Dan Bricklin's Web Site: www.bricklin.com
The Value of Bandwidth
Homes and businesses will pay a reasonable amount of money for "always on" IP connectivity because they value it.
On Wednesday, December 13, 2000, I attended a meeting of the MassBroadband Advisory Board. The purpose of MassBroadband is to have businesses and economic development organizations work together to ensure that the creation of a competitive broadband marketplace throughout the state of Massachusetts becomes a reality in the near future.
Things that several of the participants said made an impression on me. For example, Elizabeth Ames, our Governor's Senior Policy Advisor for Economic Affairs and Technology (and an early VisiCalc user during her tenure as a student at Harvard Business School in the classroom above mine) reminded us how important it was for us to make sure we make our dealings with parties, like the Telecom companies, "Win-Win". Donald Dubendorf, representing Berkshire Connect (a consortium representing the western part of the state), told how important broadband connections were to some businesses. He related a story about how suppliers to large companies are often required to hook into their customer's computer systems and cannot continue business without good connectivity. Bandwidth was not a luxury, it was a lifeblood. Others talked about the problems of getting anything other than the most basic of phone service in some rural areas. Telecom companies couldn't see why companies or individuals on Cape Cod would want broadband access to the Internet according to Eugene Curry of the Cape Cod Tech Council.
I commented about how the important thing to people is IP connectivity, not the services these companies want to sell. Built on top of IP connectivity we have TCP/IP, email, Instant Messaging, HTTP/HTML/WWW, streaming audio and video, and more still to be invented. (See also: Frankston's IP Infrastructure paper, Open Access thoughts, and IP: Meaning at the end points). There was discussion of the "last mile" (from the telecom Central Office to the home/office). The savy among us recognize that it really is the "first mile", since the object isn't to send things to you, it is for you to connect out to send or get what you want. ("Last mile" is a television view of the world. The Internet is interconnected computers all sending and receiving, not passive receivers awaiting a transmission from a centralized place.)
One of the presenters put up a slide with a quote he attributed to George Gilder. It read "The Law of Telecosmic Price Elasticity: One unit decline in bandwidth price yields a five unit rise in demand." There was also an article reprint shown that mentioned how telecommunications carriers expected to make their money from bundled services, not from selling the bandwidth.
I could see why the telecom companies were unsure what to do and scared of all this. The "Price Elasticity Law" seemed to say they'd have to keep on dropping price, never making money. The old ".com" hype made them think being a portal was the road to riches.
But then, it hit me: This isn't what happens. When you get a broadband connection, the more you have it the more valuable you realize it is. Those people in Western Massachusetts found their connections more valuable the more they used them. They needed it and didn't mind paying for it. If they didn't get it, they'd go out of business. Other parts of the state wanted it badly and couldn't have it, no matter what the price.
What really happens is that there is a step function: Once you get a certain amount of bandwidth per user, you get an immediate amount of value. The "always on" bandwidth alone is of value, not any specific service. (You determine which of the myriad of things you can use it for are important to you.) You will always pay that much for a "normal" amount of personal bandwidth, just like you pay a certain amount of your expenses for food, transportation, and other things. In fact, for most people, just having "always on" IP connectivity is the best part of a cable modem or DSL. It's faster than a good dialup connection in bits per second, but the speed doesn't help that much for many things due to latency times (the waiting time for servers to start to respond). Email, instant messaging, and most web sites are fine at 56Kbps. Today, once you have 200Kbps or so per person, you only need faster speeds on a very few applications. With 1Mbps you're set for quite a while. Today's technologies seem to be able to deliver that easily, even to the home. (Well, I did mention that the problems I hear about DSL make it sound like the "punched card ballot" of connectivity, which got a laugh from all the political people there...)
What does this mean? People will pay for "some" bandwidth of real IP connectivity. They value it very much. There is a magic price point (it seems around $30 for the home, more for small businesses).The amount they pay will not drop, it may even go up some as they become more dependent upon it. Email used to be a curiosity, now it is a necessity for many people, and has supplanted answering machines for organizing your personal life and communicating with loved ones. (See What will people pay for?)
Cell phones are like this. You first buy one on a minimum plan thinking you'll only use it for road emergencies or very important work calls. Next thing you know you're on a 300 or 600 minute a month plan and call your mother just to say hi when you're waiting in line at the Post Office. You understand the value of wireless connectivity for interpersonal communications. The average cellular customer in the US pays about $30 per month.
I'm reminded of the story of UCSD Pascal on the Apple II. The Apple II came with a maximum of 48KB of memory. VisiCalc used up most of 32KB and left just about 20KB for your worksheet. The UCSD P-system version of Pascal (the portable, byte code language of its day before Java) needed 64KB to run, so they bundled it with a special plugin board that had 16KB on it and charged about $495 for the whole thing (a lot more than 16KB was worth, even than). We made VisiCalc compatible with that board, so if you had it you got "lots" more space for your worksheet. Lots of people went out and bought the P-system, threw away the Pascal stuff, and just used the board. That capability to do their numbers as they wanted was so important, they paid more than the memory was "worth". (The people selling the P-system didn't know why their product was selling so well -- they thought it was programmers loving the language and sold it off to another company for way more than they should have, as I recall.) Telecom companies that think we want their bundled services so much are making the same mistake. The connectivity itself is like the memory card. We want it for the applications we choose.
