Science, hypotheses, and intelligent design

This is an additional post by Lars Johan Erkell made on September 8, 2021 in Swedish at the Biolog(g) site after the seven posts of his “Breakthrough for Intelligent Design?” series were made. He has requested that we post it here after the second post in that series, to explain his position in that post on what can be considered legitimate science.


A central point in the discussion I have with Ola Hössjer in the posts entitled "Breakthrough for intelligent design?" concerns the limits for what can be considered legitimate science. Since my responses to his comments on this matter are too long for the comments section, I have made a separate post on the subject.

The central question is how to formulate a hypothesis, that is, a suggested explanation for a phenomenon. Hypotheses are not formulated in an arbitrary way; you don't just grab something out of the air. A hypothesis should be based on prior knowledge. It should also be falsifiable, which means that it must be possible to show that it is incorrect. The scientific work then consists of testing the hypothesis. Whether it proves to be true or false, something new has been learned. The important thing here is that the hypothesis is formulated in such a way that it can be tested in practice with a clear-cut result. If you cannot do that, you cannot learn anything new. And if you cannot test your hypotheses, you cannot root your theories in the real world.

However, just because a hypothesis is falsifiable does not mean that it is scientific. For example, "the Moon is a cheese" is a falsifiable (and falsified) hypothesis. But pure nonsense. If we want to give it a scientific touch, we can formulate it as a classical syllogism:

Premise 1: The Moon is yellow, round and has craters

Premise 2: A cut cheese is yellow, round and has pits that look like craters

Conclusion: The Moon is a cheese

This conclusion is not just grabbed out of thin air; it is based on an analogy. But it is still nonsense. My point here is that just because a hypothesis is falsifiable, or is expressed in a formally correct manner, that does not mean it is plausible or even makes sense.

As an example of a falsifiable hypothesis that is scientifically acceptable, let us take a classic: Darwin's hypothesis of the common descent of life forms. It implies, among other implications, that older forms must lie below younger forms in the fossil record. In other words, a fossil of a reliably identified and reliably dated rabbit (or any other modern mammal) may not be found in Cambrian deposits, i.e. from a time long before the first mammals emerged. Despite millions of fossils being unearthed, nothing has yet been found to disprove Darwin's hypothesis. Still, in principle it could happen, and that would be a fundamental blow to the theory of evolution.


One can hear ID proponents argue that ID can also make falsifiable hypotheses. Stephen Meyer devotes an entire appendix to discussing the matter in his 2009 book Signature in the Cell. He writes on page 482: "… the theory of intelligent design makes predictions about the kind of features we are likely to find in living systems if they were in fact intelligently designed". What is important here is the wording "we are likely to find" - ID's predictions are thus based on what one might vaguely expect to find. Not on what one must find or on what one must not find, i.e. not on any compelling predictions.

This means that one cannot clearly falsify an ID hypothesis. As an example, let us take an ID-inspired hypothesis that Meyer puts forward: "The functional sequences of amino acids within amino acid-sequence space should be extremely rare rather than common". He argues that of all possible combinations of the amino acids that make up proteins, there should be very few combinations that result in functional protein structures. The first problem is that Meyer does not give a clear limit for what would be extremely rare and what would not, so the question is impossible to decide in practice - "extremely rare" can be interpreted in very different ways. The second, and bigger, problem is that there is no logical connection to a designer - how do we know that the designer decided that there must be only a few working combinations? We don't. So even if we could confirm the hypothesis, it would not tell us anything about whether proteins were designed or not.

This ID hypothesis is clearly not about proving ID, but about ruling out an evolutionary explanation. If it could be shown that there were extremely few working combinations of amino acids, this would be seen as an argument against the theory of evolution (which, by the way, it is not). In other words, Meyer tries to demonstrate "non-evolution" and sees this as a sign of design. This is a classic fallacy - the fact that one explanation does not hold up does not mean that another explanation must be correct.

