TeleRead: Bring the E-Books Home

News & views on e-books, libraries, publishing and related topics
August 22nd, 2005

The Law of Computer Entropy

By Roger Sperberg, New York Editor for TeleRead

A story today in The New York Times talks about the decline of the educational-software market, whose producers acted as though electronicity meant they had discovered a new way to transmute gold. Now, their gold mine seems to have played out (one manufacturer now includes flash cards with its program — you know, the physical printed cards that computer versions were supposed to wipe out).

I think their failure was to recognize the effect of the Law of Computer Entropy on sales. That’s the law that states:

The market for your computer product tends towards zero over time.

You probably know this law expressed in the reverse form — Every year, when you buy software or hardware you either get more for your money or the same features or equipment for less money. As with gravity, we have learned this law from personal experience.

For twenty-five years, since we could each have our own personal computer, we’ve all experienced this law first-hand, but it has been in effect since the 1960′s and the first great transistorized computer, the IBM 360. The hardware side is related to Moore’s Law; why this is true with software I think is because of the way the cost of additional electronic copies tends towards zero. It even manifests itself on websites.

But I’m interested in this law as it affects producers, not consumers, so I state it this way — If you don’t add features every year, you subtract customers. For the educational-software producers, the law went this way: If you don’t give people bigger and better programs every year, you lose your market. Completely. They didn’t realize how being electronic changed the requirements of their product. They didn’t offer free upgrades with better graphics. They didn’t add new levels to their learning games and lower the price. They didn’t focus on adding “must-have” features to obsolete their last version. They never thought that publishers were subject to the Law of Consumer Entropy and how that law changes consumer’s expections.

Consumers — people! — know that it doesn’t cost you anything to make the thousandth or ten thousandth or millionth copy of some electronic file, whether you call it a program, a song, a movie, a game. Or a book. They know your costs are all up-front. You think they balk at rigid DRM out of idealism? It’s all economics underneath — they know when they’re being held up, and DRM only exists to steal people’s money, from an economic perspective. They want their additional features — it’s the Law, after all — and they’ve been taking them. That’s why you read about the biggest music downloaders also being the biggest purchasers. They buy more so psychically they’re entitled to more. They’re taking what’s due them in terms of extra product.

Publishers are only vaguely aware that their purchasers’ expectation of their product is about to switch from the Colgate-Palmolive type to the mp3 type, the type wherein additional features means more of the product. They have been blinded by the download side effect — My god, they can steal our books! We’re doomed! — as though all these people are going to start selling the books on street corners. They haven’t looked at the cause. But they’d better get ready to give these people, their customers, more, somehow.

Publishers could deal with this shift in expectations in any number of ways. They could act like Disneyland or a pay library: Pay once, go on any ride — er, read any book — free all day (all week/month/year). Or let libraries be libraries and meter their readers’ access to books: Up to ten thousand reads of Half-Blood Prince before renewing your license. (Next year, the price stays the same and the library gets eleven thousand reads.) Or use the bonus approach: Buy this book — and get your choice of these additional backlist books free. A publisher like Bantam churns through so many hundred books a year they wouldn’t really lose any money with a deal like this, if they picked the right titles. And of course they’d end up building the audience for those free-book authors.

It doesn’t really matter which of these the book industry prefers if the model chosen meets that one simple criteria: you can give the customer something for nothing when you have to. This shouldn’t be hard for publishers to grasp, since so much of their product has always had a rapid decline to zero worth. Sure, focus on the books that keep their value. But respect the Law.

Twenty years ago, the typesetting industry looked at PostScript and laughed at the notion that it would impact them. In the next decade more than ninety percent of those companies went out of business. The only survivors were those that changed their business model to accommodate electronic typesetting, specifically PostScript. Technology is no respecter of market position or even of industry. The consequence of becoming electronic is submitting to the Law of Computer Entropy. And only those publishers who change their business model to account for it will survive.

Digg us. Slashdot us. Facebook us. Twitter us. Share the news.
  • Digg
  • Slashdot
  • Facebook
  • Twitter
  • del.icio.us
  • Reddit
  • StumbleUpon
  • Technorati
  • NewsVine
  • LinkedIn
  • MySpace
  • Suggest to Techmeme via Twitter
  • Netvibes
  • PDF

12 Responses to “The Law of Computer Entropy”

  1. [...] Tim Bray points out this post. The first time I’ve heard of the Law of Computer Entropy. [...]

  2. “They could act like Disneyland or a pay library: Pay once, go on any ride — er, read any book — free all day (all week/month/year).”

    This is a beautiful idea. I would sign up for that service in a heartbeat. I would even pay a monthly fee for access, and keep the subscription if I didn’t get around to reading a book that month.

  3. Bryan –

    At the publisher where I’m working, which provides legal, tax, insurance and accounting documents for research, after a certain amount of editorial work the documents go online. And from that point on, the company is less like a traditional publisher and more like a library.

    I’m with you. I think traditional publishers ought to be more like a pay library and create a direct relationship with readers and offer all their books for a single fee. So what if they all read John Grisham — and increase his readership tenfold — then they’re going to go into those other titles. This is definitely a model I would go for.

    I think it’s a good deal for both parties — more money for the publisher (and author), more books for the buck for the reader. Of course, the middlemen suffer. The book distributors aren’t needed anymore and go out of business, and bookstores become less like WalMart and more like antique stores.

