A Credibility Model for Antivirus Industry Self-Regulation

Joe Wells

WildList Research Corporation. USA

About the Author

Since 1991 he has worked as designer and/or developer on several anti-virus products (Certus, Novi, Norton AntiVirus, IBM AntiVirus, IBM Digital Immune System, VFind, and Wave Antivirus) and several automated virus analysis systems. Wells is best known for his WildList, which is a list of viruses verified as being a real-world threat, that has helped standardize virus naming and product testing. Wells is recognized as the leading authority on viruses that constitute an actual threat and is often quoted in the press. His work has been profiled by the Los Angeles Times and ABC Evening News.

Wells is an emeritus member of CARO and is on the Virus Bulletin advisory board. He is currently CEO of WildList Research and COO of WildList Organization International, Inc.

Mailing address: WildList Research, 1840 Calvada Blvd. Suite 9, Pahrump, NV 89048 USA. Telephone +1 775 537 6112. Fax: 1 775 537 6115. Email: joe@teknett.com.

Descriptors

WildList, Persuasion, Dialogical Ethics, Marketing, Product Claims, Alerts, Press Releases, Communications Science, Critical Thinking, Credibility Criteria, Credibility Rating

A Credibility Model for Antivirus Industry Self-Regulation

Abstract

The purpose of this paper is (a) to present a credibility model for the antivirus community and (b) to propose an implementation that allows the antivirus community to improve and then regulate its credibility. The industry’s current credibility is discussed. The concept of credibility is expanded and it is demonstrated how credibility can be mapped, and how specific credibility factors can be isolated, analyzed and evaluated with a view to establishing a credibility rating based on information coming out of any given antivirus organization. The factors going into a credibility model and tabulated as they are presented throughout, and then combined into a comprehensive model. However, only two portions of the model can be easily formulated into credibility criteria. It is demonstrated how these sets of criteria can be analyzed and applied. Based on model, a practical implementation of a credibility rating system for the antivirus industry is proposed, explained, and potential problems are addressed. Two supportive appendices are attached. One explains the elements of persuasion and the other sets forth principles of discussion to discussions about a practical implementation of the model.

Literature Review

Developing and implementing a credibility model, analyzing company claims, and giving companies a credibility rating is not new. It has been done. For example, the application of a credibility rating can be found in an online review of several magazine reviews of antivirus products. The website rates the credibility of each magazine’s review and applies a relative numeric value. It also provides the criteria by which each credibility rating was determined. (ConsumerSearch, 2000)

The above example demonstrates the feasibility of implementing a credibility rating system. Thereby substantiating the proposal of this paper as outlined below.

However, application of a credibility model to the antivirus industry itself is, to my knowledge unprecedented. A somewhat related application of such a model was used to rate antivirus product reviews. That is, the separate reviews were compared and rated on a set of criteria. While this does not directly apply to the antivirus industry, it does demonstrate that such application of a credibility model is both doable and practical.

Credibility in antivirus issues was addressed in my article "Lies, Damn Lies, and Marketing (Credibility in Virus and Antivirus Issues)" (Wells, 2000), which provides tools for users to gauge the credibility of claims, including alerts, marketing claims, news stories, hoaxes, etc.

Credibility itself is widely documented as a factor in persuasion theory, which is a branch of communications science.

An landmark article on the analysis of source credibility by application of the cognitive response model appeared in volume 10 of Progress in Communication Sciences (Benoit, 1991) This article was useful in developing the current credibility model and essential to the information in appendix two below titled "Credibility in Persuasion Process."

In the field of Critical Thinking, standard source credibility criteria are presented by Robert Ennis (Ennis, 1996, chapter 3). One example text in the field of communications, Communication with Credibility and Confidence, weaves the subject of credibility throughout the text (Lumsden & Lumsden, 1996). Credibility’s role in argumentation and refutation is well explained in the book Arguing Constructively (Infante, 1988)

Since this is an unprecedented application of a credibility model, most of the works cited in this paper involve logic, critical thinking, and communications. Therefore, for a more comprehensive list of available literature, see the attached bibliography.

Part 1: Review of the Antivirus Industry's Credibility

“Hyperlative, a conceptual state resulting from a high concentration of superlative sound bytes. Often bulleted. See also mind-numbing.” Coined by the author for this paper.

We have a problem. Our industry has a problem. People don't believe us and we can't blame them. Since its inception, the antivirus industry has been known for its exaggerated claims, misleading promises, competitive smearing, and a good number of other rationally immature practices.

Detects all known and unknown viruses. No updates required. The solution to your virus problem. The only antivirus you'll ever need.

Today, sound bytes are often used in advertising. Slogans are hammered into our culture by incessant repetition. People learn attitudes by rote, rather than rational thought. Yet sound bytes are nothing new.

Consider the following quote.

All effective [marketing] must be limited to a very few points and must harp on these slogans until the last member of the public understands what you want him to understand. The masses are slow-moving... and only after the simplest ideas are repeated thousands of time will the masses finally remember them.

[Marketing's] effect must be aimed at the emotions and only to a very limited degree the so-called intellect.

[Marketing's] task is not to make an objective study of the truth. Its task is to serve its own right, always and unflinchingly.

While this may sound like today's standard marketing ideology, it is actually a series of quotes from Adolph Hitler. To view the quotes in their original form, simply replace each instance of the word marketing with the word propaganda. (Kahane & Cavender, 1998, pg. 124)

Teaching show you how to think. Propaganda tells you what to think. Historically, the information produced antivirus industry that is most visible to the public is of the former type. Propaganda (a.k.a. marketing claims) are still visible on boxes, in ads, and all over the net.

