While software vendors may occasionally make policies up as they go along, they generally make the reasonable assumption that their customers don't particularly want to hear about them.
Which is why it is so surprising to hear so many of them publicly debating how they plan to license and charge for their software in future. "I can't say where we will get to in three years. We have to experiment," says Stephen Udon, head of partner marketing at Microsoft.
"We're going to try a number of different licensing models - all different types," says Mark Stradling, Citrix's managing director for northern Europe. "It's overkill, but we'll get a much better idea of what users want."
And they are not alone. Dale Vile, a senior analyst with Bloor Research, explains: "No one really knows what they are doing. Look at a new IT project going into any company today. They are looking at five or six sets of different licensing deals for the software they need."
The cause of all this confusion? "The internet," says Udon. "The internet," says Stradling. "The internet," says Vile. It appears that the net has turned licensing, previously a dull-but-lucrative pastime for software vendors, into an experiment in how to survive.
The net effect
On the one hand, the web alters the sheer number of customers that use their software. Instead of a countable number of users connected to a local area network, there are innumerable hordes of partners, customers and home-workers dialling in through the net.
And this affects the way in which customers want to buy their software: instead of purchasing boxed applications from a reseller, they want to buy licences in bulk by filling in an online form.
On the other hand, the internet also affects what users are prepared to pay. Instead of forcing customers to pay the fees indicated in country-based price lists, which bear little relation to the prices paid by large customers, the internet promises global pricing with a transparent discount structure.
The first problem particularly worries Microsoft and led to its announcement in June of per-processor pricing for its Windows DNA 2000 products. These include its SQL Server 2000 database, Commerce Server 2000 and the delayed BizTalk Server 2000, which will enable users to exchange ebusiness documents.
The change means that instead of buying a server licence and an additional client application licence (CAL) for each user, customers will now pay according to the number of processors that the server hardware runs on.
But per-processor licensing is an experiment, Microsoft says - albeit one that IBM and Oracle are also trying. So far, it is simply an option on a subset of the company's new products.
If the scheme proves workable, however, it will be extended in the next few months to cover far more offerings, although Microsoft admits that it does not know whether it will work.
Bloor's Vile explained: "It's crude, but it's the only sensible way to do it. It cuts through all the variables. You can budget for it. Financial directors love it because it is predictable."
It is a lot easier for IT managers to count the number of processors in their IT department than try to estimate the number of potential users of an internet-based application, he added.
"What we're seeing from the internet is that companies are not only giving access to their systems to their trading partners, but also to ordinary people out there in consumerland. The numbers of users of any application is much less predictable," said Vile.
While Microsoft's Udon accepts that counting processors only gives a very rough idea of how 'useful' a piece of software is, it is the only workable idea that the supplier has found to date.
"The CAL in the internet world is clearly nonsense," he says. "Some people have been able to run very large internet sites for license costs of a few thousand dollars, so we're not sure yet how we capture that value. But it certainly isn't by using our traditional licensing models," he says.
Made to measure?
Ideally, Microsoft would love to charge 'per transaction' - the more customers use its software, the more they pay. But even if customers could get used to the idea of a software licence fee that could not be predicted in advance, usage would still be too difficult to measure until customers were in a position to rent software from application service providers.
Also, when selling software licences is a company's only source of revenue, the risks of getting its sums wrong are considerable.
"Anyone running a licensing programme is making a trade-off between what they want to do, and what it's possible to do," Udon admits. "It's not a great area for shooting from the hip. The harder you make licensing to administer, the more blood customers have running out of their ears."
And Citrix's licensing experiment has left shareholders with blood running out of their ears, following a profit warning that sliced expected profits for its second quarter, which has just ended, in half. The move even prompted the resignation of the firm's chief executive.
Last October, Citrix announced that customers wanting to purchase its software could buy licences in bulk from resellers over the internet instead of purchasing boxed products.
Today, it says that three out of 10 licences are being sold this way, leaving distributors with large inventories. But the company admits it had no idea this was going to happen. "We wanted to make it easier for them to buy the product. But it was a surprise to us the scale at which our internet licensing increased," Stradling says ruefully.
Undaunted, the supplier is now about to launch several new bulk licensing schemes. Some customers will pay a one-time cost, while others will pay according to how often they use the software. Others can choose to pay per user.
Stradling admits that it is a massive experiment. "I'm not sure which ones will work," he says.
A turn up for the books
Oracle, on the other hand, is trying to teach its sales force - and its customers - how to use a global pricing model and a formal discount structure. In effect, the company has rewritten its sales manual after a quarter of a century of deal-making.
Stephen Millard, product marketing director for Oracle UK, says: "The days of some of the hard-nosed negotiations that our salesmen probably enjoyed have gone. It certainly focuses their attention because they just don't have that flexibility any more. Customers have worked us as well. They know the game they play. But that game is no longer there. They will know up front how much we will charge them for the licence."
Oracle's chief executive Larry Ellison last year promised to introduce a global price list, using the internet to ensure that pricing is clear and there is a single sales point for software.
So far this idea has proved easier to talk about than to implement. As Bloor's Vile points out, the UK sales force still advises potential customers to use the price list as a 'rough guide'. But, Vile adds, he will be surprised if a formal, global pricing structure is not operational within 12 months.
Researcher IDC predicts that revenues from ecommerce applications will boom from $1.8bn in 1999 to $23bn in 2004. And the company believes that the sales explosion will occur in innovative licensing areas such as pay-per transaction, renting applications over the internet, advertising sponsorship and shared revenue models with business partners.
This implies that the changes many software vendors are making today are simply a first step towards this goal. But suppliers also know that a single mistake in licensing policy can be far more damaging to their revenues than releasing any number of buggy products or missing endless deadlines.
So whether it is Oracle, Microsoft, Citrix or any other software vendor, the dull world of licensing is about to be shaken up. But will more companies be issuing profit warnings in future as a result? "It depends on whether they learn from our mistakes," says Stradling.
Some parts of Atacama have not received rainfall for 500 years - but a sudden deluge of water upset the Desert's delicate biological balance
Spitzer Space Telescope could not spot Oumuamua, suggesting that it is actually pretty small
Greenland crater one of the 25 largest impact craters on Earth
This long-sought progenitor star was identified in an image captured by Hubble in 2007