New items will be added to the front of this web page. Old items will be dropped off the bottom after about a month.
When I first began to shave, Gillette gave away adjustable razors to further its market for razor blades. Gillette had enough proprietary protection of its razor and blade designs that only Gillette blades fit in a Gillette razor. Someone seeking to distribute and sell manufactured goods and packaging would have had to charge at least US$2-$3 for an equivalent razor.
Today we see cellular telephone services giving away phone instruments "free" in return for signing a contract for services guaranteeing payments of US$150-$250 over one to two years. The service companies realistically expect typical users to incur air time and long distance billings on the order of US$300-$500 per year per phone rather than the minimum services contracted. Someone seeking to distribute and sell manufactured electronics and packaging would have to charge US$100-$200 for technology equivalent to the "free" cell phone.
Larry Ellison of Oracle Corporation and others are touting a US$500 network workstation -- a "skinny client" machine depending on high bandwidth to the Internet and intranets in lieu of local storage and local computing capability. If such a machine appears, will it be something actually sold to end users? Or will it be given away to users by the vendor of something else to create a market?
Price points are established around stable market channels. Better and better technology becomes available at any given price point. People involved in selling high tech products for, say, US$2,000 do not wish to sell the same product next year for US$1,600. Instead they want to sell next year for US$2,000 what today costs US$2,600.
The US$500 price point is, today, occupied by warehouse-clearing specials on PC's and by the high end of the game computer market.
For a warehouse-clearing PC special in the Spring of 1996, you can find a mini-tower 486DX33 or 486SX with 4Mbytes RAM, probably no cache (beware!), 330 Mbyte hard disk, 1 Mbyte video RAM, fixed-frequency .39 dot pitch monitor, two serial ports, one parallel port, one game port, cheap keyboard, 3½-inch diskette drive, cheap mouse, and no software bundle. By the Fall, 486DX2/66 and DX4/100 machines with 8Mbytes RAM, larger hard disk, and a .28 dot pitch monitor will probably move into this price point from the US$800 price point, where they are now offered with a modest software bundle, as Pentium 75's move into the vacated $800 price point.
For a high-end game machine you can get a modern game controller for US$250, add a US$200 communications option, and spend another US$50 for a fancy joystick or yoke.
The soon-to-be-relatively-obsolescent PC offers interoperability with lots and lots of software -- but poor video performance. The high-end game machine offers superb video, access to Internet games, access to proprietary game software, and little in the way of general-purpose productivity applications.
Other price points of interest include:
Right now, corporations have enough LAN bandwidth to make the skinny client technically feasible for localized work groups up to multi-building campuses. Outlying isolated employees still find bandwidth skimpy. No doubt, many CIO's (corporate information officers) like the idea of a next generation of replacement machines for some of the workforce at US$500 rather than US$2,000. Control freaks in the CIO community also appreciate the opportunities offered by network computing to monitor everything their users are doing and to limit employees' access to safe, sanctioned applications and data.
For those selling to the corporate market, however, I foresee little enthusiasm for the US$500 network client. Margins are already razor thin for the commoditized US$2,000 desktop and US$3,000 notebook computers - the mainstays of corporate end-user computing. Wholesalers and VAR's (value-added resellers) make money by just-in-time agility in purchasing and assembly. Not many visionaries work these market channels. Getting the inexpensive skinny client machine into existing market channels will be difficult.
Might service providers (for access or content) set up new market channels to corporate customers? Yes, if they anticipate increases in their revenue. I maintain that service providers should not expect a corporate world of skinny clients to be a gold mine of new revenue. Reasons are:
Selling a highly standardized US$500 network client machine needs an entirely new market channel built by a new entrant, a company that is not competing with itself. Is there a company with far more revenue from selling server-side machines or software than client-side infrastructure? Probably Oracle, perhaps Sun Microsystems. Are any of these in a position to build the new market channel? Perhaps not. If someone tries, Microsoft, hardware manufacturers, hardware integrators, and VAR's may erect barriers to protect existing franchises; and other players will, at best, be neutral.
A broadly based consortium of companies that stand to benefit, relatively, from a US$500 network client machine might have the muscle and incentive to overcome barriers to a new entrant. But consortia are best at sanctioning something already built. Consortia are cumbersome instruments to introduce something really novel.
I question the business case for selling the US$500 network client machine to corporate customers.
Another market for US$500 skinny client machines consists of individuals. Exclude from consideration price-insensitive, risk-taking early adopters. They have already found a "weightier" client machine to play with. Also exclude people who are fascinated by the gadget as well as the services. They can see too much value in stepping up a price point to get a full-featured PC.
