The Probability of Life

Note that this is not an argument regarding Darwinian evolution. It is primarily a probabilistic argument regarding chemical abiogenesis which must occur before evolution can begin. The thought experiment will even attempt to help such a process by thinking of it as a search engine that is actively looking for life and for the information that life requires.

This will be an argument about genetic information which is then used to generate biological products. DNA which stores genes has no chemically preferred sequences; it is a pure information storage system. Therefore DNA itself does not modify the odds on the information that it carries. DNA functions like a book that can contain letters and words on its pages, or computer files that contain bytes in any sequence. This data-agnostic charactistic of DNA has been demonstrated by encoding text, pictures and video into DNA and then accurately decoding them back out again. Like that demonstration, the DNA of life contains complex purposeful biological information.

Where did the specified complex information that life requires come from? If it came from natural causes, then these had to be probable, or at least possible. Could the universe find the information of life in the random conjuctions of natural matter? If it was a search engine, could it run fast enough and have enough time to find it?

Probabilistic reasoning about pure information is especially appropriate. Information theory depends on reasoning just like this. The modern computer tech that we use all the time depends on probabilistic methods for successful information storage, communication and many calculations.

However, note that natural processes uniformly degrade information. Information systems that do not account and correct for these errors will fail. Self-replicating systems that survive therefore must require built-in error detection and correction mechanisms. The argument here will mostly ignore this issue and focus on the problem of finding and gaining useful information. However, without error handling the rate of information loss by random processes greatly exceeds all information gain. Therefore, this information retention problem strengthens the arguments here by a substantial amount.

By conclusion, if abiogenesis is unable to happen, then evolution never would have happened afterwards. Since we are alive, if we can infer that life itself could not have begun by natural processes then we must inevitably accept that it must have come to existence by a non-materialistic means.

Living things are complex information storage and usage systems. Information of this type according to all uniform human experience only comes into existence from an intellect. Specified complex information was needed before life in this universe could begin. Therefore, by logical reasoning, the most likely cause for this information would be from an intellect external to the natural material universe.

If life came from a non-material source and if it was caused by an intellectual agent, then it is also a reasonable idea that speciation may have been caused similarly by the same external agency. And this then is not Darwinian.

The multiverse has been suggested to make our existence be more likely. See The Multiverse God. I suggest however that my arguments here will make even a multiverse be probabilistically insufficient and useless.

Wikipedia: Planck time and length

It is 1.86 x 10^43 = 1 / (5.39 × 10^44 seconds)

How Many Atoms Are There in the Universe?

Estimated from 1e78 to 1e82 atoms in the universe.

Note that some people imagine that life on earth was seeded from elsewhere in the universe. (This is called panspermia.)

This number of 1e143 chances for life appearing in the universe already includes all the (supposedly additional) chances of life from panspermia. Panspermia is already fully accounted for in my thought experiment. Panspermia does not help to solve the problem of life.

Again:

  4.29e17 seconds x 1.86e43 intervals x 1e82 atoms

= 7.98 x 10^(18+42+82)

= 7.98e142

And rounded up to be generous, is 1e143 chances. This number includes all chances for life to appear over all time anywhere in the entire universe.

This life search engine is (best-case) assuming that new random conjunctions of matter that might be a new life are continually and rapidly occurring.

This thought experiment is also generously giving each moment the ability to self-recognize a successful life search candidate even though recognition like that is not in these momentary events. Actually, a separate recognition engine outside of the universe would be required for this because the evolutionary process of natural selection does not exist before life and/or self-replication. However to simplify things, I will assume that this comes for free.

Also, this is quite unrealistically generous for search chances for the universe because each of these chances is at the level of one atom. Many, many atoms together would actually be needed for any one living thing. So there should actually be less chances given. (At least divided by the number of atoms in a minimal living thing.) However, the difficulty of the search is so great that being this generous won't even weaken the argument here.

So actually there would be 4.29e17 seconds since the beginning of time. And actually there would be 1.86e43 Planck time intervals in a second.

Multiplied together these would be 7.98e60 which I am rounding up to a larger total of 1e61 time increments since the very beginning.