Credibility is important to antivirus industry. But it is important to understand the our credibility problems go further. When the industry does something wrong, like issuing an uncalled for alert others suffer. For example, the people whose job included distributing alerts also loses credibility within that company. In addition, out contractors and consultants who send the alert to their clients also loose credibility.

Part 2: Developing a Credibility Model

When one's abilities to reason have not been cultivated and refined to a state of efficiency, one is in no position to evaluate accurately the logical force of the opinions other people express.

The Counterfeit Wisdom of Shallow Minds (Habermehl, 1995, pg. 39)

Purpose

The purpose of this paper is to present a workable credibility model that can be used to monitor information from antivirus organizations and generate credibility ratings for those organizations. While the model may be implemented by anybody, it is specifically proposed that members of the antivirus community cooperate in the implementation of this model with a view to self-regulation of the antivirus industry. When the industry implements such self-regulation, that in itself will increase the credibility of the overall industry. In addition, the credibility of individual organizations will increase when they show themselves cooperative in this effort. Once in place, the model will provide an important tool in credibility management.

Mapping Credibility

The Two Species of Credibility

The credibility of any information involves two distinct factors.

The first factor is the credibility of the source. Source credibility involves both historical patterns of credibility and current level of credibility.

The second factor is the credibility of the information content. Content credibility involves claims made in the information; along with the evidence and reasoning for each claim. Claims, evidence, and reasons may be either explicitly presented or implicit and assumed.

Source and content credibility are analyzed and gauged different. In addition each type of credibility has both abstract and concrete aspects.

The information above may be tabulated thus:

Source	Information
Abstract aspects of the information source.	Abstract aspects of the information content.
Concrete aspects of the information source.	Concrete aspects of the information content.

Both abstract and concrete aspects of credibility can be analyzed. However, only concrete aspects involve substantial data and can be measured objectively. On the other hand, abstract credibility aspects involves subjective analysis of abstract data. Therefore, this paper is primarily concerned with the concrete aspects of both source credibility and content credibility, because these aspects can be measured, mapped, and thereby monitored.

To illustrate the different natures of the abstract and concrete aspects of credibility, the table above is expanded in the table below provide more information about these differences.

Information on Source	Information in Content	Credibility Type	General Nature	Data Type	Method of Gathering	Method of Testing
Image	Perception	Abstract	Nebulous	Abstract	Observation	Opinion
History	Proof	Concrete	Material	Substantiative	Measurement	Criteria

The next four sections explain the standard method for analyzing each combination of information and type.

The How, Not the What (Abstract Source Credibility)

"I wouldn't have seen it, if I hadn't believed it for myself." Anonymous

We evaluate the abstract credibility of a source (a speaker or author) based on how we perceive the source itself. It involves our perception of the source; the source's image.

This is not an evaluation of what the source is saying (which is content and is discussed later). Rather, we evaluate the characteristics of the source itself; namely how the content is being delivered.

Three source characteristics are key to our evaluation.

· Expertise (Competence)

· Trustworthiness

· Dynamism

One researcher in credibility explains this perception (or appearance) of credibility thus:

Your credibility is the image people of you in terms of at least three factors: expertise (the extent to which you seem to know what you are saying, possess knowledge, have intelligence), trustworthiness (the degree to which the audience feels "safe" with you, believes you are reliable, feels you have his or her best interests in mind), dynamism (the projection of energy, forcefulness, and an appealing personality, the perception that you are an interesting person). (Infante, 1988, p.121)

These factors combine to form what is usually called the "image" that is projected by the source.

As we evaluate the source's manner of communication, we naturally evaluate these factors and form what we believe to be true about the source into an attitude about their credibility (compare appendix two below). However, the data we are evaluating is based entirely on observation and is thus abstract. The result is merely an attitude or opinion about the source. It is not proof.

For example, one speaker may stumble in explanations, be unclear, and completely boring. Yet that person is a highly credible expert who is interested in helping you. On the other hand, another speaker may appear completely competent, interested in you, and charismatic. Yet that person is a psychotic cult leader looking to suck you in.

In light of the above, the appearance of source credibility (taken by itself) is a poor measure of actual credibility and is completely inappropriate for measuring and rating source credibility.

Still, people will evaluate source credibility in this manner. In turn, the evaluation will affect the manner in which they evaluate the message. Thus, it is important to understand its nature and implications. It is a metric nearly everyone applies in the real world. So remember, people will be evaluating your information and mine based on our apparent credibility.

This is not to say that such credibility evaluation is completely misleading and useless. It can serve us well, but only in context. It can be useful when used with concrete credibility metrics. It might serve as a flag or trigger in other evaluation. For example, if the psychotic speaker above seemed too smooth or made you feel uneasy, it would likely trigger a higher level of skepticism, which would result in a closer examination of the actual content of the psycho's presentation.

Beware Sophistication (Abstract Content Credibility)

Many people believe the word "sophistication" has a good, positive meaning, while in fact the primary meaning is: "using subtly deceptive reasoning that sounds plausible but is fallacious." The word comes from the ancient Sophists; Greek philosophers known for their "adroit, subtle, and specious reasoning."

Such "sophistication" can be exposed by concrete content metrics (logical analysis). However, we are still discussing analysis at the abstract level. At the abstract level of source credibility, we found "image" defining perceived credibility. In the case of perceived content credibility the defining factor is "involvement."

As in image, dynamism shapes involvement. This is especially true when the content is viewed by the audience as highly interesting, involving, or important. Yet such involvement may pivot on style rather than substance. For example:

One...might utilize devices such as: 'it follows from what I have just said', when in fact nothing follows; 'I have argued that . . . ' when in fact no argument has been offered; 'I refute the suggestion that . . . ' when the suggestion is merely contradicted, not argued against, and 'it has been shown that . . . ' when nothing of the kind has been done... Effective persuasion is not just a matter of style. (Fairbairn & Winch, 1996, p.225)

A similar form of persuasion is propaganda. While rational presentation proves why something is true, propaganda merely tells you something is true. Propaganda teaches you what to think, not how to think. In this way, propaganda is much like marketing.

