Back to the map.  

Abstract:

We select the aim of {True X}, because we are interested in it, as are some other people.
We justify M1: The critical attitude.
We justify M2: Using personal experience as the foundational source of testing for any investigation with this aim.
Disclaiming originality, we propose, and justify, M3-25; these are a series of technical methods, for use in five situations, involving: Facts, Laws, Theories, Aims, and Attitudes and Institutions. Finally we assess the extent of justification for these methods.   Some are logically justified, given the aim; others are justified if sceptical doubt is quarantined; others are then justified by historical evidence, while some are justified by our knowledge of human nature.

If you would like to download a pdf version of this essay, click here.

 

THE METHODS OF SCIENCE

   Our chosen aim is {Truth}, {True X}, where X is a proposition.
   We now begin to devise the best, the most rational, reasonable, methods for achieving this aim.

 
Core Methods - proposed as timeless standards

   There is no sharp distinction between core, primary, methods and others.  Methods that follow from consistency with the aim and the meaning of 'rational' are very strong candidates. Those that express our commitment to experience, especially if we include intuition and revelation, seem strong.  Those that follow from Te and Tp (Inductive Presuppositions) are deeply embedded in human life on Earth, and unlikely to change.
   Those that follow from evidence of human nature, though very unlikely to change, are secondary, since their justification depends on the primary ones.  Similarly, those that derive from high-level guiding theories are secondary.

   Our first method follows immediately. The critical attitude is built by definition into our theoretical schema. If we agree that we are seeking for the most reasonable methods, for achieving the aim, then we are intending to list various competing methods, and then critically appraise them to select the best. Therefore a working investigation system will be most reasonable if it is doing the same (even if it has not come up with the best possible system yet, for understandable reasons).
Method (M1): The investigator should always be critical, doubting, questioning, looking for better methods, and aware of the limitations of one's best methods .
Justification: from the meaning of 'rational'; an investigator cannot consistently want to achieve her aim as rationally as possible (which implies searching for the best available methods), yet not intend to assess her methods critically.
Note: This method is an attitude, one of critical honesty. For example, if workers in a research programme have repeatedly failed to make successful novel predictions, and instead have made ad hoc complex additions, other methods do not require that they must abandon it. But M1 requires that they should honestly admit that the situation looks bad [12].    Similarly they may propose vague theories; but they should not do so with the intention of avoiding nature's guidance [13].
   We come to this uncomfortable conclusion: criticism of an activity is not abstract; it is partly criticism of the moral stature of the investigators - very personal criticism. Assessing an activity is assessing the investigators.  In an extreme case, if we criticise them for knowingly falsifying evidence, or claiming that they are justified in holding that their theories are true when they know that they do not have sufficient evidence, we are accusing them not of errors of judgement but of lying.  In less extreme cases, we may be suggesting that they have let their intense desire that certain theory be true outweigh the lack of evidence for it.  We are, in the words of Duhem [14], criticising their "passions and interests"; we are requesting that they become "impartial and faithful judges"; we are subordinating details of experimental criticism to "moral conditions".
   The critical attitude is cold.  Historical evidence indicates/IP that warm emotion, as contrasted to cold reason, is a poor guide to the truth.  Desire, need, and passionate conviction, though intensely influential, are no use to us in this quest.  To control these human emotions requires aptitude and training.
 

   We now need a criterion for 'best'.  We propose that 'best' is taken to mean: least arguable, least doubtful, most final.  A claim may achieve this because (a) it is true by definition (b) it is given to us as inarguable and final.
   We are not here considering the best ways of inventing, proposing, devising, discovering, X. We are considering the best ways of checking, establishing, justifying, X.  In other words, we are assuming that X exists - the proposition has been made.  Why, then, do we think that X is true? What is the best reason humanity presently has for making such a claim?
 

