tough nut. If a firm doesn't meet a deadline in this model, it's because there has been deliberate deceit. This avoids the horrendous issues of deliberate deceit vs. optimism. A firm's ability to make a deadline needs to be considered a probabilistic issue. There may be some interesting relationships to be teased out between the likelihood of intentional vaporware production and the frequency of unintentionally bad forecasting. It would be interesting to derive a relationship between the probability of a correct forecast and the profitability of manufacturing vaporware, which would in this context be a kind of lying by degrees, if you will.
One other assumption of Stephan's model must be discussed, but it differs from the above assumptions in that it is less global. The extent to which it is an issue in any particular industry must be examined on a case-by-case basis. The assumption in question relates to installed base and compatibility. All of Stephan's discussion refers to backward compatibility: consumers want to join the network of the preannounced product in order to benefit from the large number of users it will have.
There is no benefit here via the installed base or any other consideration for there to be forward compatibility. That is, it is just as costly to go from firm A's product to its upgrade as it is to go from firm B's product to A's upgrade. This is an essential feature of the model, since without it the known noninnovator, firm B, would never get any market share at all when the prices of the two goods are equal. But it is not a helpful simplification to make in those areas of the software industry where there are learning costs. Much standardization of software has gone on, and products are much less idiosyncratic than they were 5 or 10 years ago. Still, part of the benefit perceived by consumers in buying the preupgrade version of package A is that they expect the product to have a significant number of design features that make upgrading from A to A' easier than switching from B to A'. The two products are here stated to be perfect substitutes, but unless it is intended that they are absolutely homogeneous--which would not be helpful in considering software--some difference in switching costs would enrich the model. The profitability of vaporware may be increased by this consideration.
In summary, I think that despite Stephan's result--and I have no argument with the accuracy of the result within the context of the model--it is still well worth continuing to explore whether vaporware can truly be a profitable strategy. Stephan's model makes outstanding progress toward modeling this phenomenon, and his work highlights quite accurately that the fundamental issue is the firm's reputation over time. But I hope he will continue his work on this problem, and consider doing so in a multi-innovator model with wider technological uncertainty.
ALANNAH ORRISON, Professor of Economics, Saddleback College, Mission Viejo, CA.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Symposium: Economics of Antitrust Enforcement; response to article by Stephan M. Levy in this issue, p. 33|
|Date:||Mar 22, 1997|
|Previous Article:||Should "vaporware" be an antitrust concern?|