Therefore, as in the case of a source's image, involving and interesting content has nothing to do with correctness. It is still perception. It is still opinion. The data is therefore no less abstract than the data that shapes source image. So involvement, like image, is a poor measure of actual credibility and is completely inappropriate for measuring and rating content credibility.

In addition, it may come as a surprise to you (it did to me) that message content can actually reduce the importance of source credibility to near nonexistence. That is, highly involving content can persuade a person without regard to their view of the source's credibility. One credibility researcher discusses this point extensively and concludes "expert sources were no more persuasive than non-experts on an involving topic."

Applied Credibility Science (Concrete Source Credibility)

"One of the great commandments of science is: Mistrust arguments from authority. Authorities must prove their contentions like everyone else." (Sagan, 1996, p.28)

"In science there are no authorities; at most there are experts." (Sagan, 1996, p.210)

Concrete aspects of source credibility are measured by a set of criteria. The set of criteria presented here is based on a standard set used in the field of critical thinking. (Ennis, 1996, Chapter 3).

The criteria are (in order of importance):

· Competence

· Reputation

· Process Maturity

· Uniformity

· Conflict of Interest

The competence criterion is a combination of factors: expertise (based on knowledge and experience) and the ability to communicate that expertise. Does the source display insightful understanding? Can the source readily provide reasons for claims?

The reputation criterion involves the source's history of credibility. A good reputation being defined as an historical pattern of accuracy and truthfulness in general, but especially in the area of concern. Is the source respected by others in the field, including competitors?

The process maturity criterion focuses on how meticulous the source is. It is based on the source's past pattern of organization and care. The source's work should evidence careful methodology. Does the source insist on established procedures?

The uniformity criterion is based on the sources agreement with others in the field. Yet uniformity does not imply conformity. A good measure of uniformity, which does not involve agreement with others, is whether or not a source follows accepted industry-wide standards and recognized methods. Is the source in general agreement with other experts? Does the source follow accepted industry standards?

These four criteria are key. They are concrete and measurable. But before we go on the fifth criterion (which is more abstract), it is vital to point out that the uniformity criterion often involves the other criteria.

A source may rightly disagree with some, many, most, or even all others in the field. But a credibly source does not slander competitors. Such attacks are often a ploy to cover incompetence by attacking the competence of others.

A source may also lack reputation (and recognition) and attempt to stand out simply by projecting the image of being an innovator, different or unique. Or they may do so simply to get the media spotlight.

A credible source does not attack the status quo without very good reason and exceptional evidence. A source that proclaims that the methods used by all other sources are wrong (e.g. scanning), must substantiate the claims by proposing a superior process.

Of course a source may attack others in the field out of egotism or a contrary nature. This may not violate any criteria (except maybe the "maturity" part of number three).

The criterion based on conflict of interest is difficult to pin down. In fact, Ennis presents it as "apparent conflict of interest" and thus expands it far beyond any actual conflict of interest. Key to this difficulty is the fact that the concept of conflict of interest is abstract; it usually does involve appearance.

Conflict of interest is defined as "a conflict between the personal interests and the official responsibilities of a person in a position of trust." Common usage goes far beyond this definition to include anything and everything that might possibly cause bias or partiality.

The problem is that anyone can point at any source can cry "conflict of interest." This is because conflict of interest has the same nature as conspiracy theories; namely, it is generally impossible to either prove or disprove. Having this nature actually moves "apparent conflict of interest" out of the realm of logic and into the realm of fallacy. It is used as an appeal to ignorance; the accused cannot disprove it therefore it must be true.

The Art of Deception (a book on critical thinking, but a good title for a marketing manual) uses the conspiracy theory example and explains this deceptive ploy:

The inability to disprove your case is tantamount to a proof that it is correct. As long as nobody can show that you are absolutely mistaken, then act as if you are absolutely right. Keep on harping upon the fact that some of the things you said, even if they are trivial, are right and that the opposition cannot disprove your case. (Capaldi,1987, p.133)

Taken in our current context, unprovable claims may be taken into consideration, but only after the claim has been readjusted by our other criteria. Reputation especially can virtually (and rightly) dissolve a apparent conflict of interest. To a lesser degree, uniformity and process maturity can also diminish such claims.

For example, suppose a magazine is accused of conflict of interest when a big advertiser wins top honors. We can decrement the accusation for each of the below questions we can answer "yes" to.

· Was the tester competent (knowledgeable and experienced)?

· Does the magazine have a good reputation of being fair?

· Was the testing process acceptable for the needed results?

· Do the results reflect the results of other testing bodies?

If we answered "yes" to these (even with "unknown" to a couple), the accusation would be immediately discounted by most rational beings.

If such bias is claimed, the accused is innocent until proven guilty. The burden of proof is on the accuser. The accused is not obliged to prove their innocence, but the accuser is obliged to prove or retract the claim.

Actual conflict of interest can, of course, exist and can affect credibility. But actual conflict of interest can be proven. If a test is biased then an independent retest can prove it. If an article misrepresents facts then those facts can be independently examined. If such bias is proven, then the source's credibility should rightly suffer.

Therefore, any claim of conflict of interest must have associated proof. If it does not then it is no more than a fallacious ploy; a baseless attack on the credibility of an information source.