   Our next general method: the best reason humanity presently has for claiming 'X is true is human experience - evidence.
(i) If X can be checked by such experience, then the checking will establish truth or falsity
(ii) If X cannot be checked directly by such experience, then we may have other reasons for claiming it, but they are less good. In these cases, since personal experience (empirical evidence) is so effective, we should devise techniques for using it indirectly, in various situations, to derive maximum information concerning the truth of X.
   In all such cases, we should not pretend that our information is as good as it is in case (i); the less directly we can use the best reason we have for establishing truth, the less we should confidently claim that we have established it.
Method (M2): 'Always seek methods of checking propositions which refer back to human experience'.
Once this basic way of checking has been justified as the best available, then we are able to assess all other proposed ways of establishing truth.  Thus, following Siegel's strategy, we have reached the stage of commitment to evidence.
Justification : Why do we think this is the best reason? For what reason do we think it is the best, most final, most inarguable, reason? Are there alternative reasons which could be argued to be better?
   Our answer is:
Step (i): human experience is given to us, forced upon us, in a way which disarms criticism and doubt
Step (ii): this experience has a uniquely close relationship with certain Xs, called particular facts.  If I am having a certain set of experiences, then this is truly described as X; if X is true, then in certain circumstances I will have that set of experiences.  If I am now outside, having the experiences of wetness, drip-on-faceness, and so on, then "It is raining outside" is true.  If "It is raining outside now" is true, then in the circumstance that I go outside, I will have the experiences of wetness, drip-on-faceness, and so on.  If I claim "'It is raining outside' is true" and also "I am standing outside experiencing no wetness, drip-on-faceness, and so on" then I can be accused of being inconsistent in what I am taking 'X' or 'true' to mean. (This actually applies to unverbalisable immediate experiences rather than to complete objects in the world; while objects are already theoretical entities, therefore already potentially doubtful, the immediate experience is not doubtful)
   What types of reason are these two stages?  The first is an appeal to introspective experience, of an indubitable kind. The second is true by definition; it follows from what we have chosen "X is true" to mean - which is part of our aim; we cannot deny it without facing the accusation that we do not know what we are talking about - that we are using the word 'true' simultaneously to mean different things.

 
 
The methods (techniques) for various situations: facts, laws, theories, aims, and institutions

Note 1: Particular ideas for techniques for using this powerful tool to test for truth, complete and partial, in various kind of propositions, in various situations, have gradually developed over thousands of years.
   Some of these techniques may appear inevitable, once they are seen - rather like mathematical discoveries; they may have always been implicit in the previous ideas, but hidden.
   Some may have been delayed in appearing, because the situations to which they apply had not yet arisen.
Note 2: The symbol /IP by a method, or assessment of verisimilitude, indicates that without an inductive presupposition, the theories Te and Tp , it cannot be justified. We are assuming that this is of no particular importance.
Note 3: The list may seem unnecessary long to the reader. It has been extended partly to show that theory choice - choosing one hypothesis rather than another is just one amongst many situations.  Theory choice may interest philosophers, because grand theories give us our view of the world, but it is of no special interest to methodologists.  If, for example, it turns out that our general methods give no decisive guidance in this situation, this will in no way compromise the rationality of the enterprise we are systematising.
   If physicists then insist that their theories are decisively guided by their methods, then either they are wrong, in which case they lose some credit in our rationality-assessment, or they are right, in which case we have overlooked something.

 
Facts

Situation: Two investigators A and B disagree on the report X of a personal experience.
Method: Seek for other consequences of X being true. The most immediate is that other people should report X. If everyone except B reports that X, then propose that X is true, and that B is faulty in some way (eg. hallucinating). This theory may lead to other consequences which can be tested.
This is method M3: Interpersonal testing (/IP) .
Verisimilitude Assessment (for particular facts): Once interpersonal testing has been carried out, and X is agreed, then X is as reliably/IP true as any proposition ever will be.
Justification: Using Tp; if two independent people's personal experiences (as reported) are artificial constructs, not linked to a single external reality, then the chance that they are the same, or very similar, is very small. If the two experiences are associated with a single external reality, then the chance of agreement is very high.  So achieving agreement is strong evidence/IP that the truth of X is determined by external reality.

