"Software prices will eventually fall to zero. The open-source software movement has already started that commoditisation." That pronouncement came last week from MIT professor Michael A. Cusumano at a one-day Silicon Valley conference called "The New Software Industry".
If that sounds too good to be true to corporate IT shops that are forever squeezed by software costs, well, yeah, it is.
About a week earlier, I had dinner with open-source deep-thinker Eric S. Raymond, author of The Cathedral and the Bazaar. Raymond told me about the essay he is working on now. His conclusion, put simply: Software cannot be commoditised.
Are these guys both living in the same universe? Sure. They are just living on two different sides of Software Town.
In Cusumano's commercial-software neighbourhood, the days of the software package with a stranglehold on its market are gone -- and the reason is open source. Microsoft and Oracle may swallow their commercial competitors, but they cannot buy out the persistent competition of free open-source software.
The result: Software is becoming a commodity, and prices will collapse. Cusumano thinks the only way software companies can survive is through services -- either selling software as a service or offering add-on services along with their software products. So we will still get squeezed, but for services instead of for software itself.
It is hard to see how Cusumano could conclude that software prices will have to drop to zero -- especially in an economy where clever marketing can sell a bottle of tap water for $1.39 (£0.70).
Still, software is becoming a commodity -- isn't it?
Across town in the open-source neighbourhood, Raymond says no. Open source is not commoditising software, he argues -- just modularising it.
Software is not like hardware. After 200 years of industrialisation, we understand the value of commodity hardware. We want standard nuts that fit on standard bolts, standard tires that fit on standard wheels and standard memory that fits in standard motherboard sockets. Interchangeable parts introduce manufacturing economies of scale, while custom pieces don't add enough value to be worth the trouble.
But software is far easier to adjust than hardware. Small tweaks can suddenly make software far more useful to some customers, but without the expensive retooling that hardware requires. The cost of differentiation is small, the value high. That makes software nearly commoditisation-proof.
Instead, Raymond says, open source is forcing software to break up into modules -- stand-alone chunks of software that can be plugged together.
Modules do not have to be identical -- just act alike. So as long as the interfaces are standard and the functionality matches up, an open-source module can replace one that is proprietary.
And unlike hardware, software modules do not have to be physically in the same place to connect up. You do not need Google's software on your servers to create a Google Maps mashup. You do not need to know where Google's software is, and the mashup benefits from Google's ability to update the maps and the engine -- so long as the module keeps the same interface.
As Raymond told me, "Modularisation makes mashups possible".
What he did not say is that modularisation opens the door just as wide for proprietary software as for open source. Sure, an open-source module can replace a chunk of proprietary software -- or, instead, modules from several different vendors can compete for the same spot, each with its own distinctive advantages for some users.
That would mean no zero pricing, but maybe a lot more value for IT shops. It would mean no more one-size-fits-all, but a much better match to what our businesses need if we really can plug together the right pieces from proprietary, open-source and software-as-a-service sources.
And, just maybe, it would mean there is still money to be made in Software Town.