This post is now superseded by:
Principles of science and ethical guidelines for scientific conduct (v8.0)
Every day, numerous statements about how things relate to each other can be seen and heard everywhere. Science is normally thought of as the way of conduct that provides certainty about such relationships. However, beliefs and unsubstantiated statements can often be seen, also within science.
So, what are the principles of science then?
It is hard to say, as a well-defined set of principles for science does not seem to be readily available.
That position is supported by the following quote from National Academy of Sciences:
“The basic and particular principles that guide scientific research practices exist primarily in an unwritten code of ethics. Although some have proposed that these principles should be written down and formalised, the principles and traditions of science are, for the most part, conveyed to successive generations of scientists through example, discussion, and informal education.”
Ref.: Responsible Science, Volume I: Ensuring the Integrity of the Research Process; Panel on Scientific Responsibility and the Conduct of Research
What can be found in abundance, however, are codes of conduct like Singapore Statement on Research Integrity, EPA’s Principles of Scientific Integrity, Max Planck Society – Rules of good scientific practice or The European Code of Conduct for Research Integrity. But none of these provides a well-defined set of basic principles for science.
It also seems that even a highly influential work within the philosophy of science: “The logic of scientific discovery”, by Karl Popper, does not provide a set of well-defined principles for science. See section “3 Perspective on definitions and truth” in this work for an elaboration of this view.
The many controversies about scientific issues indicate that it would be beneficial to have the basic principles of science defined. Unfortunately, it is not obvious what these principles should be.
This work is based on the assumption that a set of principles can be defined – and is nothing less than a bold attempt to provide a set of fundamental principles for science. Principles that can be used to distinguish knowledge from beliefs – or to recognise “fake news” or “alternative facts” for that matter.
The principles provided here have not been taken out of thin air. Some principles may be recognised as principles phrased in various ways in various sources. Others are distilled from existing international standards. However, this is an original work that provides a unique set of well-defined principles for science.
This work itself, or parts thereof, can be proven wrong simply by identifying a flaw, a logically invalid principle or a flawed definition. It can also be proven wrong by identifying a concept known to be true that can not be put forward in a way that complies with all relevant principles or a concept known to be wrong that complies with all relevant principles.
The idea that a set of well-defined principles for science does not already exist may also be wrong since it is hard to prove that something does not exist. Anyhow, that idea can also be proven wrong simply by providing a link to such principles.
The reason why most terms are defined within this work is that there are many different dictionaries available at the fingertips of any reader. This work cannot rely on undefined terms or terms having various definitions, as even slightly different definitions will change the conclusions that can be drawn on the basis of this work and possibly make it inconsistent or logically invalid.
The first section of this work is self-contained and provides the principles and associated definitions. The second part of this work provides the essential arguments for each principle. The third section contains the perspective on definitions and truth that is applied in this work.
This work may be reproduced on the condition that the reproduction includes a link to the original site: https://principlesofscience.wordpress.com
1 The principles of science
Science or Fiction´s principles of science (v7.5)
§1 A scientific argument consists of clearly stated premises, inferences and conclusions.
§2 A scientific premise is verifiable. Premises and their sources are identified and readily available for independent verification.
§3 A scientific inference is logically valid.
§4 A scientific conclusion is deduced by application of axioms, definitions, and theorems or measured properties and scientific concepts that have already been verified or validated.
§5 A scientific concept consists of statements that are logically valid conclusions deduced from premises that are themselves logically valid conclusions, axioms, definitions or theorems.
§6 A scientific concept is well-defined and has a well-defined capability of prediction within a well-defined context.
§7 A scientific concept can only be validated by comparison of predictions deduced from that concept with measurement results. Whenever predictions differ from measurement results, by more than the combined uncertainty of the measurement results and the claimed capability of the concept, there must be something wrong with the concept – or the test of it.
§8 A scientific concept can only be referred to as validated for the context covered by the validating tests.
§9 A scientific statement is based on verifiable data. Data and precise information about how that data was obtained are readily available for independent verification. Whenever data are corrected or disregarded, both uncorrected and corrected data are provided together with a scientific argument for the correction.
