Community-Based Ratings for the Net

This document web lays out an alternative to current proposals for standards and ratings of World Wide Web documents. The objective of this proposal is to explain how, using current technology, groups of like-minded people can work together to provide more flexible, more personalized, and more comprehensive information about net resources.

The document is broken up into sections, each of which fleshes out a piece of the proposal in more detail. This document serves as the head document of the web and as an executive summary.

Overview

We propose using the technology called Automated Collaborative Filtering (ACF) as part of a scaleable, community-based solution to the problem of rating Web sites for appropriateness for Internet users of different ages, backgrounds and beliefs.

ACF, implementations of which can be seen in the HOMR and Webhound systems, enables people with common tastes to communicate on items they like or dislike. In particular, communities of like-minded people can use the technology to help each other decide what is appropriate, rather than relying on simpleminded keyword-blocking centralized censorhip.

Even though they may be geographically dispersed, these communities are drawn together by shared interests and provide for each other ratings of items previously seen. The software brings together people with similar tastes so that the ratings are relevant to the user. In addition, by developing personalized notions of similarity, ACF enables the system to make predictions about never-before-seen documents for a given user.

One obvious beneficiary of this proposal are parents who need to find some way of determining what Net material is inappropriate for their children, without relying on voluntary cooperation of information providers or relying solely on arbitrary outside standards. This technology also provides a path for self-identified communities, including children and teenagers, to engage in safer, more informed, self-guided exploration of valuable information resources.

Although most proposals speak in negative terms, suggesting means to filter out unwanted material, we should not forget the positive side. Discovery of desireable and useful Web sites is equally a problem, especially for people with limited amounts of time or who must pay for the time they spend or the documents they download. Wall Street Journal columnist Walt Mossberg describes this problem eloquently in his Personal Technology column (week of July 10, 1995).

Advantages

We believe our proposal embodies four advantages not found in other approaches:

Each of the advantages is explained in more detail in the proposal. In general, though, we feel that centralized solutions to these kinds of problem cannot succeed. A "centralized solution" is one where all authority comes from a single source and is distributed to all users. Such solutions do not scale up to the true size of the Internet and do not take into account its cross-cultural and cross-national nature. Most importantly, centralized solutions impose some arbitrary set of standards on all users. What is appropriate for Cambridge, MA may not be appropriate for Dallas, TX and what is appropriate for Riyadh, Saudia Arabia is unlikely to be what is appropriate in Anytown, USA.

In addition, centralized solutions require large investments of time and energy by businesses, by individuals, or by both. It is currently estimated at the time of this writing that there are over 4 million URLs. The ability of small groups with limited resources (such as local PTAs, neighborhood churches and the like) to participate in the process is very limited under current plans. Part of this problem is due to the proprietary nature of the proposed solutions. Each seems to require buying new, different software or relying on content providers to change their labeling and access schemes. We propose a solution which does not require buying any software beyond a Web client, and which does not depend on the cooperation of millions of information providers.

We believe that externally-imposed standards are inappropriate for these kinds of problems. The number of interested groups is huge; we feel that any group which wants to make its ratings available should be able to do so, much as anyone can now publish a Web magazine, not just Time or Ziff-Davis. Indeed, we expect there will be a burgeoning business in providing high-quality, reliable ratings.

Further, part of the reason communities form is common interests. Today we find informal communities of parents discussing the appropriateness of things such as the toys their children buy or the television and movies they watch. It seems strange to suppose that this sort of common interest will not exist for Web sites.

Many of the solutions being discussed are exclusive; if you buy into one approach you are likely to find yourself excluded from others. We believe that rating services along the lines of Good Housekeeping or Consumer Reports will emerge and that people should be able to select what ratings they find useful and what authorities they wish to make use of. Interested commercial providers like Disney or Nickelodeon should also be represented. Each of these should work in a way such that people do not have to worry about the details, just as interoperability of the phone system means that callers do not have to worry about the details of how their chosen long-distance provider works with their local phone company.

Finally, too many of the current proposals revolve around white-lists and black-lists which are binary decisions. This is insufficient for many real problems, such as a family with children of different ages, or children encountering non-pornographic, but potentially disturbing material. We prefer to see a system evolve which can address the issues of suitability in many different terms. Sexual content, violence presented, or religious viewpoint are only some of the many ways that parents may wish to restrict what material is viewed. These options should not be foreclosed.

Other Sections of the Proposal

The sections below can be read in any order. Each section has links at the bottom to take you to the next one in the sequence shown.

IMPORTANT: This proposal is not a "Media Lab" proposal; it is work by a group of people, some of whom are at the Lab, which has no official position on any of the matters discussed here.

The Authors

Many people have contributed their ideas and comments to this proposal. Any errors are the fault of the authors and should be reported to them. We also welcome comments and suggestions.

On to next section