The remaining prospects are people who either (1) cannot afford to pay more or (2) are intimidated by technology, want a simple interface, and yet believe that a service accessible through the gadget is valuable
Either group will consider US$500 as more than an impulse purchase. For an impulse purchase, US$300 buys a 25- or 27-inch color television; 20-inch sets are US$200. Remember that US$500 for the network workstation only entitles you to pay another US$200 per year to a plain-pipe-rack ISP (Internet service provider), plus additional fees for premium content services, and something more still to get high bandwidth into the home. Head-end equipment costs a cable company on the order of US$125 per line that uses a cable modem. The cable company, if it is even interested in marketing such service, must amortize its cost and recover it from the customer. Basic rate ISDN (integrated services digital network) is available to about half of U.S. households today, but it's on the slow side of what the "skinny" network client needs to seem responsive. Installation charges, monthly charges, and usage charges for ISDN have not stabilized. Offerings around the U.S. vary from one phone company to another. Ordering up installation is far from easy. The cost-conscious customer will find ISDN to cost uncomfortably much.
Acquiring the US$500 network client really entails a commitment of a bare minimum US$1000 over the first two years of service use. Technologically timid customers may see enough value in a "killer" service to take on this financial burden - but only if the interface is as easy as using a TV remote control wand. (And using the remote control for channel surfing, not for programming a VCR!) Cost conscious customers will find US$1000 an inconvenient sum, an amount to be financed rather than something for which to write a check.
I question the business case for selling the US$500 network client machine to individuals.
I buy a single-line telephone instrument for an amount equal to one to three months of local telephony service fees. I buy a television set for 3 to 12 months worth of cable TV fees. Will I buy a network client up-front for an amount equal to 24 months of Internet access fees?
This brings us, by a commodious vicus of recirculation, back to the original theme -- giving away a razor in order to sell razor blades. It may prove too difficult to sell US$500 network client machines. Can a business case be made for giving away such machines to create new markets for services?
The razor blade case depends on proprietary differentiating technology, protected by patent. The cellular phone case depends on long-term contracting in the absence of proprietary differentiation.
The US$500 network client is touted as "open" rather than a proprietary solution. It is not clear that a machine usable only for America OnLine or only for Lexus/Nexus has appeal. Both the Microsoft Network and, more recently, CompuServe are abandoning plans to do business as proprietary horizontal content integrators. They seem to believe that micropayments on the Internet can produce subscription and per-use revenues for premium services. This leaves only America OnLine and struggling Prodigy with broad-based horizontal proprietary services. (CompuServe will keep its new- kids-oriented WOW! as a proprietary service having some restricted Internet access.) AT&T and MCI will probably bundle open data network access with bearer services, staying away from content services for the nonce.
To "give away" a network client machine in exchange for a services contract, a vendor needs the customer to sign up for services for perhaps two years at perhaps twice the price charged to a customer who owns a PC. This makes the cost of the gadget too apparent. Leaving potential customers with a finance-versus-buy-outright decision in a two-tiered service rate plan is just too complex. They could delay decisions indefinitely.
What makes more sense is some sort of partial subsidy -- perhaps zero-rate financing, perhaps covering half the cost of the machine in exchange for a service contract.
Why not simply forget the idea of the US$500 skinny client machine? The idea can be resurrected in two to three years as a US$200 skinny client machine. That would make the most business sense. However, the US$500 is explained not only by a business plan. The hubris of other CEO's wanting to deny Bill Gates his vision also looms large. For those who can't stomach the idea of computer in every home and every computer running Microsoft software, waiting may not be an option.
Sun Microsystem's Java is little beyond a novation of Sun's Oak technology. All the important technical details were already in Oak. Oak connotes sturdiness, and even nutritive value for a squirrel. Java, however, connotes action, caffeine-based alertness, or perhaps the mysterious orient -- definitely a less stodgy image than the old oak tree.
Microsoft, never willing to concede any significant software market to a competitor, answered Sun by a novation of its own. Microsoft's distributed OLE control technology, suffering from an unpronounceable and nonevocative initialization, OCX, got renamed as Active-X. Active is good, and X suggests openness even when applied to proprietary approaches.
Which brings us to AT&T, or rather the remnants of AT&T. According to the Tuesday, May 7, 1996, New York Times,
Lucent Technologies … said today that it would soon begin marketing an operating system developed by Bell Labs for use on interactive networks.
The operating system, named Inferno [emphasis added], is intended to operate either on a stand-alone basis in new computing and communications devices, or as an adjunct to existing operating systems like Microsoft Windows 95 and Windows NT or UNIX.
The article goes on to note that Inferno comes with its own programming language, called Limbo, that is designed to coexist with Java and other software.