    And Darryl (at randomthoughts) –

    I wouldn’t say I discovered this law, nor that I’m the first to write about it. Still I couldn’t find anything written about it, so I coined the label and the expressions of it in this article.

    – Roger

  4. Pity you can’t apply that law to gasoline: buy more, get more. The trucking industry would be pleased.

  5. Roger, thanks for the feedback on my site. My comment was mostly that I had never heard the label. It’s catchy :)

  6. [...] TeleRead: Bring the E-Books Home » The Law of Computer Entropy is great post about how the Law of Computer Entropy (The market for your computer product tends towards zero over time) impacts the publishing industry. Interestingly, one of the model Mr. Sperberg suggest to deal with this is “Pay once, read any book — free all day (all week/month/year)” is already being used by O’Reilly with their Safari Bookself. [...]

  7. A d’oh comment: it never actually gets to zero unless you have a zero-install, zero-maintenance, non-evolving block of code. The code business is too diverse for your law to apply to real business.

    For any shared code (say interfaces to State or Federal systems for administrative data), the business decision is who pays initial high costs and how are those amortized across markets of variant sizes. For main product code, this is relatively easy to do. For interfaces, it is harder given that they tend to be more custom and their market is more fragmented. Too often, your spreadsheet experts identify a market and make assumptions based on it having a smooth coherent solution surface. Later as you actually sell into that market, you discover it is really a lot of little mini-markets that overlap but are distinct; in other words, the closer you get to the ground, the bumpier it is. The overlaps are actually the hardest to price because where you may think they will converge on a fixed or limit attractor, they are actually strange and continue to orbit without recrossing.

    So, the law in theory is good, but unreal in practice. You are treating code as if it were a manufacturing process, and that is an all too common mistake of naive modeling. Code management is more like farming or husbandry than manufacturing, at least, living code is. Otherwise, it is like selling packaged chickens.

  8. [...] ebooks post, very interesting [...]

  9. Len —

    You’re right: It does cost something to make that nth copy. And package it. And distribute it or serve it from the web. And support the user. And models that assume the production costs go to zero have faulty assumptions, as you point out.

    But I’m talking consumer mindset here, not producer economics. For shampoo, or gas, or paperback books, the price goes up every year (and sometimes despite that the quality goes down) and consumers accept that. They understand that it costs more to make those things and so the cost of the finished product has to go up.

    Publishers have to be aware that their customers won’t be in that mindset when they buy electronic books, but in the mindset of consumers who’ve seen prices fall, continuously, in every regard, for 25 years, or since the onset of the personal computer. (Longer than that actually, but not really widespread enough to be noticed.) I remember thinking in 1978 what a great deal I got on my computer because for the same amount of money as my friend who bought in the spring, I got 64KB of memory and not just 16KB in the fall. That was more memory than the computers NASA used to send men to the moon. And just think what you could get today for $4,000.

    And if you as a producer can’t come up with some way to give your customers additional features or additional product or something, your customers will abandon you. (Of course, I’m counting lower pricing as something “additional,” in this case additional value per dollar.)

    We don’t expect poets to write ever better poems, or musicians ever better songs, but we do have rising expectations of software coders. Who today would accept a character-based application? It might do the job as well, or faster, but what user wouldn’t expect a GUI front-end?

    No, code isn’t manufacturing, and markets aren’t homogeneous. But that’s your worry as a producer. From my perspective as a consumer, you owe me. If you’re electronic.

    — Roger

  10. Feedback cycles do not of necessity produce ever falling costs, but even if that is expected, it isn’t irresistable.

    I do owe you something and the question is is it more of the same on a faster cheaper platform, or an improved version with better features or simply emotionally satisfying in the current buying cycle. If the law of recapitulation holds, I have to give you the same thing and still sustain modifications in every subsequent version. Because the law of entropy holds for any communication system with a fixed energy budget, either the environment is becoming more sophisticated and can sustain the costs of modifications, or the environment collaspses and the market bifurcates. However, costs are not the only measure. There is a stickiness based on memory-driven patterns that create sustainable behaviors, even illogical ones.

    I have the same songs on ever fewer disks. At some point, those disks become non-playable and I have to replace even more songs on fewer disks again. As a consumer, I accept that as the cost of maintaining my cultural desires. Replacement value is never zero.

    Simulation of human intelligence based on AI has always been a tough problem because where based on logic, it fails to reckon that social bonding is based on emotional attachment, so the principle of rationality is a weak predictor of human and market behavior. We keep buying SUVs, and even if we stop for awhile as we did in the 70s, as soon as another Reagan emerges to tell us that unlimited consumption is our birthright, we’ll start buying them again just to fit in. In the short term, we will defy the energy budget and pray for it to rain oil. In the long term, we will create alternatives. Investing is a game of timing and opportunity.

    The connundrum of market behavior is the willingness of the consumer to sacrifice necessity for popularity and convenience. As I said on my blog, my neighbor is out blowing the leaves off of his lawn. The kneepads help.

  11. The Law of Computer Entropy

    Roger Sperberg talks about why computers and software (and anything that uses computers or software) gets cheaper over time. He then applies this thinking to alternate pricing strategies for the publishing industry.

  12. [...] [3] See The Law of Computer Entropy. [...]

Leave a Reply

Subscribe without commenting