Digital Reputation Principles

This article appeared as my column for Connect Magazine in December 2006.

Most online interactions are devoid of many of the cues that people use in the physical world to make judgments about the character, stability, and reliability of the people, systems, and other entities. Yet these cues are critical to sophisticated interactions and transactions.

Related to that idea, your online identity isn't something that accumulates over time--becoming more and more trustworthy as it ages. Online persona are fragmented; there's no reliable way for me to know that you at eBay is the same you at LinkedIn, for example. Whatever credibility you've built up at eBay is non-transferable.

Of course, online reputation is often accessed informally by people using their experience on the Internet. For example, I may not trust email I get from the domain because I've received substantial Spam from addresses in the domain in the past.

On the other hand, formal systems can collect and process online information in an attempt to calculate reputation scores in a more algorithmic manner. We call such systems reputation systems. The results from a reputation system are used by other systems that need to assess past performance of a person or other entity in making decisions about their likely future performance or their intent.

I'm convinced that over the next several years online reputation systems will be built to help facilitate important transactions on the Web and that managing your online reputation will be as important to your financial and social well being as managing your reputation in the physical world is today.

There are some simple online reputation systems that have probably affected how you use the Web already. One such system is Google's page rank algorithm--the system that determines the order of search results. You can interact with this reputation system indirectly, by performing a search, or directly, by asking Google using it's Web services for the rank of a specific URL.

Another online reputation system that you may have used is eBay seller and buyer ratings. These scores let other eBay users know what kind of experience other buyers and sellers have had with that user.

Neither of these systems is general purpose--their reputation scores are limited to their specific domains. Yet they inform our understanding of how a more general purpose reputation system ought to work.

Over the past year, I and others have been working on a set of principles that we believe should govern the operation of general-purpose, online reputation systems. Here are a few of the more significant ones:

Reputation is someone else's story, not yours. I can't control what you say about me although I may be able to affect the factors you based your story on. General purpose reputation systems should let each user compute their own story about someone or something based on various factors.

Reputation is based on identity. Reputation isn't part of your identity, but is based on it. Reputation requires linking identities together.

There is a natural tension between reputation and privacy. Privacy depends on decoupling identifiers, while reputation relies on linking them together. Think of your credit score that enables you to borrow thousands of dollars based on your financial reputation. That benefit comes at a high cost to your privacy. Reputation systems should be built in such a manner as to afford people as much control as possible over how their identities are linked.

Reputation exists in the context of community. Any given context will have specific factors for what is important in determining reputation. This is different than saying "communities have a reputation about someone." Communities do not have beliefs, only people have beliefs including beliefs about what others believe.

Reputation is a currency. Reputation can be used as a resource. Reputation researcher Paul Resnick and others have shown that a positive eBay reputation has real economic value.

Reputation is narrative. Put another way, reputation varies with time. Reputation is dynamic because the factors that affect it are always changing.

Reputation is based on what you believe others believe. A reputation isn't just based on facts, but is also based on what other people think about the subject of the reputation calculation. These beliefs are signaled to others in various ways depending on the context.

These reputation principles guided the design and implementation of a pilot reputation system we built in my lab at BYU. While we view these principles as universal, we recognizing that there is still room for discussion on the particulars. I'd like to hear your thoughts.

Phil Windley teaches Computer Science at Brigham Young University. Windley writes the popular Technometria blog on technology at and is the Executive Producer of IT Conversations, the Internet's premier technology podcast. Contact him at

Last Modified: Tuesday, 31-Oct-2006 09:21:56 MST