§10 A scientific measurement report contains traceable values, units and stated uncertainty for well-defined measurands in a well-defined context.
§11 A scientific prediction report contains values, units and claimed capability for well-defined measurands in a well-defined context.
To keep traceability to a full account with definitions and explanation, these principles can only be reproduced on the condition that a link to this site is included: https://principlesofscience.wordpress.com/2017/02/26/the-principles-of-science-v7-5
Definitions for The principles of science
|argument: a conclusion inferred from a set of premises|
|attribute: a characteristic used to describe or define a thing|
|axiom: a statement that is self-evidently true and accepted as a true starting point for further deduction|
|belief: an acceptance that something is true without proof|
|calibration: comparison of a measurement with a reference having a known uncertainty|
|capability: maximum difference between predictions and measurements|
|clearly stated: stated in a manner that is only open to the intended interpretation|
|comparison: quantification of the difference between|
|concept: any expression of a relationship between two or more measurands|
|conclusion: a statement inferred from one or more premises|
|context: a set of those things that have an influence on an observed, measured or predicted value|
|contradict: demonstrate that a statement is not true|
|correct: replace a measured or predicted value with another value|
|data: measured or predicted value of a measurand or relationship between measurands|
|deduction: a combination of premises into a conclusion by means of mathematics and logic|
|definition: a set of distinguishing characteristics|
|disregard: remove a value from a series of data that is used as a premise|
|document: an identified collection of words, numbers and symbols|
|false: a statement that can be contradicted by a sound argument within the defined context|
|hypothesis: a propounded statement or concept that has not been verified or validated|
|independent: not under influence of the party propounding a concept|
|inference: logical connection between premises and conclusion|
|logically valid: the truth of the premises guarantees the truth of the conclusion – it is impossible for the premises to be true and the conclusion nevertheless to be false.|
|mathematics: a consistent and logically valid system of symbols and operations on these symbols|
|measurand: well-defined property that can be observed or quantified by a measurement|
|measure: quantify a measurand by establishing the ratio between that measurand and a reference that serves as a unit – and assign a number representing that ratio, and the associated unit, to that measurand|
|measurement (result): a measurand quantified by a value and an associated unit|
|nature: any thing and any relation between things|
|observe: conclude if an attribute is in accordance with a definition|
|precise information: sufficient for replication by an independent person using equal tools|
|prediction: quantification of a measurand without any foreknowledge about an eventual measurement result|
|premise: a statement used to infer a conclusion|
|property: an attribute that can be observed or measured|
|prove: demonstrate the truth of a statement by means of axiom, definitions and theorems.|
|readily available: available without further request|
|reference: a measurement device or procedure that has an unbroken chain of calibrations to the definition of the unit|
|relationship: a quantified change in measurand A is followed by a quantified change in measurand B|
|science: systematic discovery of truth|
|sound: a conclusion that is logically valid and based on true premises|
|source: identified document containing a premise|
|statement: a logical proposition that can be either true or false within the defined context|
|test: an activity that can verify or validate|
|theorem: a conclusion that has been proven and that can now be used as the basis of other proofs.|
|thing: whatever that can be defined|
|traceable: having an unbroken chain of calibrations to the definition of the unit|
|true: a statement that can not be contradicted by a sound argument within the defined context|
|uncertainty: quantified accuracy|
|unit: a well-defined quantity that has one unique value|
|validate: demonstrate the truth of a concept within a well-defined and applicable context|
|verify: demonstrate the truth of|
|wrong: not true|
2 Arguments for the principles of science
It should be noted that one of the ideas with this work has been to provide fundamental principles of science – in a compact manner. A significant effort has been invested in limiting the amount of text to an essential minimum.
Regarding §1 A scientific argument consists of clearly stated premises, inferences and conclusions.
The constituents of an argument can be recognised in §1 and the associated definitions.
Within science, it should be possible to verify that an argument is sound – that the argument is based on true premises, and that the truth of the premises guarantees the truth of the conclusion.
An essential characteristic of science is that arguments should be independently verifiable. To be able to verify that an argument is sound, the intended interpretation must be clear. An argument that is open to multiple interpretations can not be verified by an independent party as it can not be known which interpretation is the correct one to verify.
