Coalition for Networked Information 
Information Policies:  A Compilation of Position Statements, 
Principles, Statutes, and Other Pertinent  Statements


Nelson's Canons

Theodor H. Nelson
Microsoft Corporation
1 Microsoft Way
Redmond, WA 98052-6399
206-882-8080

Source:  Nelson, Ted.  "Nelson's Canons:  A Bill of Information 
Rights," electronic file.
A Bill of Information Rights

Easy and Arbitrary Front Ends.
Smooth and Rapid Data Access.
Rich Data Facilities.
Rich Data Services Based on These Structures.
"Freedom from Spying and Sabotage."
Copyright.

1. Easy and Arbitrary Front Ends.

The ten-minute rule text must move.

2. Smooth and Rapid Data Access.

Freedom of roving.

3. Rich Data Facilities.

4. Rich Data Services Based on These Structures.

A. Anthological Freedom.
B. Step-out Windowing.
C. Disanthological Freedom.

5. "Freedom from Spying and Sabotage."

Freedom from being monitored.
Fiducial system for telling which version is authentic.

6. Copyright

It is essential to state these firmly and publicly, because you are going to see a lot of systems in the near future that proport to be the last-word cat's-pajama systems to bring you "all the information you need any time, anywhere." Unless you thought about it you may be snowed by systems which are inherently and deeply limiting. Here are some of the things which I think we all want. (The salesman for the other system will say they are impossible, or "We don't know how to do that yet," the standard put down. But these things are possible, if we design them from the bottom up, and there are many different valid approaches which could bring these things into being.
These are rules, derived from common sense and uncommon concern, about what people can and should have in general screen systems, systems to read from.

1. Easy and Arbitrary Front Ends.
The "front-end" of a system -- that is, the program that creates the presentations for the user and interacts with him -- must be clear and simple to use and understand.
"The Ten-Minute Rule." Any system which cannot be well taught to a laymen in ten minutes, by a tutor in their presence of a responding setup, is too complicated. This may sound far too stringent; I think not. Rich and powerful systems may be given front ends which are nonetheless ridiculously clear; this is a design problem of the foremost importance.
"Text Must Move," that is, slide on the screen when the user moves forward or backward within the text he is reading. The alternative, to clear the screen and lay out a new presentation, is baffling to the eye and thoroughly disorienting, even with practice.
Many computer people do not yet understand the necessity of this. The problem is that if the screen is cleared, and something new then appears on it, there is no visual way to tell when their new thing came from: sequence and structure become baffling. Having it slide on the screen allows you to understand where you've been and where you are going; a feeling you also get from turning pages of a book. (Some close substitutes may be possible on some types of screen.)
On front ends supplied to normal users, there must be no explicit computer languages requiring input control strings, no visible esoteric symbols. Graphical control structures having clarity and safety, or very clear task oriented keyboards, are among the prime alternatives.
All operations must be fail-safe.
Arbitrary front ends must be attachable: since we are talking about from text, or text-and-picture complexes, stored on a large data system, the presentation front end must be separatable from the data services provided further down in the system, so the user may attach his own front-end system, his own styles of operation and his own private conveniences for roving, editing and other forms of work or play at the screen.

2. Smooth and Rapid Data Access.
The system must be built to make possible fast and arbitrary access to a potentially huge data base, allowing extremely large files (at least into the billions of characters). However, the system should be contrived to allow you to read forward, back, or across links without substantial hesitation. Such access must be implicit, not requiring knowledge of where things are physically stored or what the interest file names may happen to be. File divisions must be invisible to the user in all his roving operations (freedom of roving): boundaries must be invisible in the final presentations, and the user must not need to know about them.

3. Rich Data Facilities.
Arbitrary linkages must be possible between portions of text, or text and pictures; annotation of anything must be provided for; collateration should be standard facility, between any pair of well-defined objects: "Placemark" facilities must be allowed to drop anchor at, or in anything. These features imply private annotations to publicity- accessible materials as a standard automatic service mode.

4. Rich Data Services Based on These Structures.
The user must be allowed multiple rovers (movable placemarks at points of current activity); making possible, especially, multiple windows (to the location of each rover) with displays of collateral links.
The system should also have provision for high-level ?mooting and the automatic keeping of historical trails.
Then, a complex of certain very necessary and very powerful facilities based on these things, viz.:

A. Anthological Freedom: the user must be able to combine easily anything he finds into an "anthology," a rovable collection of these materials having the structure he wants. The linkage information for such anthologies must be separately transportable and passable between users.
B. Step-out Windowing: from a place in such an anthology, the user must be able to step out of the anthology and into the previous context of the material. For instance, if he has just read a quotation, he should be able to have the present anthological context dissolve around the quotation (while it stays on the screen), and the original context reappear around it. The need for this in a scholarship should be obvious.
C. Disanthological Freedom: the user must be able to step out of an anthology in such a way and not return if he chooses. (This has important implications for what must also really be happening in the file structure.)

Earlier versions of public documents must be retained, as users will have linked to them. However, where possible, linkages must also be able to survive revisions of one or both objects.

5. "Freedom from Spying and Sabotage."
The assumption must be made at the outset of a wicked and malevolent governmental authority. If such a situation does not develop, well and good; if it does, the system will have a few minimal safeguards built in.
Freedom from Being Monitored. The use of pseudonyms and dummy accounts by individuals of files that also exist on public systems, it is necessary to have "fiducial system for telling which version is authentic." The doctoring of online documents, the rewriting of history--cd. both Winston Smith's continuous revision of the encyclopedia in Nineteen Eighty- Four and "The White House"-- is a constant danger. Thus our systems must have a number of complex provisions for verification of falsification, especially the creation of multilevel fiducials (parity systems), and their storage must be localizible and separate to small parts of files.

6. Copyright
Copyright must of course be retained, but a universal flexible rule has to be worked out, permitting material to be transmitted and copied under specific circumstances for the payment of a royalty fee, surcharges on top of your other expenses in using the system.
For any individual section of material, such a royalty should have a maximum: i.e., "by now you've bought it."
Varying royalty rates, however, should be the arbitrary choice of the copyright holder: except that royalties should not varying sharply locally within a tissue of material. On pubic screens, moving between areas of different royalties cost must be sharply marked.

info@cni.org