The Molecular Disintegration of Reason (Concrete Content Credibility)

"A simple person believes every word he hears; a clever person understands the need for proof." Proverbs 14:15 New English Bible

Analyzing content credibility is simple and precise. Measuring the results of analysis is completely objective. This is due to the fact that content claims can often be reduced to a binary level. They are either valid or invalid (true or false). The process of analysis are based on traditional methods of argument analysis. For a claim to be anything more than mere option, it must meet several requirements. If a claim does not meet all the requirements, it is invalid. If it meets all the requirements it is valid.

A claim is nothing more than a logical formula.

A claim is a conclusion based on one or more premises. Each premise represents one of two components, evidence or reasoning.

Most claims actually have multiple premises. Some of which may be implied rather than actually stated. Look at the following claim and analysis:

Tests show MalWall stops more malware!

Premises:

· Evidence (Explicit) Tests show MalWall stops more malware.

· Reasoning (Implied) If a product stop more malware then you should use it.

Conclusion (Implied):

· Therefore, you should use MalWall.

The first level at which claims can be evaluated involves validity of form.

There are two logical forms that a claim may take, deductive or inductive. Deductive forms are meant to prove the claim is true. Inductive forms are not meant to prove the claim is true, rather they are meant to prove the claim is likely (believable and probable).

In a valid deductive claim, if the premises are true then the conclusion has to be true. In a valid inductive claim, if the premises are true than the conclusion is probably true. Thus, while it may be more difficult to evaluate an inductive claim's conclusion, the key element in both forms involve the truth of the premises.

Note especially, at this point, that the term "true" will not be used in the rest of this discussion. A good number of books on logic use the term "acceptable" rather than true. The reason "acceptable" will be used here rather than "true" is simple. A premise can be technically or arguably "true" is some since and still be unacceptable. To clarify the difference, think of acceptable as meaning "obviously correct."

Take the example of the XM/Laroux virus. A top virus professional discovered the virus and immediately sent samples to other virus professionals. Several days later, a different company issued a press release that they "discovered" the virus. When pressed on how they could "discover" a virus that had been sent to them earlier, they weaseled out by stating that they had received it from a customer and therefore "discovered" it on the customer site. So in a since the claim was technically true in that sense. However, the claim deceptively mislead the press and public. The claim made their company appear to be better. The claim was completely unfair to the actual discoverer and a abuse of her trust. So while the claim may have been arguably true, but it was by no means "obviously correct" (acceptable).

For the overall claim to be acceptable the following conditions must be met:

· Premises are acceptable (obviously correct)

· Evidence is sufficient to support the conclusion.

· Evidence is relevant to the claim.

· Evidence is unambiguous.

· Conclusion is actually based on premises.

Developing a Model of Measurable Credibility

Substantial Credibility

The above information allows us to create a model that is based upon the measurable variables that define substantial credibility.

Credibility of Source	Credibility of Information
Competence Reputation Process Maturity Uniformity	Acceptable Relevant Sufficient Valid
Conflict of Interest	Conflicting Interpretation

Substantial Dialogical Ethic

Implementing a code of dialogical conduct serves two purposes. First, it defines a standard of effectiveness based on standardized practices that reflect rational maturity and are most often successful as credible communications. Second, it defines a standard of ethics which is not based on what is considered unethical or "immoral" communications, but based on practices that most people agree are fair and ought to be practiced. Thus, it is much easier to define, agree upon, and apply a dialogical ethic than it is traditional ethics statements.

Presented below are the four key aspects from a standard model of dialogical ethics used to examine message content. (Lumsden & Lumsden pg. 25)

· Truth: Statements are accurate, complete, and presented honestly. For example, material that misrepresents facts by exaggeration, omission, or other manipulation is considered to be unacceptable.

· Freedom: Avoids manipulating others in ways that subvert the their rights to decide for themselves. For example, material plays on a user’s fears (of viruses, of managers, or losing their job) is actually an attempt to coerce the user. Such an approach attempts to distract the user from alternate solutions.

· Fairness: Does not manipulate information to devaluate other approaches. For example, material that presents unfair claims (such as an out of date reviews or attacks on a competitor that exaggerate data) is an attempt to imply that the information source is better by trying to make others look bad.

· Respect: Demonstrates respect for the intelligence of the recipient and does not address them as inferior. User abuse involves mental bullying. For example, material that specifically talks down to users (such as by using terms the user probably does not know, or by focusing on the source as an unquestionable authority) demonstrates that the source does not respect the user’s intelligence.

Note a similarity between the four examples that violate the four aspects. Each one involves an effort to persuade the information receiver without providing clear claims based on solid evidence and sound reasoning. These examples are therefore dialogically unethical. This illustrates the importance of these aspects in rating credibility.

To illustrate, a message may contain no support for its claims. This in itself is not a problem. The message may have no effect on credibility simply because there is nothing to analyze logically. However, if the message uses tactics similar to one or more of the negative examples, it can be analyzed and rated based on the dialogical ethic factors.

To implement the credibility model being presented, these four concepts should be hammered into a general dialogical ethics standard for analyzing message content from antivirus organizations. Thus we may add this set of factors to the model in order to complete it.

A Complete Credibility Model for the Antivirus Industry

The table below presents a complete credibility containing all the factors discussed thus far.

However, the practical application of the model should focus only on two of the groupings.

· The concrete (Metrics) aspects of information credibility.

· The dialogical ethic aspects of information credibility.

The rest of this presentation will refer almost exclusively to the two sets of criteria.

	Source (S)	Information (N)
Metrics	S(1) Competence Reputation Process Maturity Uniformity Conflict of Interest	N(1) Acceptable Relevant Sufficient Valid Conflicting Interpretation
Image	S(2) Expertise Trustworthiness Objectivity Dynamism	N(2) Competence Trustworthiness Coorientation Involvement
Dialogical Ethic		N(3) · Truth Freedom Fairness Respect

Using the Credibility Model

Initial application of the model would involve two processes. The first involves application of logical disintegration to information containing claims. In this case the most important factors can be represented in a trinal form. That is, they can be evaluated as true, false, or undetermined.