 
Laws

Situation : Investigators seeking a generalised law find that two properties are associated, in a situation where they have no experience to suggest that there are any special circumstances present.
Method : Universally generalise the association between the two properties. This is M4: Generalise facts into laws, if there is no evidence to the contrary (/IP ).
Verisimilitude: The law is not as reliably true/IP as the fact. Nonetheless, it can be regarded as reliably true (/IP).
Justification: From Te.
Note: This is abduction (to use Peirce's term) as well as induction, in that it is a process of devising, and of testing, at the same time. Others of the following methods concern rational ways of devising propositions.
Note: If a critic asked us why this assessment of verisimilitude was reasonable, we could respond "What good reason/IP do you have to doubt it? We have experienced an example where it is true, and no example where it is not true; we have no good reason, yet, to think that any special circumstances applied. Quarantining sceptical doubts about the reasonableness of generalising, why do you doubt its truth?". Since ex hypothesi she has no evidence, we cannot see what reason she can have. And in this sense it is undoubtedly (1995) true. Which, the argument concludes, is the closest - in this situation - we can ever reasonably get to certain truth, reliable truth, indubitable truth, or, indeed, truth.
   Refusal to accept the need to establish general truth more conclusively than by human consensus, public knowledge, is characteristic of, for example, physicists (see, for example John Ziman Public Knowledge ).
(i) On the one hand the best available methods may lead us to propositions whose truth cannot reasonably be doubted; on the other hand, these propositions may not be true. We thus need not deny the existence of the truth. Majority claims can be false. Fallibilism is not the same as democratic relativism.
(ii) The implicit tyranny of the majority is correct. If Galileo had failed to convince his peers, then, to the best of human judgement, it was false that the Earth moves. Faraday regarded persuading his peers - especially those in Germany who disagreed with him - as essential; it established, to him, that he was not a crank; if his reasons could persuade even them, then they were good reasons.

Situation : Investigators are devising a new law L2 in an area of phenomena already covered by an old law L1.
Method : L1 has verisimilitude/IP, so L2 needs to retain that verisimilitude, while correcting the false consequences. The change from the Ideal Gas Equation to the Van Der Waals Equation is an illustration. This is M5: Devise a replacement law so it captures the verisimilitude of the original
Verisimilitude : The new law may have no more verisimilitude than the old law had in its heyday. But the old law now needed correction because of experimental inconsistency. This is progress/IP by running to stand still, which is better than slipping backwards.
Justification : from Aim.

Situation : Investigators are trying to show that two properties A and B are causally (necessarily) connected.
Method : Show that whenever A is present, with other properties different, so is B. Show that whenever A is absent, with other properties the same, B is absent. Show that when the size of A is altered, the size of B alters. These are M6: Mill's methods of Similarity, Difference, and Concomitant Variation (/IP)
Verisimilitude : The causal connection can be regarded as reliably true.
Justification : From Te.

{An example of the use of M6 by Faraday}

Situation : Investigators are trying to show that two properties A and B are causally connected.
Method : They should try to set up active experiments, rather than passively observing. This enables them to control the presence and absence of various factors, when using M6. This is M7: Active Experimentation (/IP)
Justification : To gain advantage of M6.

 
Theories

Situation : Investigators are devising a new theory which cannot be tested directly.
Method : Ensure that the theory has consequences which are directly, interpersonally, testable. This is method M8: Falsifiability .
Justification : From M2, since there is no other way of establishing truth.

Situation : Finding that one consequence of a theory is interpersonally false. This does not imply that the theory is entirely false, but the inconsistency implies that somewhere there is falsity. This is method M9: Take inconsistency seriously .
Justification : From the requirement of consistency, since otherwise the investigator is claiming that a proposition is both true and not true.

