I am not alone in  – in fact most of the atheist agree with me on this – being an agnostic atheist regarding a deist god, and there is a principle behind this that I call “the principle of confirmation”.

In philosophy this principle is known by several expressions, like “evidentialism”. Michael Martin calls it the principle of agnosticism1. In logical positivism a similar principle can be find by the name “principle of verificationism”:

“Since the meaning of a word is determined by its criterion of application (in other words: by the relations of deducibility entered into by its elementary sentenceform, by its truth conditions, by the method of its verification), the stipulation of the criterion takes away ones freedom to decide what one wishes to “mean” by the word. If the word is to receive an exact meaning, nothing less than the criterion of application must be given; but one cannot, on the other hand, give more than the criterion of application, for the latter is a sufficient determination of meaning. The meaning is implicitly contained in the criterion; all that remains to be done is to make the meaning explicit.” Rudolf Carnap: “The Elimination of Metaphysics Through Logical Analysis of Language2

For those, who are versed in philosophy, I have to emphasize, that I do not just accept the principle of verification as the criterion of meaningfulness, but also respectively confirmation as the criterion of accepting the truth. In short: a sentence is meaningful, if it can be tested at least in theory, and it shall be regarded to be true if and only if it is confirmed, and it shall be regarded as false, if and only if it is falsified. Moreover I accept the weak verification criterion, i.e. confirmation does not have to be a proof (by deduction) and falsification does not mean absolute refutation.

“If by verification is meant a definitive and final establishment of truth, then no (synthetic) sentence is ever verifiable, as we shall see. We can only confirm a sentence more and more. Therefore we shall speak of the problem of confirmation rather than of the problem of verification.” Rudolf Carnap: “Testability and meaning”3

So in case of synthetic statements4 verification can only expected to consist of confirmation5. The statement “god exist” is a synthetic sentence, since it is stating that god exists as an entity in reality. I cannot interpret and never heard any meaningful interpretation that god would exist in another way. I hold this claim “god exists in another way” as a meaningless statement, I don’t know what it should mean. The “there exist” mathematical quantifier cannot be meant with gods claimed existence (that would then be an analytic statement), because it makes no sense to pray for an abstract concept, and to believe that an abstract concept created the Universe ex nihilo.

So verification means confirmation in the case of synthetic questions. Therefore it is not infallible and it is not certainty. This confirmation is the most that can be expected in empirical matters. Those, who claim certitude about synthetic statements, those claim something logically impossible, since synthetic statements cannot be strictly speaking proven, for reasons of logical grounds, and we cannot have certitude, especially about universal statements.6

First, because those base statements that would be the premisses of a deduction, namely our experiences are uncertain themselves. At least, when we create an interpretation of them. For example one cannot deduce the existence of something from any experience with certainty. This can always only be highly truthlikely7.

Second, because there is no logical deduction from one, or a finite number of singular experiences to a universal, theoretical sentence. For example even if we could be sure that there exists a finite number of black crows, we cannot deduce from this, that all crows are black.

Moreover, for the real existence of something, there cannot be an ontological (logical, or mathematical) proof, that does not even contain experiences. Such so called ontological arguments confuse analytic statements (like mathematics) and synthetic statements, that are about real existence of things. In a mathematical (ontological) proof we can only deduce something about an abstract concept within a so called axiomatic system. But that never means that it is likewise true in reality.

The famous example for this is that in the Euclidean geometry the sum of the angles of the triangle is 180 degrees, in the so-called Lobatchevsky-Bolyai geometry it is less, and  in the so-called elliptic geometry it is more. In the Riemannian geometry it may vary from triangle to triangle. But in reality, the real, physical space may be like any of these abstract geometries. Which one of them is true in reality is a synthetic question that cannot be decided without empirical data. Similarly, the existence of a real god cannot be deduced ontologically. Even if god existed in real. Therefore all ontological proofs of god are utter nonsense in the first place.

Regarding empirical science, it works with some simplification so that we make observations, and by this we can confirm the existence of some things, and other not. We may also dis-confirm the existence of some things, like that of a a perpeetum mobile that is very not likely to exist according to our highly confirmed knowledge.

