Pricing Intangibles

How do you price intellectual goods that can be manufactured and distributed for no significant cost compared to the cost of their creation? Whether you're talking software, configurations, architectures, or written works, if it can be distributed virtually without ever being placed/printed on physical media (such as a book), what is the basis of value and pricing?

Back when I started working for Computer Associates International in the late 1990's, I tried to explain to my brother why mainframe software was so expensive. After all, it generally didn't have any more lines of code than PC software that sold for a few hundred dollars at the most. At the time, I suggested that the price was generally a fraction of the cost savings that it brought.

I still stand by that assertion, but over the years, I discovered just how hard that is to prove. Once a piece of software has become embedded in an environment over many years, there's no simple way of knowing how much cost it's saving, because removing it could bring everything to a halt, which would be a completely different order of magnitude of cost.

Interestingly, pricing written works can involve similar issues. If someone asks me to write up a white paper, article, or recommendation, should I be paid by the word?

I'm reminded of a joke, a quotation and an anecdote.

The joke is about a highly-experienced mechanic who is faced with a car that has stopped working, and no one can figure out why. They ask if he'll fix it, and he agrees to for $1,000. Eventually, his price is accepted, and he goes to work.

To the external observer, the mechanic appears to be dancing around the car in a manner reminiscent of Mr. Bojangles - crouching down low, leaping up high, and almost seeming to be performing a rain dance of sorts as he looks over every nook and cranny of the car. Then, he suddenly takes out a ball peen hammer, strikes the car engine with an exacting blow, and pronounces it fixed.

The owner tries it out and, sure enough: the car now works. Then the mechanic presents his bill for $1,000.

Skeptical, the owner asks for a price breakdown. The mechanic replies, "That's $1 for hitting the car, and $999 for knowing where to hit it."

The quotation, attributed to many people, but probably most famously to Mark Twain is: "I would have written a shorter letter, but I did not have the time."

The anecdote is about the production of enameled steel pots and pans during the Soviet era. Apparently, it became harder and harder to find small ones while there was an overabundance of large ones. This was not because having larger ones somehow improved the cooking experience, but rather because the factories that produced them were incented by the amount of material used, rather than any measure of useability or rates of sales.

What all three of these have in common is that the value of things is not always connected to the simplistic, some might say "common sense," measures we apply to commodities. In fact, other than the most tightly-controlled and homogeneous commodities, I suggest size is rarely a good measure of value.

One of my favorite examples of this is the Windows operating system, which I've heard may be the largest software system ever created in terms of sheer number of lines of code. And yet, you can buy a copy for a few hundred dollars, and even get a PC thrown in (or vice versa). Compare that to z/OS, IBM's premier mainframe operating system, which costs a few orders of magnitude more than that. Yet, I doubt IBM would claim there's a commensurately larger number of lines of code; likely, there are fewer.

So, how do you price mainframe software in a way that is fair? I suppose it makes sense to take a lot of factors into account, from the capacity of the mainframe (they vary greatly in capacity), to the cost savings and benefit that a typical customer is likely to derive, to the cost of creating, maintaining, and deriving a reasonable profit from a piece of code that is only installed on a portion of the 10,000 or so mainframes in use in the world today.

That's a very different approach from pricing generic consumer software, of which millions (maybe even billions?) of copies are sold.

Likewise, when pricing the generation and delivery of other intellectual property that has great value to a relatively limited but very economically significant audience, it seems to me that paying per word is like asking for excessively large cookwear, rather than looking for that perfect touché in as few effective words as possible.

Of course, like any good question, this is more of a journey than a destination. But it's one of the important questions in the mainframe world, and one that is of particular interest to me as I look to generate written works of significant value, and don't think it appropriate to charge by the word, since I can often achieve more with a small number of well-chosen-and-aimed words than with a 10-page white paper.

What do you think about the pricing of such intangibles?