Situation : Investigators are devising a new theory which cannot be tested directly.
Method : Avoid devising a theory which has no further experienced facts and laws as consequences than the ones, already assessed as true, that it was designed to have.  Ensure instead that the theory has consequences which are as unlikely (improbable) as possible if the theory is a human fiction; the theory should take risks.  No vague theory can achieve this. It can be achieved by (i) having precise numerical consequences (ii) not being aware of some data when the theory is being devised. This is M10: Seek for severe testability (/IP) .
Justification : From the ability to use M11.

Situation : A theory T is proposed which cannot be interpersonally tested directly, but has various consequences which can. Method: If the consequences were highly unlikely to have been true (survived interpersonal testing) if the theory was entirely a human fiction, then this theory has some truth in it, and is worth developing. This is M11: Use severe testing as an indirect guide to the presence of truth (/IP) .
Verisimilitude Assessment (for theories): there is no assessment of the exact amount of truth-credit that a theory can gain by this method; it is a matter of judgement.  The theory is not as reliably true as a fact or a law. (/IP).
Justification : From Tp.

Note: This is successful novel fact prediction. For more details, see Novel Facts . This chapter includes some examples of past thinkers who have recognised its importance, and struggled to justify it.

Situation : Devising new theories in an area of phenomena
Method : "Make (hypotheses) ... only one after another" (Poincaré p. 134-5). If a new theory involves many new untested hypotheses, and then (i) it passes a severe test, which part should take the truth-credit? (ii) it fails a test, which part is faulty? This is M12: Minimise the number of new untested hypotheses in a new theory (/IP)
Verisimilitude : If a new theory has many untested hypotheses, then passing a severe test leaves unknown the location of the truth credit.
Justification : From logic, given the aim.
Note : A good example of an abductive method.

Situation : Devising an experiment to test a new theory by testing one of its consequences.
Method : Arrange that the chosen consequence follows from some key hypothesis, combined with other hypotheses which already have established truth credit. This is M13: Design crucial experiments (/IP) .
Verisimilitude : If the consequence comes out false, then the new hypothesis is to blame.
Justification : From M12.

Situation : A theory with some amount of verisimilitude is inconsistent with interpersonal evidence.
Method : One possibility is to alter the theory until it is consistent with the evidence. This is M14: Make ad-hoc adjustments to theories (/IP)
Verisimilitude change : None. This does not increase or decrease the verisimilitude.
Justification : From non-use of Tp.
Verisimilitude change : Less.
Justification: As in Lakatos' degenerating research programme, this depends on whether there are good reasons (ie. ultimately based on experience) to suggest that the altered theory now has a structure which makes it less likely to be true.  If simplicity is a justified guiding theory, and the ad-hoc change decreases it, then such a change justifiably reduces the verisimilitude of the theory.  This reduction in verisimilitude is therefore secondary, at best.

Situation : A theory has a consequence which is found to be false. How can the investigators locate the source of the inconsistency?
Method : If the method (M12) of minimising the number of untested/IP hypotheses in new theories has been followed, then the investigator can justifiably start by supposing that the new, untested/IP, part of the theory is the source of the inconsistency. This is M15: Suspect first those parts of a theory which have least truth-credit (/IP) .
Justification : From logic, given M12.
Note :
(i) We here deny the holistic view that the entire structure of theory, linked to experience at its edges, is homogeneous with respect to truth-credit, such that, when an inconsistency arises, the fault could equally well be anywhere.  The view arises from ignoring the (logically invisible) temporal order of theory construction - the justificatory pedigree of different parts (see Tennant (1994) p. 885 "Scientists ... set about rational revisions ... by being able to recall, and retrace, and reassess, the links of justifications in the pedigrees they have for their various beliefs, individually and as a community".  If investigators have used the Galileo-Poincaré method of piecemeal, step-by-step, construction of the structure, using specific regions of the edge to check/IP the truth-credit of each theoretical construction, as far as possible, before building further parts, then the resulting theoretical structure will be differentiated with respect to truth-credit .  Inconsistencies can then be rationally blamed on parts of the construction with the least truth-credit.
(ii) If various options nonetheless remain open to the investigator, then good experiments will need to be devised to decide between them.