The simplest application involves claims that can be determined to be either true or false (acceptable or unacceptable). If the claim is substantiated then increment credibility value, if falsified then decrement it. Claims that fail standard evaluation are either nulled out or, if the claim is unacceptable (misleading, unbelievable, or deceptive) then the credibility value may be decremented.

It is often easy to see whether information involves disclosure or deception. It is also possible to see where a companies priorities are by asking a question about that "critical" announcement or alert. Did it serve to accurately inform or did it serve simply to increase sales and placate stockholders.

Phase 1: Process Outline

The following outline could also be presented as a flowchart. It shows the series of questions that can be applied to information content to determine the validity of claims made in an information object.

Note: Under heading III. An unacceptable claim is simply one that cannot be proven or disprove. These are usually ambiguous, vague, or metaphysical: smart users (undefined) use our product; most users surveyed prefer out product (what percent of how many users? 51 out 100 or 2 out of 3); nearly all intelligent life forms in the universe use our product (try to disprove that one).

Analysis Sequence of Concrete Information Analysis

I. Are any claims made?

II. Are any claims clearly implied?

III. Are any unacceptable (unprovable or easily falsifiable) claims made?

IV. If any claims are made of implied then:

a. Is any evidence presented to substantiate the claim? If evidence is presented then answer these questions.

i. Is the evidence acceptable, that is, clearly correct?

ii. Is the evidence recent?

iii. Is the evidence relevant? Does it actually support the claim?

iv. Is the evidence sufficient to support the claim?

v. Is the evidence consistent with known facts?

vi. Is the evidence from a reliable source?

vii. Is the evidence ambiguous, easily interpreted differently?

b. Is there a line of reasoning based on the evidence? If so then answer these questions.

i. Is the reasoning actually supported by the evidence?

ii. Does the reasoning rely on other, unproven assumptions?

iii. Is the reasoning consistent with the rest of the evidence and reasoning?

iv. Is reasoning from cause to effect valid?

v. Are comparisons in the reasoning fair and applicable?

vi. Is the reasoning fallacious?

c. Is any standard principle of dialogical ethics violated including:

i. Dialogical Code of Conduct

1. Fallibility (unproven or unprovable claims)

2. Clarity (vs. jargon or market-speak)

3. Fair Representation of Competitors

4. Commitment to User Interests

ii. Ethics of Message Content [Lumsden and Lumsden pg. 25]

1. Truth (accuracy and completeness)

2. Individual Freedom (choice)

3. Fairness

4. Respect

For more information on methods on analyzing argumentation see (Kahane & Cavender, 1998, ch. 8), (Infante, 1988, Ch. 6), (Conway & Munson, 1997, Ch. 2), (Browne & Keeley, 1998, the entire book).

Phase 2: Applying Functional Credibility Calculus

The formula P=S+N was presented a formula presented at the beginning of this paper where:

P = Persuasion

S = Source Credibility

N = Information Credibility

Assume S and N to be variables of a type called "C" where C is a value used to "rate" any given claim.

In turn, C is calculated as the summation of two sets of values.

· The first set, v, contains a number variable factors.

· The second set, w, contains a number of constant (weighted) factors.

Every member of set v has a corresponding member in set w. Therefore every w_n contains a constant that is mapped to the variable v_n and visa-versa. Thus, the value of C is calculated by multiplying each w_n and v_n then adding the result to C.

Variables within the each set represent values assigned any given criterion.

Thus, the first element in set w would be a weight based on the relative importance of Expertise in analyzing credibility; whereas the first element in set v would be a value assigned as based on analytical data.

The weighting of variables become critical in extreme cases. For example, a completely credible source may not be considered dynamic (in their speaking or writing), especially if they are struggling against that awesome foe of communications, the English language. At the same time, a person with zero credibility might come across dynamically and persuasively,

The model is not meant to grade the ability of speakers' or writers'. Such aspects, while relevant in the credibility universe, are of minor importance.

The good news is that this model will not be applied to speakers. It deals with information streams that are electronic (or possibly printed).

Moreover, no one is saying that all factors have to be included. The model is presented to be comprehensive; to the point of excess.

For this reason, we may safely ignore S (the variable of type C representing source credibility) and focus exclusively on N (information credibility). In addition we can focus on the concrete factors.

Criterion	Yes	No	Unknown
evidence acceptable
evidence recent
evidence relevant
evidence sufficient
etc
Sum

An implementation of this model would calculate all or selected input to produce a value for C, the source's credibility rating.

Implementation of the Credibility Model for Self-Regulation

Overview of a Credibility Rating System

The following chart depicts a functional credibility rating system.

Blue arrows (top, down to monitor) represent the information flow from an organization to the public. Red arrow (bottom right) represents feedback from the world. Gray arrows (center) represent the credibility stream.

To implement application of the model to any given information source several steps must be taken. The first step could be the hardest. It involves initializing C, the source's current level of credibility. It may be best to initialize C to a neutral state.

The credibility rating itself is a functional component. It functions as an adjustment to what could be called the system’s level of skepticism. To illustrate this consider what goes through your mind when listening to a conference speaker. If you are suspicious of the person, you listen as a skeptic. As that person speaks you listen and analyze the claims made, measuring the soundness of their argument. If however the speaker is someone you trust you will listen and view the information as reliable and build your own ideas from the claims.

In a similar fashion, the credibility rating raises or lowers the “skepticism” level in the filter component. If a organization is highly credible, the amount of information objects sent to the monitor is lower. On the other hand, if the organization has little credibility, more information will be sent to the monitor.