Wikipedia: Minimal genome

Note however that 256 is a theoretical minimum genome and no actual organism exists with a genome this small.

In the case of our organism with the minimal genome, we have roughly equated these 256 uniquely different genes with 256 different proteins. Note therefore the proteins must be sufficiently complicated that there are no duplications between them (because otherwise we wouldn't have needed all of the genes).

Wikipedia: Protein

Again, orders of magnitude is the size of the exponent in scientific notation. It is the size of the count of digits in the number.

The universe has 1e143 chances to find life. However, the likelihood of finding life by the measure here is 1 out of 1e768. (That is 1 followed by 768 zeros!) This gap is 768 - 143 = 625 orders of magnitude. (That is 1 followed by 625 zeros!) It is 1e625 times too improbable to have happened.

Note that the complexity of the proteins in the experiment are not being measured. The measure is about the whether a random protein of indeterminate form has a function and the rarity of those with function.

Functional proteins from a random-sequence library

Living systems are highly integrated, and important life processes depend on coordinated function. Said another way: life fails if the protein functions are not coordinated.

There are some functions are done by single protein engines. However, there are very many cases where multiple proteins coordinate to complete a function. Hemoglobin, which carries oxygen in our blood, is made of four protein units. A more dramatic example is the bacterial flagella which is a propulsion organelle composed of 20 different proteins; and it requires every one of them for correct function.

A requirement for two coordinated proteins multiplies the odds against it. 1e77 x 1e77 which is 1e77^2 = 1e154   Since the chances in the universe = 1e143, then 154-143 = 11 orders of magnitude implausible in our whole universe.

Other papers suggest proteins are not quite so rare at 1 in 1e40. This lesser rarity could be true for selected proteins because they come in a large range of lengths.

(This rest of this note will make more sense after development of the line of reasoning in a few more paragraphs.)

The details may be different, but the argument that follows will still hold because even this smaller value is still strikingly rare. The all-searching universe could at best find three coordinated proteins of this rarity 1e40 x 3 = 1e120 which is less than 1e143.

However, the rarity of the full genome would be 1e40^256 = 1e10240, which is 10240 - 143 = 10097 orders of magnitude implausible.

See also New Book: New Proteins Evolve Very Easily

Note that some think that functionally useful proteins would not be difficult for evolution to find because there are many possibilities for useful proteins. The thinking is that since there should be many target patterns for functional proteins then it is easy to get the new proteins that are needed in new organisms. However we can see from other things that is not so.

If we look at "convergent evolution", we see cases of organisms which could not have come from common ancestors, but they have common features. Sometimes certain distinctive proteins are in common in those organisms as a part of the functioning of the feature. One of the most famous of these is the protein prestin which used for echolocation in both bats and whales.

As Günter Bechly has noted, if there were many targets for functional proteins, then we should not see convergent evolution of proteins. If functionally equivalent non-similar proteins were available, then they should have been found instead of the identical proteins. The fact that we actually see these particular proteins in common between unrelated organisms is very strong evidence that they are the only ones that will do. Therefore from this we can know that useful protein targets actually are extraordinarily rare.

See Convergent evolution and Echolocation in bats and whales based on same changes to same gene

A Dentist in the Sahara: Doug Axe on the Rarity of Proteins Is Decisively Confirmed. “… only 1 in 1e77 sequences corresponds to a functional fold/domain within the protein…”

Estimating the prevalence of protein sequences adopting functional enzyme folds “… implies the overall prevalence of sequences performing a specific function by any domain-sized fold may be as low as 1 in 1e77…”

1e77^256 = 1e19712

Chemical abiogenesis is hopeful self-deception because the chemistry is adverse to success - in an RNA world or in any other chemistry. There is poisoning of the reaction with by-products, lack of purification of the reaction products and lack of protection from destructive environments. Even if some researcher intelligently determines a candidate for a self-replicating molecule, this does not make abiogenesis be more likely because the adverse chemistry still dominates in all natural contexts. So this chemical approach ignores that chemistry always works to make such a process fail.