Situation : Investigators are devising a new theory T2 in an area of phenomena already covered by an old theory T1.
Method : T1 has truth-credit, so T2 needs to retain that truth-credit, while correcting the false consequences.  The sequence of changes from the Dalton/Boltzmann atom to the Thomson atom, the Rutherford atom, and the Bohr atom, is an illustration.  The truth-credit in the T1 needs to be located by studying which parts of it actually led to the improbably correct consequences, and which parts were - to use Poincaré's phrase, neutral hypotheses. This is M16: A new theory should include the truth-credited features of the old theory it replaces (/IP).
Verisimilitude : The previous truth-credit of T1 can be transferred to T2; if T2 has no false consequences, this may be progress over T1. Otherwise, it is running to stand still. If T2 can also gain further truth-credit, then it will be further progress over T1. (/IP)
Justification : From Tp and the aim.

Aims

Situation : Discussing subsidiary aims.
Method : {True X} alone underdetermines the investigation. Investigators need to consider whether particular subsidiary aims seem to be assisting, or hindering, the search for {True X}. Physicists should discuss the benefits of searching for macroscopic realistic models of physical phenomena; if the search for Truth seems to be being hindered perhaps they should search for mathematical models from which the phenomena can be derived. This is M17: Discuss subsidiary aims.
Justification : From linguistic consistency, given desire for rationality.
Note : If a subsidiary aim is given precedence over Truth, then an investigation completely changes its character, and new methods apply.

Situation : Discussing the overall aim - {True X}.
Method : An investigation is only fully rational if the overall aim of the investigators is assessed from time to time. Do they still value Truth as highly as they did in the Renaissance? Perhaps it is subsidiary to the valuable aim of {Solving Problems Of Living}; perhaps it is hindering progress towards this grand aim. If so, then large parts of the investigation would be irrational. This is M18: Discuss the value of Truth.
Justification : From linguistic consistency, given an avowed desire for rationality.

 
 
General Attitude and Institutions
 

Situation : Attitude to new ideas.
Method : New ideas should always be welcomed and tested, because all available ideas are fallible. This is M19: Open mindedness .
Justification : From the aim.

 
 
Secondary Methods - some examples

These include methods which depend for their justification on evidence from the history of investigations in a particular area, indicated by the letter H. It also includes those which depend on evidence concerning human nature - human psychological, social, and animal, behaviour - indicated by the letters HN.

Situation : A single controversial claim that X is true is made by an investigator who stands to gain personally from its truth.
Method : Human beings succumb to wish-fulfilment, and lying. The claim needs to be checked by several independent investigators, who have no emotional involvement one way or the other in its truth. This method is M20: Public Repeatability Of Controversial Observational Claims (HN)
Verisimilitude Assessment : Without such public repeating, X has low verisimilitude.
Justification : From laws summarising experience of human behaviour, insofar as these exist.

Situation : Investigators are trying to find the possible causal factors which lead to state S1 of a system.
Method : Assume that there are differential laws describing the variation of the factors with time, which, when applied to the immediately preceding state S0 of the system, determine state S1. This is M21: Differential Laws (H)
Justification : A high-level guiding theory, justified by having led to the choice of lower-level theories and laws which have gained truth-credit.

Situation : Choosing between alternative forms of a law or theory.
Method : Prefer the theory which has the simpler structure. This is M22: Prefer simpler laws and theories (H).
Justification : Even supposing that the problem of the natural language in which the theory is to be stated (to avoid merely syntactical simplicity) can be solved, the justification for this method remains problematic.  It can be justified by belief in God as the creator of an elegant Universe, though whether this belief, as a theory, can be given adequate verisimilitude by the operation of our methods is arguable.  It can be justified, area by area, by reference to the empirical success of simple laws and theories; if the guiding theory of simplicity has led to truth-credit in these areas, it has some claim to be true in those areas. Whether this is sufficient is questionable.  As Poincaré remarked, Physicists have as often found complexity behind superficial simplicity, as they have simplicity behind complexity.
   Perhaps it is merely a deep-rooted human prejudice.