The purpose of this sliding scale is to minimize the amount of information sent to the filter and thus minimize the work.

This brings to the filter and a critical clarification. The filter must not divert information for testing, rather the information flows on to its intended receivers, while copies are sent to the monitor. The reason for this is important are obvious: The system is not a censor. It must not interfere with information. Bear in mind that, in a practical implementation, the monitor would simply be a subscriber to the monitored organization’s mailing list.

So, based on the monitored organization’s credibility rating, some amount of information goes on to the monitor for analysis.

The monitor performs analysis based on the process outline and criteria in the last section. Upon this basis, that information object is rated. In most cases, the rating will be neutral and the overall rating will be unaffected. If the information receives a positive or negative credibility value then the overall rating may be affected.

In practical terms, the monitor is probably a human who spot checks press releases, web pages, email alerts, and organization newsletters. In this case, single information objects will probably have no effect in themselves. Rather, the monitor will document the value and watch for a pattern in those values. The documented pattern can then be used to reset the source organization’s credibility rating. The exception to this would be where a source sends out a totally unacceptable object, which might affect the rating immediately.

Drawbacks of the Credibility Rating System

There are a number of negative factors that must be considered.

The initial system will most likely be a network of people connected with the antivirus industry. Thus, personal opinions (view towards friends and enemy) must be considered. Also the monitor must spend at least some time in monitoring. This must be approached in a manner that does not interfere with more critical tasks.

Warning: If you volunteer to do this in your spare time, the rest of us will immediately question your credibility. You may have to provide evidence of said “spare time.”

The sheer volume of information to choose from makes it hard to manage. This is where an automated pre-filter would come in handy. This could be for example flags set in the mail reader that immediately deletes most messages and looks for credibility warning flags (there is much common wording in questionable and wrong antivirus product claims).

In an industry with huge egos (including corporate egos) getting cooperation must be a consideration. Yet cooperation at this level is not unheard of, as in the case of the WildList. As usual the system must have clear benefits to both users and the individual organizations.

Where to Next

More steps must be taken before a practical implementation of this model is feasible. The following is a list of prerequisites.

· Terms (like “acceptable” or “vague”) must be clearly defined and agreed upon.

· A standardized dialogical ethics statement must be drafted and agreed upon.

· A specification for the filter must be drafted and agreed upon.

· A set of heuristics used to trigger the filter must be agreed upon.

· The value and weight sets for information credibility must be agreed upon.

To help expedite this next phase, it is further proposed that the discussion and agreement process be pursued within a rational framework, namely “The General Principles of Rational Decision Making” presented in Appendix three below.

Acknowledgements

The author would like to thank Trend Micro for their cooperation in making this presentation possible. All of the research and most of the writing was done by the author while CEO of WarLab, which was a wholly owned subsidiary of Trend Micro. Before the paper was complete, WarLab was dissolved and the author went started WildList Research Corporation, yet Trend took steps to insure that the presentation would be given.

Appendix 1: Definition of Terms

Belief Consequence

The strength of a belief is measured on the person’s view of the consequences of that belief based on how desirable and probable the consequence is. The more desirable and probable the consequence is, the stronger the belief.

Cognitive Dissonance

A state of internal conflict where a persons beliefs and attitudes or values are in conflict. This motivates the person to change one or more of these factors to alleviate this internal conflict.

Cognitive Response

When a person takes in information about some issue, he or she responds by analyzing claims and either accepting or rejecting them. The person’s views on that issue are strengthened, weakened, or modified as a result.

Cognitive Structure

The communications science term for the relationship between beliefs, attitudes, and values.

Content Credibility

A measure of credibility based solely on the evidence and reasoning in an information object.

Credibility

The degree to which an information object or a source of information is viewed as being knowledgeable and reliable. An expert who is trustworthy is generally viewed as credible.

Fallacy

Any error in reasoning. See (Damer, 1995, the entire book).

Information Object

Any stand-alone collection of information. Examples: press release, article, alert, web page, or claims on packaging.

Self-Persuasion

The actual method by which people change their attitude. Arguments, in themselves, do not persuade people. Rather the material provides evidence and reasons upon which hearers build their own arguments and thereby convince themselves.

Source Credibility

A measure of credibility based on a historical pattern of information from a source.

Appendix 2: Credibility in Persuasion Process

Understanding the Cognitive Structure

The following list is based on common models in Communication Sciences; most notably in relation to persuasion theory. It represents a three tiered description of our cognitive system as involved in decision making.

· Our values rarely change. When they do, it is a slow process.

· Our values are the foundation of our cognitive structure.

· Our values are based on our attitudes.

· A value is completely conceptual.

· Our attitudes change. The change rate can range from slow to sudden.

· Our attitudes combine to form our personal set of values.

· Our attitudes are based on our beliefs.

· Attitude can be gauged. It ranges from most favorable to most unfavorable.

· Our beliefs continually change. The change is usually quick and painless.

· All we believe to be true about an object, forms our attitude about it.

· A belief is simply something you believe, something you consider it to be true.

· Beliefs can be tabulated; with x and y axes.

Beliefs are measured on the basis of belief consequence.

As stated above, any given belief can be tabulated with x and y axes. These axes represent the feelings that define how important the belief is to us.

The axes are defined thus:

· ·Desirability of consequence (ranging from desirable to undesirable)

· ·Probability of consequence (ranging from likely to unlikely)

Importance of consequence is added add as a third element by some experts, other experts view importance as being defined by the others (importance = desirability + probability). Therefore importance defines the weight of any given belief and thus its relative effect on attitude.

