The W3C, P3P and the Intermind Patent

November 3, 1999

Lisa Rein

Last week, the W3C published an analysis from Pennie & Edmonds LLP on whether implementations of their Platform for Privacy Preferences Project (P3P) would infringe a patent held by Seattle-based Intermind Corporation.

The analysis confirmed that it would be possible to implement P3P without infringing the patent—great news for implementors and users of the new standard—but the episode raises important issues about protection of the freedom to implement open standards.

A peculiar twist to this case is that Intermind themselves were involved in the W3C working groups on P3P, and had declared during that time that they had a patent pending that may impact the new standard.


Overview of P3P
Intermind Corp.
Intermind's Patent
Analysis of P3P and US Patent 5,862,325 by the W3C's legal counsel.
P3P 1.0 Working Draft
APPEL Working Draft

P3P itself is a good example of the flexibility and extensibility that is the promise of XML and RDF. Essentially, a P3P implementation informs a user of a web site's privacy practices and allows them to control what information they disclose to a site, and how the site is allowed to use it. For a summary of P3P's features, see the sidebar "Overview of P3P".

Part of P3P consists of rules enabling automatic negotiation of privacy concerns by a client-side program linked to the browser with the web site's server. This negotiation will be performed by an agent program on behalf of the user.

It is at this point where the conflict with the Intermind patent begins: agents automatically transferring data and interacting on behalf of both and information provider and consumer.

This model is not restricted to P3P: software agents performing automatic negotiations are likely to be a prominent feature in the e-commerce systems of the near future. Thus the issue of exactly what the Intermind patent does and does not claim is of prime importance for today's XML implementors.

In this article we examine the claims of the Intermind patent, the reasons P3P doesn't infringe the patent, and review the implications for the future.

What is covered by the Intermind patent?

The Intermind patent, entitled "Computer-based communication system and method using metadata defining a control structure", essentially defines an automated communication system where a provider computer transfers to a consumer computer an object (referred to as a "communications object") containing data, metadata and methods.

The transfer of "methods" means that a server computer can transfer rules to a client computer that can be executed at a later time. For example, a rule could be sent to the client which dictates schedules and conditions for retrieval of content from a publisher. These rules control "IF...THEN" aspects of the communications relationship between a consumer and provider.

So what distinguishes Intermind's technology from prior art in this area? After all, transfer of data, metadata and program code from server to client is nothing new.

"The distinctions between other metadata technologies and ours are very real," Peter Heymann, one of Intermind's founders, insists. "What makes ours unique is that the metadata is used to define the control processes."

"For example, groupware technologies rely heavily on metadata, such as the topic names in a Notes discussion database. However, the overall system uses a fixed architecture defining how information is delivered and shared. Metadata-based channel technologies will "open this up" the same way the Web has freed us from proprietary, closed document formats."

"Many people observe that metadata is involved in many other communications technologies, and that's absolutely right. In fact, metadata is the very foundation of the Web—the tags in every web page are metadata that describe the information contained between the tags. A browser is a program that knows how to process that metadata to display and link that information."

"Our patent is focused on a new way to automate communications. Our technology relies on the transfer of metadata from an author to a subscriber in order to control the subsequent delivery of information in either direction between the two.

"What makes it unique is that the metadata is used to define the control processes. Specifically this means a process to automatically detect and transfer updates from the publisher to the subscriber, or a process to automatically deliver subscriber information back to the publisher."

The applicability of such a technology is obviously broad. In fact, Intermind registered 126 claims on its patent for this technology, covering among other things its application in conjunction with existing technologies such as digital signatures, push and pull communications, and a variety of protocols.

The point of issue between the Intermind patent and the P3P work is whether the transfer of data and metadata within a P3P implementation, and negotiation between client and server, constitutes the utilization of a "communications object" mechanism as defined by the patent.

Peter Heymann commented on the scope of their patent:

"Does it cover XML and RDF, absolutely not. Our patent position doesn't cover either one per se. XML and RDF are standards defining an extensible markup language and extensible semantics, respectively. A range of communications control structures that are likely to be built with XML and RDF will infringe, however."

The Opinion of the W3C's Legal Counsel

First of all it is important to clarify that there are really two different legal situations to consider in this particular situation. The first is whether or not P3P implementations would infringe upon the "communications object" patent claimed by Intermind. The second is whether or not the patent itself is valid; an issue that the W3C's counsel did not set out to address in the course of its analysis, although they did note"that the independent claims are rife with serious ambiguities."

The problem facing the W3C, however, was dealing with a developer public that was becoming hesitant to implement the P3P technology, due to the potential of having to deal with intellectual property complications.

The W3C's counsel first established that Intermind's patent describes "communications object" that would be construed by a court as "limited to a combination of metadata, data, and instructions organized as an object-oriented combination of metadata, data, and methods for operating on the data, to encapsulate location transparency and completely define the communications relationship between two computers in both directions via any network, protocol, or middleware."