Situation : Choosing between theories.
Method : Given the difficulty of assessing the truth credit of a theory, this is hard. Factors to be taken into account include those above.  Duhem concluded that the decision is ultimately one of good taste.  Individual prejudices will be cancelled out by ensuring that a large number of people are involved in the judgement (more than the twelve people on a jury, but doing a similar job). The method is M23: Choose between theories by awaiting a Consensus, for example on the balance between truth-credit and simplicity (HN).
Verisimilitude : Given the difficulty of choosing, we should not place as much confidence in the truth of our theories as we do in facts and laws - unless we can locate those parts which deserve the truth credit.
Justification : The effect of individual bias is reduced by insisting on agreement from a qualified group, and by the passage of time, as inflexible supporters die. There may be historical evidence that theories chosen this way eventually gain more truth-credit than ones which were championed by individuals, but failed to achieve consensus.

Situation : Investigators have proposed various conflicting theories in an area; there is no consensus.
Method : M24: Continue to work on all theories (H)
Verisimilitude : Low for all of them.
Justification : From historical evidence that theories which seem unsatisfactory to start with, can sometimes be made better by persistent development, even in the face of serious empirical problems.
 

Situation : Human institutions of investigators and administrators.
Method : Humans tend to form into a pecking order, in which dominant individuals, older and more powerful, who hold certain traditional theories, which they have taught for years, have a disproportionate influence on whether new theories receive proper attention. Institutions need to show awareness of this danger, and take steps to reduce its effect - such as by rotating top jobs, and by reducing the formal influence of senior members on junior researchers. (This tends not to happen, since the senior members may be the only ones in a position to make such changes) This is method M25: Discourage Institutional Pecking-orders (HN).

 

Justification for the methods

   Investigators in some areas may not greatly appreciate being rated 0/5 for rationality. They may respond by denying that our methods are universally applicable,  objective, standards.
   Firstly, they may object, these guideline methods are compromised. The ideology of science is concealed within them - making them unfair tools for criticism of investigators in different frameworks.
   Religious Believers, for example, sometimes argue that their theories need to be assessed in their own terms, using their own language rather than one imposed from outside. This is the holistic theory of meaning. To avoid its problematic consequences, our assessment is couched not in the language of Science, but in a relatively neutral common language. If investigators insist that there is no room for them in this framework, and can show why - by implication already using a relatively neutral language - then we will devise another. "Admittedly, we shall find ourselves again in a framework, but it will be a better and a roomier one" [20].
   Secondly, they may object, the methods are arbitrary and unjustified. What is so marvellous about a critical attitude? Why should all theories have to make novel predictions, which then turn out to be successful? Who needs public testing? What forces them to agree with the methods? Our justifications have been:

(a) appeal to linguistic consistency
(b) practical consistency - pointing out that all investigators assume them anyway
(c) empirical evidence - referring to the history of investigations.

   We now discuss these in turn:

(a) Philosophical arguments

(i) No-one who avows the aim of {Truth} can consistently deny M 2,5,8,9,19.
(ii) M 1,17,18 follows from an analysis of the meaning of rationality. To deny them, one would have to argue for an alternative way of using the word.

(b) Practical Consistency- the Inductive Presupposition

This argument is only available to the applied philosopher. Although methods M 3,4,6, 10,11,12,13,14,15,16, are not justified by evidence, and do not seem to be justified in any other way, everyone behaves all the time as though they are reasonable - to such an extent that they have become an aspect of the ordinary meaning of the word 'rational'. Everyone is in the same boat. They are reasonable/IP. Investigators tempted to deny the acceptability of using inductive methods therefore face an accusation of hypocrisy, unless, determined to display intellectual consistency, they behave in ways which do not involve inductive presuppositions. This will involve them in expecting that their chair may not continue to support them; that at any moment the air in their region may become unbreathable; that touching pen to paper may cause an atomic explosion. Such admirably logical behaviour may unfortunately imply that further philosophical discussion with them only becomes possible at visiting times.
   As a matter of fact, no normal person (we exclude philosophers, of course) in our experience has ever objected to the use of inductive methods.
   Agreement on this vital step enables us to bypass the problems which have dogged objectivists; refined common-sense becomes available for use - not because it is proved to be sense, but because it is common.
   We can now use these methods, which by the nature of their justifications are not time-dependent, to assist justification of the remaining methods.