For a better understanding of this concept see (Lumsden & Lumsden, 1996, ch.16) and (Infante, 1988, ch. 8)

Understanding Attitude Change

Every attitude is based on a set of beliefs. Attitude change is a result of persuasion. The process involves replacing old beliefs with new beliefs. In other words, what we believed to be true about some object has been proven false So now we believe something different to be true about the object, thus our attitude changes.

Desirability is the stronger aspect of belief and probability is the weaker. For this reason persuasion is usually intended to diminish the perceived likelihood that the belief is true. By providing evidence and reasons the probability scale goes from certain, to probable, to plausible, to possible, to improbable, all the way to impossible.

Belief to attitude change can be symbolized: B®A

However, your attitude toward some object can reverse suddenly and without regard to your beliefs. Such change is generally caused by emotional pressure. After your attitude changes, beliefs follow. In fact, if the beliefs do not change, a state of internal conflict appears (cognitive dissonance), making you very uncomfortable.

Attitude to belief change can be symbolized: A®B.

There is an all too common form of this process. It happens as a reaction to peer pressure.

The desire to be accepted by your peers suddenly clashes with you attitude toward some object. The new attitude is adopted and it clashes with the old beliefs. To avoid cognitive dissonance, beliefs are changed. Usually the probability of (formerly "wrong") consequences is lowered by rationalization.

Another form of attitude-to-belief change involves having a bad experience that reverses your attitude toward some object. As above, beliefs follow.

This form is of special interest to us here, because when you have a bad experience someone else (your manager, a tech support person, an antivirus salesman), the results often include the person's credibility plummeting.

In general, any A®B attitude change is bad.

To summarize:

· B®A is rational change, the result of persuasive reasoning.

· A®B is reactive change, the result of emotional reaction.

Understanding the Cognitive Response Theory

(How We Evaluate Information)

According to research, B®A change is not the direct result of persuasion. Rather, it is the result of our response to persuasion. When we receive persuasive input, most of our thought goes into analysis of that input. When the input is through, we probably cannot recall the actual content, but our beliefs and attitude have changed.

The information input stimulates a cognitive response process. The input also provides raw material for that process. From the input, we select various ideas (facts, claims, evidence, reasonings, etc.). We then use these ideas, along with our own knowledge base and experience, to form either supportive or counter arguments of our own (which are also shaped by our values and attitudes). In this way, we change our own beliefs and attitudes). This is called self-persuasion. Note the following explanations.

Ultimately, any change your audience makes is because of self-persuasion. Your influence depends on your providing material that activates the audience members' motives - their needs, values, beliefs, attitudes, and goals. (Lumsden & Lumsden, 1996, pg.440)

Cognitive response is a mediating variable between persuasive communication and attitude change. Thus, this theory is essentially a description of how individuals process messages. It makes the assumption that human beings are not influences by unmediated messages, but that they think about the message (their cognitive response to it). (Benoit, 1991, pg.3)

Understanding How Credibility Affects Persuasion

In light of the above, it becomes clear that beliefs and attitudes change as a result of self-persuasion - persuasion that is based on the supportive or counter arguments we ourselves build from the information.

This bring us to the main point of this discussion.

Cognitive response is the bridge between persuasive information’s and attitude change. Knowing this fact allows us to understand the critical role that credibility plays. The information source's level of credibility (as viewed by the hearer or reader) has a direct and powerful effect on the cognitive response process.

Benoit formulates it as, P (persuasion) = S (source credibility) + M (message content) wherein S is negative in the case of low credibility, thus diminishing content persuasiveness.

In fact, the credibility of an information source often has more influence on self-persuasion than the information itself does. The reason for this is simple.

The information itself is sifted by the recipient for relevant data. That data goes into supportive or counter arguments. The user's perception of the source's credibility determines the critical level of data analysis. Note this point:

Receivers are more highly motivated to carefully scrutinize and produce counterarguments in response to messages generated by low credibility sources. On the other hand, receivers are less likely to be critical of sources they respect. (Benoit, 1991, pg.7)

We may as well admit it. We all do it. We are skeptical about a source. So, when we receive persuasive information from them, we look for weaknesses: ambiguous terms, weak arguments, irrelevant evidence, fallacies, non sequitor reasoning, and such. Even when the evidence and logic are solid, we look for a different interpretation or a reasonable alternative solution.

Our own credibility (or lack thereof) directly affects any information we send out. Persuasive information will automatically be recalculated by the Benoit formula (P = S+C). If our credibility is low, any persuasive argumentation will automatically be devaluated, no matter how good your evidence, or how solid your reasoning.

Changes in Credibility

Building credibility is a slow process. Rebuilding credibility is even slower. Credibility erosion can be all too quick. Complete credibility collapse can be seemingly instantaneous.

To build or rebuild credibility requires a steady output of information with high-credibility content. At the same time, a source builds credibility by displaying competence and being trustworthy.

Competence is demonstrated when:

· Communication is clear

· Complex material is explained.

· Expertise is evident.

· Knowledge of user needs is evident.

· Flexibility is evident.

· Information is never misleading.

Trustworthiness is demonstrated when:

· Promises are kept.

· Feedback is timely (if not immediate).

· Problems are explained rationally.

· Information is unbiased.

· Errors are admitted and corrected.

· Users never feel like they are being conned.

Credibility erosion is caused by being incompetent and untrustworthy. Therefore, taking the steps needed to build or rebuild credibility will (quite obviously) slow and hopefully halt the erosion.

Credibility collapse is another matter. It may or may not be deserved. It usually results from evidence or allegations of apparent wrongdoing; something that is perceived as unethical, illegal, dishonest, self-serving, or prejudicial.

You do (or apparently do) something that is perceived as a violation of a person's set of values. In response, the person reacts. Thus their attitude change is reactive. (Remember A®B? It is the bad one.)