With the help of Roget and Webster, observe that …
I always wondered about Lucent's fuzzy circle logo -- most company symbols seem, well, crisper. I now recognize that it is probably derived from the crown of thorns of religious iconography.
All of AT&T's marketing people must be reorganized out of Lucent. I don't think I could devise two product names having more negative connotations than "inferno" and "limbo." It is a shame that, what might very well be good technology, is dead on arrival because of marketing naiveté.
The information systems industry would not be where it is today had it not been for Bell Labs -- UNIX, highly transmissive optical fibers, electric battery technology. Companies that primarily develop software are in no position to challenge Microsoft's hegemony over software used in set-top boxes, PCS devices, video phones, and information kiosks up through high-end workstations. A competitive challenge to Microsoft for a seamless operating system solution can come only from the likes of AT&T or MCI. I greatly respect Microsoft and usually buy Microsoft products, but I doubt that they should get a "free ride" just because nobody at Lucent knows how to name a product.
Ten-cents-a-minute is as low as long distance rates can go. I recall that Dr. Shukri Wakid of NIST told me a year ago that the industry-average cost of line-item accounting and billing for telephony was US$0.19 per call. A one-page facsimile transmission at 9600 bps uses about two minutes -- so the US$0.20 billed for off-peak transmission of the fax pays only its accounting and billing. Leaving a voice mail message when playing content-free telephone tag also takes about two minutes on the meter, just breaking even.
The carriers don't make money on a long distance call at a dime-a-minute until you send a two-page fax or start a real voice conversation.
Sprint does a lot of advertising of its off-peak dime-a-minute rate. This is the same company that completely botched its accounting and billing when introducing its Spectrum™ (TDMA digital cellular) service in the Washington, DC area; so I really doubt that the dime lady has achieved any revolutionary productivity increase in the accounting and billing arena.
So, why shouldn't long distance minutes be free? In provisioning land lines for the U.S. Department of Defense, the cost of a transcontinental line is about equal to the cost to go the last ten miles. With this cost structure, shouldn't the long distance carriers price services the way the regional Bell operating companies price local service? The RBOCs charge for local service predominately on an unmetered subscription basis.
I know that my local phone service with tax comes to US$37.89 for two residential voice grade lines each month. I'm not inclined to spend any more time than I need to making or receiving local phone calls simply because there is no marginal cost to me of doing so. I don't really worry that somebody else with two lines may get twice as many minutes of local connect time as I do for the same price, nor do I think that anyone has a legitimate gripe for being charged $37.89 to use the phone only half as much as I.
The reasons that long distance is metered and local service is by subscription are:
Actually, as local telephone service becomes competitive, I am unlikely ever again to know that my local service comes to US$37.89 per month. That makes it too easy for me to switch to a competitor offering two residential land lines for $35.50. Instead, the rate scheme will be complicated so that service differentiation by advertising will be effective.
In yesterday's New York Times (4/30) an article appeared in the business section noting that MCI intends to introduce a consolidated billing system that bundles long-distance, paging, and Internet access in a single bill. This is a smart move. This could give MCI more agility than competitors in trades between subscription elements and metered elements. It will, of course, be marketed as a convenience to the customer.
Metered service is required to ration usage where there is excess demand for a scarce commodity. In telephony, this mostly applies to wireless bandwidth. With a given transmission power dictated by the equipment in end users hands, the service providers can only make the cell sizes so small. Although telemarketers have so far kept mostly to land lines, increased use of wireless can be expected. Since the receiver decides whether the last mile is by land line or wireless, it is the receiver that must pay and must be given economic incentive to ration usage. Sprint Spectrum's scheme of including the first minute of incoming calls within subscription pricing rather than metered air time appears to solve the problem of incoming junk calls.
Subscription service is truly needed when the payer doesn't know ahead of time the value of the particular item, only that in aggregate the items obtained through the service have value. Web browsing is an example. If I had to persuade you ahead of time that reading this essay was worth, say, US$0.35, you wouldn't be reading it. Because there are subscriptions at US$10 to $20 per month, you take a chance on reading this.
There is a gray area where metering is not required for rationing and subscription is not required because of a priori valuation. Here we can expect to see conceptually incommensurable metering schemes used as obstacles to direct price comparison. Given that this will happen, how do we get the accounting and billing cost below US$0.19 per line item? One way is to provide detailed billing on demand, perhaps with a fee attached to discourage robotic requests for itemized bills that will not be inspected. Another is to provide bills through electronic data interchange rather than printed on paper and delivered through the snail mail.
Somehow, in all of this, there should be a business opportunity for somebody to do pricing and costing comparisons to help the little guy decide among competing offers and claims.