The counsel then proceeded to argue that the P3P proposals (renamed to "policies" in the November Working Draft of P3P, but hereafter referred to as "proposals" for consistency with the legal analysis), used to define a site's privacy practices, were a static declaration that did not include anything resembling "methods".

As well as "proposals", the P3P platform includes APPEL (A P3P Preference Exchange Language). This language allows the expression of a user's privacy preferences. Furthermore, this expression may be transferred between computers, by the transfer of an APPEL file. Although such a file does constitute a "control structure" in that it dictates rules for the negotiation of privacy preferences, it cannot be construed as the control structure claimed in the patent, as it is not transferred from the client to the server.

In some cases, claims can be expanded beyond their literal terms under what is called the "doctrine of equivalents". According to the legal opinion, the amendments and arguments contained in the Information Disclosure Statement (IDS), filed by Intermind with the Patent Office, placed additional restrictions on the special type of control structure its patent defines. Anything disclaimed during any amendments is automatically excluded from any future proceedings.

Specifically, the W3C's counsel held that because Intermind had qualified its "control structure" in its IDS filing—in order to distinguish it from prior art, and thus justify its patent application—it had in fact placed significant limits on the patent by foregoing the right to assert infringement by any other type of control structure under the doctrine of equivalents.

Intermind conceded in its IDS that prior art included "control structures transferred from provider to consumer to control communications" and that their control structure was different because it was a "metadata-defined control structure" within a "communications object", capable of abstracting the instructions required for transport.

This then led the counsel to conclude:

"The use of P3P does not involve a communications object, much less a communications object that completely defines the communications relationship between a provider and consumer, or that provides location transparency. Accordingly, P3P does not infringe any claim of the '325 patent under the doctrine of equivalents."

The following table provides a summary of the various disqualifying factors which mean an implementation of P3P would not infringe the Intermind patent.

Requirements of Intermind "Device" P3P Disqualification Factor
Encapsulated in the object-oriented sense P3P files purely descriptive, declarative, static statements
Data and methods sent to client, and executed Data interpreted by client application
Two-way communication via object One way to client, one way back to server via client application
Information transfer control performed by communications object Information transfer control requires client application & HTTP
Location transparency (all information required for communication encapsulated) Static file containing data structures only, no control structures
Encapsulated data and code and processing instructions (OO object) Data only (no code, no processing instructions), not an object
Control structure "abstraction of client-server relationship completely defined" Incapable of controlling communications

Where does this leave the world of open standards?

After establishing that P3P didn't infringe upon Intermind's patent, the W3C commented that they hoped not to make such analyses "a regular practice".

A certain amount of protection for standards is afforded by antitrust regulations. Professor Pamela Samuelson, a specialist in intellectual property at the University of California, Berkeley, explains:

"If you are participating in the standards process, and you are aware that a patent is in process, and you don't give notice to people, you can't just do a sneak attack."

"If you try to use your patent intentionally to monopolize the marketplace beyond the boundaries of normal exploitation, that could get you into trouble. You have to be relatively prompt in asserting your patent. That is, you can't sit back and quietly watch till everyone adopts a technology and it becomes a standard and then expect to be able to get money from everyone who's using it because of a patent."

"Some standard-setting organizations have rules that if you participate in a standards effort, and you don't reveal that you are getting a patent in the course of the standards negotiations, and you eventually get a patent, you will waive your right to get more than a reasonable license fee for its use."

ANSI, for instance, will not standardize on a patented technology unless the owner of the patent agrees to license it on fair and non-discriminatory terms.

Could more have been done earlier by the W3C?

Should the W3C have taken action to clarify this issue earlier in the P3P process? Intermind's Peter Heymann stated that the P3P working group were made aware of Intermind's work:

"We have repeatedly asserted to the standards group the existence of our intellectual property. This is something we have taken very seriously, at the risk of being perceived mildly abrasive."

Certainly it was made clear to the membership at the Push Workshop in September of 1997, because Intermind stated it clearly and included it in writing in their presentation at the workshop. A Microsoft Word file of the presentation has been available online since late fall 1997, so the membership was notified long before the patent. Still, the W3C took no action in dealing with the situation until the patent was received in early 1999.


What does all this mean for XML programmers and implementors of W3C standards?

The good news for the W3C is that it is highly unlikely that any court of law will judge that a P3P implementation infringes the Intermind patent. This is specifically because P3P includes no "communications object", the technology that distinguishes Intermind's patented communications system from prior art.

The bad news is that, however much the W3C doesn't want to repeat this process, it is likely that an issue of this sort will arise again before long. The Internet, XML and associated technologies are fast becoming the platform on which the majority of future commerce will be conducted. Were anyone able to attain it, the advantages of control of those means of communication will be great.

If the innovation and open development processes which have so far characterized the rapid growth of the web are to continue, then careful measures must be taken to avoid the capture of open standards by individual corporations.