If the collapse is undeserved, you must respond immediately. One way to counter an A®B reaction is to show that the allegations are unfounded. If they are unfounded then there is no proof, no evidence, no support for the claims. Point out that the source of the allegations has the burden of proof. They must prove their allegations. You do not have to prove your innocence, they have to prove your guilt. But of course, being the competent, trustworthy organization (or individual) you are, you feel obligated to reassure those who depend on your information. Though not obligated to do so, you should answer the charges and prove the allegations false.

If the credibility collapse was deserved, then you got what you deserved.

Appendix 3: The General Principles of Rational Decision Making

The principles presented below are based on “A Code of Conduct for Effective Rational Discussion” in the book Attacking Faulty Reasoning. (Damer 1995. Ch. 8)

The Principle of Fallibility

Each participant in the discussion should acknowledge that possibly none of the alternative positions presented are correct and, at best, only one of two opposing positions can be more reasonable. Therefore, one must be willing to thoroughly examine the issue and change positions if the opposing view is demonstrably more reasonable (more defensible).

The Principle of Truth-Seeking

Each participant should be committed to the task of earnestly searching for the truth or at least the most defensible position on the issue. Therefore, one should be willing to (a) examine alternative positions seriously and without prejudice; (b) look for insights in the positions of others; (c) and allow other participants to present arguments for, or raise objections to, any alternative position being discussed.

The Principle of Burden of Proof

The burden of proof for any positions rests on the participant presenting that position. Therefore, if and when another participant asks, the proponent should provide evidence and reasons to support that position.

The Principle of Charity

If a participant’s position is restated or reformulated by another participant, it should be expressed in the strongest possible version that is consistent with the original intention of the first participant. Therefore, if there is any question about that intention or about implicit parts of the position, the first participant should be given the benefit of the doubt in the reformulation.

The Principle of Clarity

The formulation of all positions, defenses, and responses should be free of any kind of linguistic confusion (such as jargon, vagueness or ambiguity) and clearly separated from other positions and issues. Therefore, each position on the main issue should be stated clearly and simply with focus on the main issue.

The Principles of Evidence

Each position presented should be (a) directly relevant to the issue; (b) sufficient to support the position; (c) and demonstrably true. Therefore, any position presented as true (or acceptable) should be a relevant and complete presentation of facts. The position should also stand up to rebuttal.

The Principle of Resolution

The purpose of rational discussion is to examine and weigh alternative positions and ultimately to decide what to do or believe. Therefore, after all positions have been presented and defended, an issue should be considered resolved by all participants if (a) the proponent for one of the alternative positions successfully demonstrates and defends that position; (b) and that position is clearly supported by the strongest evidence and reasoning. In this case, participants are obliged to consider the discussion fairly resolved and cooperatively follow the course of action dictated by the accepted position.

The Principle of Suspension of Judgment

If no single position comes close to being successfully demonstrated and defended, or if two or more positions seem to be demonstrated and defended with equal strength, one should, in most cases, suspend judgment about the issue. If however practical considerations seem require an immediate decision, one should weigh the relative risks of gain or loss connected with the consequences of suspending judgment and, on those grounds, suspend judgment or resolve the issue (perhaps as a temporary resolution).

The Principle of Reconsideration

If a successful or at least good position has been adopted, and that position is subsequently found by any participant to be flawed in a way that raises new doubts about the merit of that position, one is obliged to reopen the issue for further consideration and resolution.

For practical examples of using this later list, along with standard logical analysis, see The Counterfeit Wisdom of Shallow Minds (Habermehl, 1995), which makes application of these principles throughout the book.

Bibliography

Benoit, W. L. (1991). A Cognitive Response Analysis of Source Credibility. In B. Dervin (Ed.), Progress in Communication Sciences (Vol. 10. pp. 1-19). Norwood, NJ: Ablex Publishing

Browne, M. N. & Keeley, S. M. (1998). Asking the Right Question. A Guide to Critical Thinking. Upper Saddle River, NJ: Prentice Hall.

Capaldi, N. (1987). The Art of Deception. Amherst, NY: Prometheus Books.

ConsumerSearch (2000). ConsumerSearch: Product Information: Antivirus Products (Copyright 2000) [On-line]. ConsumerSearch. Available: Error! Reference source not found.. Path varies. Follow links from home page: “Office” then “Antivirus Software” under the Computers label then the button labeled “All Reviews.” Last Accessed: 2001, February 1.

Conway, D. & Munson, R. (1997). The Elements of Reasoning. Belmont, CA: Wadsworth Publishing.

Damer, T. E. (1995). Attacking Faulty Reasoning. Belmont, CA: Wadsworth Publishing.

Ennis, R. H. (1996). Critical Thinking. Upper Saddle River, NJ: Prentice Hall.

Fairbairn, G. & Winch, C. (1996). Reading Writing and Reasoning: A Guide for Students. London: Taylor & Francis.

Habermehl, L. L. (1995). The Counterfeit Wisdom of Shallow Minds. New York, NY: Peter Lang

Infante, D. A. (1988). Arguing Constructively. Prospect Heights, IL: Waveland Press.

Kahane, H. & Cavender, N. (1998). Logic and Contemporary Rhetoric. Belmont, CA: Wadsworth Publishing.

Lumsden, G. & Lumsden, D. (1996). Communicating With Credibility and Confidence. Belmont, CA: Wadsworth Publishing.

Sagan, C. (1996). The Demon Haunted World. Science as a Candle in the Dark. New York, NY: Ballantine Books.

Wells J. (2000). Lies, Damn Lies, and Marketing (Credibility in Virus and Antivirus Issues). WarLab Journal [On-line]. Available: Error! Reference source not found.. Last access: 2001, February 1.