Telecom companies are worried that people (at home) will pay $30-40 for IP connectivity today and only $5 tomorrow. I think it's the opposite. They'll pay $30 today and would value it at $40 tomorrow (though telecom costs will go down, so they really can give them a bit more bandwidth at the same price and still make money). We don't want to put the telecoms out of business. It's OK for them to make money. We just want them to be telecom companies, not our only newspapers or TV channels or bookstores. We need them to understand that we value what they can provide and we'll pay if they give it to us the way we want. We want the general capability, not just today's hot service.
The cost to people of AOL has not dropped (and maybe went up) over the years, yet millions of people still think access to information, email, and IM is worth it, and continue to tell their friends to do the same. Remember the survey that said people would rather have an Internet connection than TVs or most other types of communication if they were stuck on a desert island. They are spending an increasing amount of time using the Internet in their daily lives every year as we find more uses and more people join the Net.
For quality of life and vitality of the business environment, all homes and businesses should have access to a "decent" amount of IP connectivity. That will make it even more valuable to all. For that, I'm sure that all the users will gladly pay a certain amount and continue to pay that amount for quite some time. This is a chance for a constant cash flow like local phone service.
People don't want specific technologies like "cable modems" or "DSL" or anything. They want high-speed (200Kbps-1Mbps), always on, IP connectivity whatever way you can give it to them. This is clear. It's amazing how Mitch Kapor called it years ago when he said we should forget about all these schemes to get very high speed lines into our homes for "Interactive TV" and who knows what else. "Just go with ISDN (115Kbps)," he said. "It's here today, we can use it today." "But," they replied, "a few people will pay hundreds for those lines. Why would anybody want it for $50 for just email?" The siren songs of streaming movies to homes and interactive couch potatoes, as well as the pockets of the few who were already depending on the bandwidth, just blinded everyone so they couldn't see what the Net pioneers knew instinctively and we all see now.
As PCs evolved, certain peripherals became part of the "standard" package and a certain level of performance was needed for that function. The difference between having a printer and no printer is much greater than the difference among printers. The same is true of sound cards, digital cameras (not a standard thing, yet...), and more. "Always on" IP connectivity is now another one of those "standard" things. It is part of having a personal computer. PC usage isn't going down, just sales of new ones as people spend their money on other things that connect to the PC. In many cases, IP connectivity to the Internet figures into that new purchase. In fact, for the last few years "getting on the Net" was the reason to buy a home PC.
- Dan Bricklin, December 14, 2000
After I posted the above article, I got several emails that are worth reading. Here is a sampling:
Very good points, and I enjoyed the thrust of the argument. However the focus on bandwidth was extremely painful, continuing a dangerous confusion.
You should be more upfront about "always on" being different from "bandwidth" - bandwidth is a constant, non-varying maximum capacity. Always on is minimizing one component of response time, or latency. DSL over PPOET, for example, can be configured so your computer is NOT always on, in fact it has to "dial up" an ISP from the DSLAM, which means a delay, a possibility of a "busy signal", and an inability to maintain a stable IP address.
I hate that people use the term "bandwidth" to mean "always on" or "fast response time".
If you read the FastCompany article, you understand why that confusion is not merely benign, but actually malignant - frequently leading to policies that actually result in worse response times and "always on, but never available" service.
Of course the fact that the coalition is called the Mass "Broadband" coalition is another big error. Most of the best example applications are very low bandwidth, but need to be "ready to go" in an instant. The least customer-useful are the "video couch potato" apps like "interactive TV" and so forth, because they are already well-served by broadcast cable channels.
Though I got called a "socialist" for bringing it up at Lessig's "policy implications of end-to-end" conference, I think it is worth looking at Stockholm's experience.
The city deployed "unterminated" fiber throughout the city, and allowed anyone who wanted to, to lease a fiber link for a modest cost based on its lifetime depreciation and maintenance costs. (Note that fiber continues to have value, it is the terminators that modulate it that become technically obsolete, and the city does not provide those - an example of getting depreciation accounting aligned with technology change).
The result is that a wide variety of competitive providers are delivering such services as 100 Mb Ethernet to the home or apartment, at prices comparable to MediaOne cable modems.
As long as we grant cable cos the right to completely control their markets in exchange for laying cable, we won't have much innovation-driven deployment. But if you take the argument about not wanting the streets torn up too often away, there's really not much support for monopoly positions.
- David Reed, an Internet pioneer
Bob Frankston concurs. He has problems with the term "Broadband" in that it buys into the dangerous view of the world that it isn't just IP connectivity we want but rather a cable with lots of "bands" of frequencies used for mainly cable-TV. It lends "...legitimacy to the 99% of the cable that they have kept for themselves so they can delivery 'broadband' which is unreformed CableTV." (The 1% is the frequency bands they allocate for IP connectivity, the 99% is the frequency bands they allocate for normal TV modulation.)
I also received this:
Dan, I just read this essay on your site, and I thought I would offer this observation. The dial-up access form of Internet connectivity makes the Internet a place you go to. That is, in order to access the Internet you have to do something. That effort almost makes accessing the Internet like going to the library. It's somewhere out there.
Broadband changes that because now that you are always connected the Internet isn't some place you go to. Instead, it is right there in your home and office. Once the Internet is always there you begin to think about it in an entirely different way. It becomes more a part of a normal routine. In fact, it almost becomes "appliance like."
- Frank McPherson, MCSE
Windows CE Knowledge Center: www.fmcpherson.com/knowce
© Copyright 1999-2010 by Daniel Bricklin
All Rights Reserved.