When you delegate the tasks of mowing your lawn, washing your car, or cleaning your suit, very little privacy is at stake. By contrast, when you hand over the management of your medical, legal, or financial affairs to another human, the performance of those tasks depends on your willingness to reveal very private and personal information. While oaths and laws may protect some confidentialities, there is no real regulatory shield against the leaking of intimate knowledge by human assistants. That is achieved solely through trust and mutual respect.

In the digital world, such high regard and real confidence will be more difficult to accomplish, given the absence of actual or inferred values in a nonhuman system. In addition, a society of electronic agents will be able to communicate far more efficiently than a collection of human cooks, maids, chauffeurs, and butlers. Rumors become facts and travel at the speed of light.

Since I constantly argue in articles and lectures that intelligent agents are the unequivocal future of computing, I'm always asked about privacy. However, the question is usually posed without a thorough appreciation of how serious an issue privacy is. As many of my speeches are delivered to senior executives in fancy resorts, I sometimes announce that I have arranged with the hotel management to receive a list of the movies watched by members of the (usually) male-dominated audience in their rooms the night before. As half the faces in the audience turn red, I admit I am joking. But no one is laughing. It's quite telling, but not that funny.

All of a sudden, our smallest actions leave digital trails. For the time being, these "bit-prints" are isolated instances of very small parts of our lives. But over time, they will expand, overlap, and intercommunicate. Blockbuster, American Express, and your local telephone company can suddenly pool their bits in a few keystrokes and learn a great deal about you. This is just the beginning: each credit-card charge, each supermarket checkout, and each postal delivery can be added to the equation. Extrapolate this trend and, sooner or later, you are but an approximation of your own computer model. Does this bother you?

Hermes and hermetics
It doesn't bother me, and this is why. The data concerning whom I called, what I watched, and where I ate is not very interesting in comparison with either why I did so or any consequential information from my doing so (I liked the meal, my guest liked it, or neither of us liked it, but didn't want to admit it). The fact that I ate someplace is almost meaningless if the intent and the result are unknown. Purpose, intent, and subsequent feelings are far more important than the action or choice itself. I leave only a few digital crumbs for the direct-marketing community by revealing, for example, that I dined somewhere. The interesting data is held by the agent who made the reservation and later asks me how the evening went.

Today, marketers reverse-engineer a consumer's choice to infer why a decision was made. Advertisers cluster such demographics to further guess whether I might be inclined to purchase one soap flake versus another. Tomorrow, this will change. We can opt to tell a computer agent what we want, when we want it, and, therefore, how to build a model of us - the collective reasoning of the past, present, and future (as far as we know it). Such agents could screen and filter information and anonymously let the digital marketplace know that we are looking for something.

Two kinds of agents will exist in that scenario: one will stay at home (on your wrist, in your pocket, in your radio) and one will live on the Net, surfing on your behalf, carrying messages back and forth. To some degree, the homebodies can be hermetically sealed. They will read bit streams about products and services broadcast in abundance through wired and wireless channels. They will scoop off subsets of information of personal interest - an act as simple as grabbing a stock quote for you, and as complicated as determining your interest in a segment of a talk show. These agents will be "all ears."

Messenger agents will be more complicated. They will function as we do today when they cruise the Net looking for interesting things and people. We are at a time in history when the Net is sufficiently small for some to believe that Mosaic and other browsing tools are the only future. They are not. Even today, the people surfing the Net are distinguished by having the time to do so. In the future, there will be almost as few humans browsing the Net as there are people using libraries today. Agents will be doing that for most of us.

These Net-dwelling agents are the ones we need to worry about when it comes to privacy. They need to be tamper-proof, and we must find ways to preclude new forms of kidnapping (agent-napping). Sounds silly? Just wait until the courts begin to agonize over whether intelligent agents can testify against us.

Clipper ships
Security and privacy are deeply interwoven. The government is asking us to sail in an ocean of data, but it wants the ability to board our (Clippered) ships at any time. This has outraged the digerati and has become the object of enormous debate in WIRED and other places. I yawn. This is why.

Encryption is not limited to a single layer. If I want to send you a secret message, I promise you that I can, without any risk of anyone else being able to decode it. I simply place an additional layer of encryption on top of the data, using an unbreakable code. Such codes need not be the wizardry of mathematicians or the result of massive electronics, but can be simple but secure.

To prove this, I have put 105 rows of 12 bits on the spine of my book, Being Digital. These bits contain a message. I bet that you will never be able to decode it. If classrooms of hotshot math students want to try, be my guest. WIRED magazine will honor you at great length. But don't spend too much time. It is not nearly as easy as the title of this story: James Bond.

Next Issue: The Balance of Trade of Ideas

[Copyright 1995, WIRED Ventures Ltd. All Rights Reserved. Issue 3.03 March 1995.]