The principle of science is that we accepts those things to exist, that are confirmed, and we do not believe anything about the other things. Science cannot start in the first place with accepting those things to exists that are not confirmed, because then what would confirmation be good for? It is not even a working method to accept these thing temporarily to exist, and then sort out those, that we can dis-confirm. This does not work, because we would need first to accept infinite many things temporarily to exist. Moreover, we needed to believe in many things that contradict each other. For example, if there was a historical person, and we do not know it yet, then we had to believe at the same time that she had 1, 2, 3, … children. Up to infinity. This would be absurd, even if we limited the infinity by a practical upper limit. It is better to admit: “we do not know how many children she had”.

Even if we avoided the error of accepting theories that contradict each other, there is the second problem that we cannot evaluate an infinite number of theories. And we cannot examine all the infinite number of hypothesis of non-existent thing just to refute all of them. So we would never finish up, and we would end up today still, to believe that there are infinite kind of things that exist. But we could not list up them, therefore we could not even start with refuting them.

As an illustration let us imagine, that we need to find and list all our relatives that are within the distance of 5 steps of relations from us. Let us consider one step our immediate relatives, the parents, the children, the brothers and sisters and partners.8  Now the evident search algorithm is to ask first our immediate relatives and ask them about their immediate relatives, and continue with that, and do this until we covered the distance of 5. This is the classical tree search algorithm. Because it is limited, there is no fatal problem with cycles.

With this algorithm we finish in finite time, and we mapped (“graphed”) up our relatives upto the distance 5. If there is a possibility of error in our method, we maybe would not find all the relatives, our would find false relatives. But we have reason to think that in most of the cases this method is quite good and finds our real relatives up to a distance of 5.

Another method could be that we first list up all human beings as our relatives, and then we investigate each of them, if they are relatives within the distance of 5. Now all intelligent being understands, that this is an unnecessarily cumbersome method, that is practically impossible. Even if the set of all human being is finite. On the other hand, the set of all hypothesis is infinite, and in this case the problems are not only practical, but are theoretically insurmountable. This is why this method does not work, and this is why we have to proceed step by step. The criterion of confirmation in science ensures that we proceed step by step, i.e. that we follow a method that works.

Now this step by step method sounds very conservative, old stylish. Therefore I have to emphasize that this concerns only science and knowledge. Here I think the right, actually the only possible method is to go step by step. Diametrically opposed to this I am relativist  and liberal in aesthetic an ethical questions, and I would not prescribe people to go step by step in these question. This opposing views is because I think ethics is subjective and therefore relative, and therefore there is no truth, and no method needed to discover the truth in reality about ethical questions. Science is about objective questions, ethics is about subjective questions.

1 Michael Martin: „Atheism: A Philosophical Justification”, Temple University Press, 1990, p. 29.

2 Rudolf Carnap: “The Elimination of Metaphysics Through Logical Analysis of Language, Erkenntnis, 60-81, 1932.

3 Rudolf Carnap: “Testability and meaning”, Philosophy of Science, Vol. 3, No. 4, Oct., 1936, pp. 419-471.

4 Synthetic statements are about reality. Analytic statements on the other hand are true independently of the contingent facts of reality, because of the very meaning of our concepts. According to logical positivism theoretical mathematics is analytic.

5 And this is not the confirmation of protestantism.:)

6 Universal statements are those that state something like “about all things of this sort this and this is true”. So they really state something about potentially infinite things. They include a universal quantization, like “all”, or “every”.

7 The word “verisimilitude” is the word of Karl Popper, because he maintained, we cannot assign real probabilities to theories, moreover that a theory that is probable is not really a good hypothesis. Interesting hypothesis are those, that are not probable, but still not falsified and have high “verisimilitude”. See Karl Popper: Conjectures and Refutations: The Growth of Scientific Knowledge, London: Routledge, 1963.

8 From a mathematical viewpoint it would be advantageous not to consider siblings as immediate relatives, because then we have short cycles in the graph of relationships. We would have bigger cycle anyway, because it is possible to marry a relative, especially a further one.