This is my attempt to speculate on the impact of widespread availability of high-speed wireless connectivity. I am not suggesting when this state of Bandwidth Utopia will be achieved. What I am trying to do is guess the impact of BU, if and when it should occur.

Fewer Phones

We have five phone numbers (one home phone number, two cell phone numbers, and the phone numbers at our two offices), and our home and my office both have phone lines dedicated to computers for dialing into the Internet. Just for local phone service, we receive four different phone bills a month--and we do not get a bill for Jackie's office phone.

Even with two phones per capita, we can have a shortage of available phones. When all of us are home, no more than three of us can be on separate calls. With two adults and three teenage girls, the limit of three calls can constitute a constraint.

With BU, I could see us cutting back to one phone per capita. We would get rid of our home phones and office phones. Instead, each of us would have our own cell phone, with a unique phone number. Cutting back to one phone per capita would mean that when I ask for your phone number you no longer rattle off a daytime phone, an evening phone, a cell phone, and a number for a pager.

Of course, when we have BU, you might not need to know someone's phone number. Your phone might access an online database to find the person's number without your having to know it. Just as I do not have to know your IP address to find your web site, I should not have to know your phone number to connect with your phone. However, the ready availability of phone numbers would give rise to attention-theft issues, which will be addressed below.

Dumber Terminals

With BU, my dumb terminal will be connected to a "home" server that performs data storage and processing functions remotely. There are servers from Citrix that have this capability today.

I want this because

• I do not want to hassle with software upgrades.
• I do not want to worry about an operating system.
• I do not want to worry about my hard disk crashing
• I do not want to have to deal with "synchronization" when I work on something other than my normal workstation.
• I do not want to lose important information or work in progress because a stupid backup battery fails while I'm not looking.
• I do not want to worry about computer viruses.

1. You want to be able to do work even when the computer network goes down.

   Well, I think that we should be able to develop enough confidence in networks so that this fear will disappear. Keeping a personal computer around in case the network fails is a bit like keeping a typewriter around in case the printer breaks. Printers do break, but by now most people have decided that this is not a contingency that justifies reverting to typewriters.

2. You do not want the IT bureaucrats to be in control, the way they were the last time we had dumb terminals, in the mainframe era.

   This concern ignores a critical difference between dumb terminals connected to the Internet and dumb terminals connected to a mainframe. With the mainframe, your IT department has monopoly power to decide what functions are available. With the Internet, you can choose any supplier of disk space and applications.

My dumb terminal would not know how to run Java or even how to browse the web. It only knows how to refresh my screen, so whatever web pages I see are those that are captured by a browser at my "home" server and sent to my dumb terminal. This could create a problem for web pages designed with client-side programs (Java, JavaScript, plug-ins, etc.). Those types of programs either must be executed on my "home" server or on the server providing the Web page.

The dumb terminal that I carry with me actually could have some pretty fancy capability. Basically, it might be an ordinary cell phone. However, whenever I have enough surface area and enough power available, I can attach devices to make it more powerful. I could add a tablet screen. I could add a keyboard. In an office or hotel room, I could connect to a high-resolution monitor. In my home, I might connect to high-quality speakers and a screen suitable for viewing movies.

Because my dumb terminal is flexible, I can easily move back and forth between text and voice conversations. I might start a dialog in email, but then my friend and I may decide to switch to voice.

Smart Attention Sentries

The fundamental error that plagues most visions of BU is the assumption that we will desire more demands on our attention. In fact, the biggest issue with BU will be our ability to control the demands on our attention.

• If I have only one phone, it is with me all of the time, then I can no longer fend off unwanted calls by leaving my office and turning on voice mail.

• If I can be located easily in phone and email directories, then I may be subject to ever-increasing volumes of phone solicitation and spam.

  All of this suggests the need for smarter attention sentries. An attention sentry will be software on my "home server" that filters my calls and emails.

  Today, some people use caller-ID or voice mail as attention sentries for phone calls. One could take this step further to create a system for prioritizing calls.

  Suppose that my telephone sentry shows me a list of ten recent phone calls and asks me to rank them in terms of priority. Calls from friends, relatives, and business associates would rank high. Calls from mass-market solicitors would rank low. The sentry can capture information about the calls, including the time of day and the id of the caller. Eventually, the sentry probably can predict that a call at dinner time from a number that I have never called is likely to be a low priority, and the sentry can screen it out.

  People will need different types of telephone attention sentries. A corporate CEO will want to screen out calls from strangers. For a doctor, it may be less important to screen out strangers but more important to conduct "triage" to help prioritize the call.

  With email, today's filters are very crude. I have to choose to assign an email to a particular bucket based on a single characteristic, such as the phrase "make money fast" in the subject line. Using this simplistic filter, I might accidentally delete an email from a friend. Even worse, I fail to delete all of the spam that uses subject line phrases for which I do not have filters.

  A better filter would be a statistical model that incorporates many factors. Currently, I delete a lot of my email without reading it, because I am confident that it is spam. Most of my unsolicited email fits into patterns that a statistical model ought to be able to detect, even though it does not match exactly any particular phrase.

  Conclusion

  Implicit in this vision of Bandwidth Utopia is the assumption that person-to-person communication will be the most important function. BU will facilitate making it easier to reach people. This in turn will require smarter sentries in order to protect people from intrusive demands on their attention.

  For computing, BU will make network computing more reliable and functional. This in turn should tend to increase the proportion of dumb terminals in use.

  What about all of the companies investing in wireless infrastructure and applications? Eventually, I believe that the most successful firms will be those that are aligned with the vision of fewer phones, dumber terminals, and smarter attention sentries. Perhaps this vision is not valuable enough to repay the billions of dollars that have been sunk. But I feel quite certain that "shopping alerts" and "stock market alerts" are never going to be billion-dollar industries.