Regarding §2 A scientific premise is verifiable. Premises and their sources are identified and readily available for independent verification.
A premise can only be verified if it is properly referred to. Both the premise itself and the source containing the premise should be identified, and the source should be available for verification.
If a premise can not be verified, the premise can only be accepted on the basis of a belief in the proponent of the argument.
Regarding §3 A scientific inference is logically valid.
If an inference is not logically valid, it follows from the definitions that the truth of the premises does not guarantee the truth of the conclusion – it is possible for the premises to be true and the conclusion nevertheless to be false. Hence, the conclusion can then only be accepted on the basis of some kind of belief.
Regarding §4 A scientific conclusion is deduced by application of axioms, definitions, theorems or measured properties and scientific concepts that have already been verified or validated.
A scientific conclusion may be applied in an argument for or against a propounded statement or concept, or as part of a scientific concept.
A logically valid construction that ends up in a conclusion has to be based on something. In the case of abstract constructions like mathematics, the basis for the construction will be axioms, definitions, and theorems.
In the case of constructions intended to provide a correspondence between an abstract construction and observations and measurements of nature (like physics), the axioms, definitions, and theorems may be about nature or about the correspondence between an abstract construction and nature. In this case, the construction may also be based on observed or measured properties or scientific concepts that have already been verified or validated.
As an example, it will normally be acceptable to base a scientific conclusion on a concept like Newton´s laws of motion within their validated context, or on a measured property like the gravitational acceleration (approximately 9,8 m/s^2 on earth). The application will dictate how accurate that measured property will have to be – whether 9,8 m/s^2 is sufficiently accurate or if a more accurate value is required.
Regarding §5 A scientific concept consists of statements that are logically valid conclusions deduced from premises that are themselves logically valid conclusions, axioms, definitions or theorems.
The entire concept will have to be a logically valid construction that has a well-defined and true basis. If there are any logical fallacies in a construction, the result will be that the concept can only be accepted on the basis of some kind of belief.
A concept that is under construction, or has not yet been validated, should be clearly identified as an hypothesis to avoid premature application of the concept.
Regarding §6 A scientific concept is well-defined and has a well-defined capability of prediction within a well-defined context.
To facilitate independent judgment, the concept itself will have to be well-defined. If the concept is not well-defined, it can not be tested by an independent party. An independent party will not know what to test and how to test it.
Concepts are only valid within a context. One example of this is classical physics: “Beginning at the atomic level and lower, the laws of classical physics break down and generally do not provide a correct description of nature.” (Ref.: Wikipedia; classical physics; at the date of publishing this work). To facilitate judgment of a concept, the context for which the concept is claimed to work well will have to be defined.
Many concepts got a capability of prediction of the value of a measurand, but not exactly. A concept may have a capability of prediction with some uncertainty. To facilitate judgment of a concept, the capability of the concept will have to be defined. If not, there is no way to tell if the concept performs as claimed or not, or whether it is useful for an intended use or not.
Regarding §7 A scientific concept can only be validated by comparison of predictions deduced from that concept with measurement results. Whenever predictions differ from measurement results, by more than the combined uncertainty of the measurement results and the claimed capability of the concept, there must be something wrong with the concept – or the test of it.
Any collection of words, numbers, and symbols is an abstract construction that may or may not correspond with observations and measurements of nature.
Within many areas of human expressions, like in politics, religion, love, hate, humor or whatever; it may not matter if an expression corresponds with nature. An essential characteristic of a useful scientific concept, on the other hand, is that of correspondence between predictions of that concept and observations and measurements.
Even if a concept complies with §1 to §6, there is no guarantee that the concept is a complete construction that also provides a correspondence between that concept and observations and measurements of nature. Without testing it, it can not be known for sure that the concept is complete, that there are no errors in it, that the concept is correctly constructed or that the concept actually has the claimed capability of prediction.
The only way to know that a concept performs within the claimed capability, within a defined context, is to deduce predictions from that concept, measure nature within the same context and see if the difference between predictions and measurements is within the claimed capability of the concept.
In judging the results of the test, the uncertainty of the measurements will have to be taken into account. Repeated tests are required to ensure that the results are representative.