(c) Empirical arguments

   How could we justify M 21,22,24 empirically? There are three possibilities.
(i) We could check our methods against the practice of investigators.  But we find ourselves in a vicious circle.  Before we could use the practice as evidence it would need to be labelled as 'justified' and 'unjustified'.  But we have no independent method of doing the labelling - other than intuition or general agreement.  If, for example, we humbly ensure that typical decisions in Physics, the ones widely agreed to be justified, are justified by our methods, and then later we give physicists 5/5 for rationality, we are open to an unanswerable accusation of partiality.
   There is no way out of this circle.  If we want our methods to have teeth we cannot support them with pre-labelled historical behaviour.  The labelling of the behaviour as 'justified' or 'unjustified' will be done by our methods.
   We cannot even be descriptive methodologists, using the current consensus amongst, say, physicists, as the touchstone of our system. We must be confident prescriptive methodologists. We are proposing the standards that a good investigation should meet.
   No doubt, if our labels turn out to be very different from those widely agreed (or indeed if our methods themselves are different, though the labels end up the same) we would look carefully at the arguments presented for the different labelling. At the same time, those in the wide agreement would look at our arguments. Neither the methodologist, nor general practice, has final authority. We are engaged in a complex co-operative investigation to seek the methodological truth. As usual, we need a consensus judgement, and the truth may elude us.

(ii) We could use historical evidence of the relative success of the investigations based on some methods to justify preferring them (as long as we have the use of other methods, justified in other ways; otherwise this is viciously circular). History suggests that some methods are our best present methodological conjectures concerning effective ways of investigating phenomena; they seem to have worked better than their competitors. 'Better', in this context, is the word which shows that we are using other methods, not justified historically, to assess the success of these methods. It is not chance that physicists, consistently using these methods, have made more progress than astrologers.
   This claim is not circular, as (i) was. We are making a factual methodological claim.
But can success - progress - be assessed?  Given the common aim of {Truth}, can the set of propositions in physics be compared to the set in astrology, in terms of their relative success in achieving the aim?  If it can, then the methods used in Physics can take some of the credit. Or can the more recent set of propositions in physics be compared with an older set?  If what has changed is primarily the methodology, then again this could take the credit.
Relative success can be assessed, though not mechanically.  Consider personal conviction - faith based on revelation as a method of seeking the truth.  Theoretically, it could work.  But in practice it has been spectacularly unsuccessful.  It has consistently led to claims later accepted as false; it has led to novel predictions later accepted not to have occurred; it has led to the need for ad-hoc adjustments.  Repeatedly the beautiful ships of the believers have foundered on rocky facts.  Our non-time-dependent standards provide a way of assessing relative success.
   One desperate escape route for low-rated investigators is to deny the importance of the everyday interpersonal consensus - to deny that the facts are the facts. Mr.Cerullo could insist that a woman has been miraculously cured of her cough, despite interpersonal agreement that she still has it. But without this relatively basic consensus, we become unable to distinguish sanity from insanity; reality fragments.

(iii) Given our non-time-dependent standards, we can use evidence about human behaviour (Human Nature HN ) to support the use of some methods: M 20,23,25.

 

Conclusion

   The reader may feel that this is all rather excessive.  "Aren't there just a few key methods, perhaps for theory choice?", she may feel.  This impatience with the unwieldiness of such a long list of methods is understandable, but inappropriate.  Demarcation, for example, has proved to be an intractable problem if we focus on just a few methods.  An area of investigation may be assessed as low in rationality, not because of dramatic failure to use just one method, but because of failure to use a whole set of the best available methods.

Back to philosophy of physics map page