algebraist's Journal: Shrinkwrap Revisited

I've written elsewhere about the sea change that's come over the software industry, principally in a move en masse to "shrinkwrap" software for even enterprise scale applications. This benefits both customers and the software developers because the cost of developing and maintaining the software is amortized over N customers. If N is sufficiently large, in terms of immediate cost this is a clear win. Evidence is available that this is a much bigger factor causing the shrinking labor market for software developers than, say, competition from overseas developers or H1B folks. For one thing, even India is experiencing a slowdown. (This reference is available upon request. It was originally on Reuters.com, but they have since pulled the article and I am not aware of any Reuters archive.)

I understand entirely why businesses are doing this. In a nearly deflationary economy, with major sectors already experiencing deflation, the best financial option is to hold cash. It will be worth more tomorrow.

In information technology this deflation is aggravated by the continuing improvement in processor and memory performance per unit cost, meaning demand is down for everything from semiconductors to replacements for software that exploit higher performance hardware. But, like most things in life, we're learning there are other costs, less immediate but real nonetheless.

First, just like costs are 1/N of what they might otherwise be if a business wrote their own software, it's sensible that the vendor's sensitivity in terms of response to problems or requests for customization at a customer site is also 1/N of what it might otherwise be. Actually, it is probably less since the more a particular customer demands fixes or customizations without paying for these the less valuable that customer is to the vendor of the shrinkwrap. It is the most profitable customers, the ones who can use the software just as is, who are the most valuable.

Second, and more recently, centralization on a handful of shrinkwrap applications incurs legal and other risk. I am, of course, speaking of the PeopleSoft-J.D.Edwards-Oracle battle and of SCO-versus-the-world, notably IBM.

In the case of Oracle's predatory attempt to take over PeopleSoft, when the takeover was first announced, they loudly they would discontinue the PeopleSoft product line should they succeed. Worse, they said they'd fire PeopleSoft's employees. This message is actually being focussed upon PeopleSoft's customers. Mr Ellison must want to damage PeopleSoft's market position even if he does not acquire the company. PeopleSoft's customers must now feel they made a poor choice in going with PeopleSoft, but the same outcome might await them if they went with anyone's pre-packaged solutions, including Oracle's.

These clients now base their central business processes depend upon that solution. It is in danger of disappearing and, even under the best of circumstances, should Oracle succeed the client's own customers are likely to suffer poorer quality of service during the changeover.

One might blame Oracle for this situation, as the management of PeopleSoft and J.D.Edwards are clearly trying to do with their lawsuits, and it is odd that Oracle is essentially punishing customers for going with PeopleSoft, but actual responsibility also falls on the decision to use shrinkwrap for essential business capabilities. The forces which are driving companies to outsource are the same which are driving consolidation and acquisitions. Mr Ellison of Oracle might say this is inevitable with his pronouncements about the end of software as we know it. But despite his opinions, Oracle isn't immune from these tactics either.

In the other case, that of SCO going on a rampage to renege its licenses for everyone to use Unix, also poses a risk, however less likely the SCO suit is to succeed than Oracle is to buy PeopleSoft. The idea of IBM's AIX being suddenly unavailable to thousands of businesses must have their CIOs have waking nightmares. It is also a price of centralization. IBM may have pockets deep enough and legal talent strong enough that settlement is the most likely outcome, however far off in the future that settlement might be. But there will be costs in that case, and I cannot see IBM doing anything but somehow passing these onto users of AIX. There are, too, also many other licensees of Unix who may not be as robust as IBM and who really will be damaged along with their customers.

In summary then, there are some pretty ugly possibilities facing businesses who have consolidated their software functions in an ERP like PeopleSoft. If common sense has its way, the markets, the government, and the courts will see the harm which the Oracle predation and the SCO litigation frenzy would do if they succeeded. But, then, this is management trusting the future of their businesses to an outcome determined by greedy and fickle markets on the one hand and a justice's whim on the other.

No, I don't think I'd like to be a CIO right now, certainly not one managing an ERP solution. Imagine if someone tries to do something to Oracle?