Can we make this multiple choice? What are the possible answers? Consider the alternatives. One story might go that the possibilities come under the main headings, roughly, of the formal model's theory of a good argument, or the model we get from inductive logic and theories of scientific reasoning, or perhaps the fallacy approach can provide at least a piece of the standard of a good argument. Or, just for the sake of good form, none of the above.

The correct answer is none of the above. Each of the alternatives named above fails importantly to be able to supply us with evaluations to the effect that an argument is good. Look at each in turn.

The model of argument we get from formal logic is incapable of justifying us in saying whether an argument is relevant, trivial, crucial, or better or worse than any other formally correct argument. Further, that model is not able to tell arguments from non-arguments (viz. the mortality of Socrates). Except in philosophy and highly mathematical disciplines, it is very seldom that anything remotely like formally valid arguments are offered in the ordinary give and take of scholarly debate. Even when they are offered, the evaluation of those arguments is conspicuously not just an effort to apply the formal theory to see if the arguments measure up. Further, the formal model only begins to do its work, only can begin to be applied, after the work of reading the argument and translating it into the proper form is over. Thus, the formal model cannot help with any judgements as to whether the argument has been fairly represented in the form to which the model is brought to bear.

The situation is more complicated in the case of inductive models. In a way this is to be expected, since inductive models were hatched as a result of a cooperative process on the parts of logicians of formal stripe, scientists (most of whom believe in logic), and philosophers of science. Logicians have tended to think their discipline has something to offer people who are trying to figure things out, and who better than scientists? Inductive and scientific models of good arguments are what happen when logic meets the world.

The result is an improvement on logic, perhaps, but still has its problems. Because scientists are often trying to defend, not the certainty, but rather the likelihood of some claim or solution, much of the work in applying logic to science has been in developing a model which will admit of degrees of evaluation rather than being limited to "this argument is valid" or "this argument fails." A great deal of attention has thus been paid to elucidation of the forms of arguments from sampling and to the role of disconfirmation of hypotheses in advancing scientific progress.

We need not deal with these in detail to see that some of the same problems still will be inherited from formal logic. In particular, in so far as the evaluation of arguments rests on an appeal to the argument's form, it is not possible to say whether the argument has to do with the issues at hand, nor can the evaluation help with our deciding whether we have found a form which properly represents the argument.

And nothing in the form of an argument can tell us whether a sample has been drawn in a manner appropriate to the content of an argument, so that the differences in doing a good job of getting, say, blood samples, soil samples, demographic samples, atmospheric pollution samples, or makeup color samples, cannot show up in the argument form. As a matter of fact, when scientists disagree they do not use the materials or formal concerns they inherit from logicians or philosophers of science or those of their own who are logically inclined. We will take a look at a couple of these later.

Fallacies? Well, of course not. The fallacy approach never pretended to tell us what a good argument is, only bad. The fact that an argument does not commit a fallacy says next to nothing about its being a good argument. Even if we knew all two thousand-odd fallacies cold and never missed an instance, no one has ever claimed that these are all or even most of the possibilities. Too many of the fallacies also depend on context for their being labeled a fallacy to be just. Saying something which would be a vicious ad hominem in a city council meeting may be exactly the right thing to bring up in a criminal trial when the credibility of a witness is a legitimate issue. The fact that an argument commits a fallacy is not a matter of determining the argument's characteristics separate from its setting; to make a fallacy label stick requires us to be able to evaluate the argument separate from the theory--that is, the fallacy name and characteristics. That is, being able to tell whether an argument commits a fallacy requires us to be able to tell separately whether is it is a good argument.

What then is the standard of a good argument?

I suggest there is no standard of good argument separate from individual issues and debates. As we have seen, the method advocated in this work is what I call the method of worry. It requires us to hear all the arguments which bear on the issues, to provide the best possible answers to each argument, and then to array the arguments in front of us on the way to arriving at our own positions on the issues. It is a joke to call it a method, since it is the opposite of any shortcut, and it apparently yields nothing in one field which can be profitably brought to another. Are there any arguments for this approach rather than another? Here is one: if any other procedure is to be proposed for helping us to evaluate arguments, it must make its case partly by being able to do better than this procedure of worry. That is, this procedure will in fact be the standard against which any other evaluative procedure must be measured. For a standard of good argument to prevail it must do better at sorting out good arguments from bad in particular issues than the actual debate on that issue does. To engage in thoughtful debate on the issue is to use the method of worry. The standard of standards, then, just is the method of worry.