There are many ways to adjust a concept to match observations and measurements. Many kinds of curve fit, parameterisation, change of definitions or addition of hypotheses can be used to adjust a concept. The problem with adjustments, however, is that adjustments may hide that the concept does not have the claimed capability of prediction.
Some concepts may need some kind of basic calibration and adjustment, but if a concept really has the claimed capability to predict the value of a measurand, there should be no reason to adjust the concept to a particular test.
The reason why it is so useful to compare predictions with measurements is that all kinds of adjustments of the concept to match measurements are logically impossible. It is impossible to adjust a concept to match something that is not yet known. Prediction excludes all kinds of adjustments of the concept to match the measured values.
There may be other ways to validate a concept, but all other ways leave a possibility that the concept has been adjusted to match measurements. Hence all other ways to validate a concept should also be followed by a scientific argument proving that the concept has not been adjusted to match the measurements of that particular test. Without such proof, the concept can only be accepted on the basis of a belief that the concept has not been adjusted particularly for that test.
If a concept is not tested by an independent party, the concept can only be accepted on basis of a belief in the party propounding a concept.
Regarding §8 A scientific concept can only be referred to as validated for the context covered by the validating tests.
A test is performed within a context. Obviously, the test is only valid for that context. As a principle, the concept can only be referred to as validated for the context covered by the validating test.
It may be that interpolation or extrapolation to some extent can not be contradicted by a sound argument, but that is not normally the situation.
However, the party propounding a concept might be able to put forward a scientific argument for the validity of interpolation or extrapolation, and it might be that no opponents are able to put forward a counter argument. Anyhow, extrapolation or interpolation should be followed by a scientific argument.
Regarding §9 A scientific statement is based on verifiable data. Data and precise information about how that data was obtained are readily available for independent verification. Whenever data are corrected or disregarded, both uncorrected and corrected data are provided together with a scientific argument for the correction.
Whenever a statement is based on observations or measured or predicted values, the data should be readily available for independent verification. If not, the statement can only be accepted on the basis of a belief.
There might be errors in the experiment that produced the data. Such errors can possibly be revealed by an investigation into how the data was obtained or by an independent replication of the experiment.
Anyhow, the statement can only be verified if precise information about how that data was obtained is readily available. If not, the statement can only be accepted on the basis of a belief.
Finally, it can be irresistible to disregard or correct data. There may be scientific arguments for doing that. If so, those arguments should be verifiable. If not, data should not be corrected, discarded or disregarded.
Regarding §10 A scientific measurement report contains traceable values, units and stated uncertainty for well-defined measurands in a well-defined context.
Obviously, a measurand will have to be well-defined, how else can anybody know exactly what has been measured?
Also, the measurement result will also have to be provided as a value together with the associated unit. A value without a unit is meaningless.
By using a unit in accordance with the International System of Units, the unit will already be well-defined. If the unit is a non-standard unit or even a hitherto unknown unit, the unit will have to be properly defined in the measurement report.
Whenever a measurement is performed by some kind of measurement device, the measurement device should be traceable by an unbroken chain of calibrations to the definition of the unit. Without a traceable measurement device, there is no way to know if the measurement is accurate, there is no way to quantify the uncertainty of the measurement.
Regarding the uncertainty of a measurement, the introduction to the following free and readily available guideline: Guide to the expression of uncertainty in measurement; JCGM 100:2008, explains why quantification of uncertainty is essential: “When reporting the result of a measurement of a physical quantity, it is obligatory that some quantitative indication of the quality of the result be given so that those who use it can assess its reliability. Without such an indication, measurement results cannot be compared, either among themselves or with reference values given in a specification or standard.”
For the principles provided in this work, it is regarded sufficient to state that it is essential that the uncertainty of a measurement is provided in the measurement report. Obviously, there are benefits in providing the uncertainty in accordance with an international standard or guideline. By not providing the uncertainty in accordance with a standard or guideline, there is a risk that the measurement report is regarded insufficient and that no judgments can be made on basis of that report.
Finally, it is also essential that the context for the measurement is well-defined. All the things that are known to have an influence on the value of the measurand should be identified.