Meyer in his appendix has a list of twelve "ID-inspired" predictions of which the above example was the last one. The other eleven suffer from the same weaknesses. Some of them are pure arguments against evolutionary processes and have nothing to do with design. Others assume that the designer is working with a particular type of design which you then want to identify. But you cannot know what type of design the designer uses, so what you are actually testing is the idea you have about how the designer should work. Without actual knowledge of the designer, it is impossible to know how to formulate precise hypotheses. Everything becomes pure conjecture.

As an example of empirical work based on a hypothetical designer, consider a study of the impact of intercessory prayer on the recovery of patients who have undergone a particular type of heart surgery.1 This is a large study with 1800 patients, carried out according to all the rules of the art. The end result was that no significant effect of prayer was seen. Does this mean that prayer has been shown not to work? Not at all. The results only show that intercession does not work with the particular experimental design chosen. Perhaps it would have worked with other patients, other experimental groups, other control groups, other prayer rituals, other people praying? We don't know. Since we have no concrete knowledge of the designer, we do not know how the experiment should be set up to show with certainty the effect of intercessory prayer. We can guess endlessly. This is not how good hypothesis testing is done - that requires precise, testable hypotheses.


Let us leave the hypotheses and move on to what is required of a scientific explanation. Explaining something means clearly demonstrating the causes of some phenomenon; you connect cause and effect. You also have to show how the cause leads to the observed phenomenon. A good explanation, then, involves linking a phenomenon to a known cause through a known mechanism. Darwin's theory of evolution is a good example. His idea of common descent was quickly accepted by his contemporaries; it was so obvious that it explained the patterns of nature well. But with natural selection it was not so. Darwin could not explain the mechanisms of heredity, which was a prerequisite for understanding natural selection. That is why not many people fully accepted his theory. However, by the turn of the 20th century, the mechanisms of genetics were starting to be elucidated, so now the process of evolution was being understood. But it was still not clear exactly how evolution happened. It wasn't until population genetics had developed sufficiently in the 1930s and 40s that the theory became widely accepted. By then, the whole chain was clear: known causes (heredity and variation) caused a known phenomenon (evolution) through a known mechanism (natural selection).

Today we can understand the entire causal chain. We can understand it at the molecular level as well as at the population level. We know the molecules and processes that cause evolution. We can observe and measure them, which allows us to have an evidence-based discussion about the importance of different factors, and how well different models fit reality. That is how it should be with a good scientific explanation. It also allows us to formulate and test new hypotheses based on the knowledge we already have - for example, can cooperation and helpfulness be explained in evolutionary terms? It turns out that it can. The theory of inclusive fitness provides a robust evolutionary explanation.

Compare this to an ID explanation of how a complex biological structure would arise. The cause, the designer, is unknown, and so are the mechanisms by which the design would be implemented. Thus, we cannot assess how plausible the ID explanation is, or how it compares to other explanations. An explanation based on unknown causes and unknown mechanisms is, strictly speaking, no explanation at all - it doesn't make us any wiser. It provides no concrete knowledge. It has no explanatory value. Nor does it provide a reliable basis for further hypotheses. If you are already firmly convinced that there is a designer, it may work as a religiously based explanation, but it is scientifically uninteresting.

The fuzziness of ID explanations is characteristic. It is also reflected in the lack of precision in the concepts used. For example, "complex specified information" is not defined in a way that makes it measurable or practically useful. Similarly, it is impossible to clearly determine whether a structure is irreducibly complex or not - distinct criteria are lacking. It is obviously impossible to conduct meaningful research with such fuzzy concepts. Yet ID proponents refer to these concepts as if they were clear, meaningful and established. But they are definitely not.

The conclusion is clear: ID cannot provide the kind of mechanistic explanations that are the basis of a scientific theory. Thus ID cannot compete with established science, and can never be a scientific alternative. On the other hand, ID could be a complementary explanation on another level: one is free to imagine a governing intelligence behind all the mysteries of existence, while science provides the mechanistic explanations of how the world actually works.


1) Benson, H., et al: Study of the Therapeutic Effects of Intercessory Prayer (STEP) in cardiac bypass patients. American Heart Journal (2006) 151, pp. 934-42