(This principle has been based on section 7.2.1 in the freely available international guideline: JCGM 100:2008; GUM 1995 with minor corrections; Evaluation of measurement data — Guide to the expression of uncertainty in measurement.)
Regarding §11 A scientific prediction report contains values, units and claimed capability for well-defined measurands in a well-defined context.
This principle is an analog to §10 about measurement reports, this should be no surprise since predictions are supposed to be comparable with measurements. A claimed capability may be expressed and documented in the same way as the uncertainty of a measurement.
3 Perspective on definitions and truth
As mentioned in the introduction to this work, even a highly influential philosophical work like “The logic of scientific discovery” by Karl Popper does not provide a set of well-defined principles for science.
The first version of this work (“10 theorems for ideas about how things work”) started out as an attempt to identify a set of principles, or methodical rules, as established by Karl Popper. That turned out to be a bit challenging, as his methodical rules were not defined and identified in a clear manner.
The following quote may shed some light on why:
“It is, I now think, the fact that most philosophers regard definitions as important, and that they have never taken my assurance seriously that I do regard them as unimportant. I neither believe that definitions can make the meaning of our words definite, nor do I think it worth bothering about whether or not we can define a term (though it may sometimes be moderately interesting that a term can be defined with the help of terms of a certain kind); for we do need undefined primitive terms in any case.”
Ref.: The logic of scientific discovery; Page 463; (Addendum, 1968)
In this work, that view is opposed by the position that definitions are of uttermost importance for an evaluation of the truth of a premise, inference or conclusion. Take for example the symbol: “+” in mathematics. Without a definition, it would just be a meaningless cross.
Even though it may seem that a definition can never be precise enough for all possible readers, the set of principles provided in this work is based on the axiom that: It is possible to define terms so precisely that a propounded statement is only open to the intended interpretation. If that is not the case for a particular statement in a particular context, a meaningful argument about that statement will not be possible.
Another issue with The logic of scientific discovery is the perspective on truth, as illustrated by the following quotes:
“It should be noticed that a positive decision [test result] can only temporarily support the theory, for subsequent negative decision [test results] may always overthrow it. So long as theory withstands detailed and severe tests and is not superseded by another theory in the course of scientific progress, we may say that it has ‘proved its mettle’ or that it is ‘corroborated’.. by past experience. Nothing resembling inductive logic appears in the procedure here outlined. I never assume that we can argue from the truth of singular statements to the truth of theories. I never assume that by force of ‘verified’ conclusions, theories can be established as ‘true’, or even as merely ‘probable’.”
The perspective on the term truth that has been taken in this work is that, if all definitions are in place and relevant tests have been performed, we are able to conclude if it is true that a well-defined concept really has the defined capability within a well-defined context.
Engineers will probably be familiar with this definition of truth. Engineers will be used to demonstrate the truth of their constructions – to verify and validate that a construction has a defined capability within a defined context. In particular, verify and validate are both terms that are also used in the widely used international quality standard ISO 9001.
It should be noted, however, that even though a concept can be true by the definition used in this work, another concept that has a better capability of prediction or is valid for a broader context may be eventually be discovered or invented.
By this definition, Newton´s law of universal gravitation can still be regarded to be true in the sense that the concept has a definable capability of prediction within a definable context. While Einstein´s general relativity can also be regarded to be true, but that concept has a better capability of prediction within a wider context.
It should also be noted that even if a concept is true, it can still be useless. It may be a true prediction that the precipitation at a defined location on a defined date will be between 0 and 1000 mm, but that prediction will also be of no value.
An example of this perspective on truth can be a television system. A television system can transmit a movie via a fiber and display it on a screen (whenever everything in that system performs in accordance with its design). That functionality is true – it can not be contradicted by sound argument.
It is remarkable that a lot of things have to be true for television system to work properly, more things will have to be true than an individual person can fully understand. However, even a kid can tell if it is true that it works.
This work can only be reproduced on the condition that the original source is identified by a link to https://principlesofscience.wordpress.com
This is a work by “Science or Fiction?” with invaluable support and scrutiny by “Gnomish”. – See how it evolved.