This is even acknowledged in the famous Miller-Urey experiment. The molecular products that they were able to generate had to be rescued from the source materials to avoid ruin. No such laboratory rescue would be available to any naturalistic abiogenesis.

In addition to the raw chemistry problems, unless this self-replicating RNA would have virtually perfect reproduction, its copies will very quickly degenerate and become non-viable. This is what lab experiments always show. This is a practical re-statement of the original problem.

Note that philosophically, even if an imagined chemistry of self-replicators could be achieved, they do not change the context because they have no instinct to survive like living things do. They will just as “happily” die (that is: fail) as they will live. They won’t try to live, so they won’t survive. The “will” of living things to continue living is a non-trivial component of life.

As I heard Elon Musk say in an interview, a repeatable process to make a space rocket is an order of magnitude more difficult than making one rocket, and a machine (a factory) to make make rockets is again an order of magnitude more difficult. He was not even suggesting that the rocket was a self-replicator.

This was not meant to be an argument about the adequacy of evolutionary processes in living things. However since evolutionary processes are suggested to function in pre-living chemistry, the problems on the theory must be addressed. The model for transition of a car to submarine is analogous to the assumed evolutionary transition from land animal to whale. The non-living thought experiment (the car to submarine) shows the extreme unlikeliness of a successful path between the two such types in living entities and the very certain lack of fitness in the intermediaries.

Changes like this had to have occurred in any path from a first self-replicating molecule (non-living chemistry) to LUCA (the Last Universal Common Ancestor, which was living). Therefore, the necessary path to life had to have gone through states that were non-viable.

An analogy:

Imagine you are to implement a procedure for changing a novel written by Dickens into a novel written by Stephen King. You are allowed to do this by randomly adding, removing, or changing around words and then selecting for the change being a meaningful story.

Could this be successful?

It is a ridiculous thought experiment with no chance of success ever. A key thing is that the meaningful story is an analog for a viable evolving organism. Evolution has similar likelihood of success.

(Idea from an interview of Günter Bechly.)

Even if a simple self-replicating molecule existed since the beginning of the universe, and even if it was trying combinations of improvements as fast as possible (at Planck intervals), it would (by definition of abiogenesis) still have to be able progress up to the complexity of life as we observe it. The same rarity argument from before still holds. There would still only be 1e143 opportunities to come up with anything in the known universe. However as previously pointed out, the rarity of life from an over-simplified view of proteins is 625 orders of magnitude too rare to be produced in our material world, and from a more realistic view of protein rarity, existing life is 19569 orders of magnitude too rare to exist.

In the thought experiment of a life-searching universe, the mechanism for recognizing successful searches was given for free without a mechanism. Here, evolution “taking over” is finally just taking on that necessary task. This means that the context has not changed for the better from the original thought experiment.

In fact it is much worse because the for-free recognition mechanism was up to full Planck speed, but an evolutionary mechanism is much slower. In practice, evolution will have a recognition interval on the order of seconds to years instead of 1e43 times a second. Also, the imaginary for-free search engine was perfect in recognizing fitness for life, but natural selection is deeply imperfect.

A self-replicating molecule as a solution to the problem of abiogenesis is a profoundly naive self-deception.

Note that if the CBT was only something ordinary, then more than one variation or kind of toolkit should have evolved. That fact that there is only one CBT is evidence that the CBT is spectacularly special.

Essentially, that which I am calling a Common Biological Toolkit (CBT) is the functional basis for all living things. Therefore, on evolutionary theory it must have existed since the last universal common ancestor (LUCA). It includes DNA expressed through RNA into proteins, metabolization of fats and sugars, ATP as an energy intermediate, a cell enclosed by a lipid membrane, ion transporters across the cell wall, cellular division, and etc.

Wikipedia: LUCA

The thoroughgoing materialist will say that life as we know it could have been different if time was run again from the beginning. They don’t have any attachment to life as it exists now. Instead they speculate materialistically that with a new run though time, life could have taken a different path and a different kind of life would have emerged. Therefore they let arguments based on the practical problems of abiogenesis or of evolution just roll off their back. This is because they see observable life as just an example of the life-developing process. They don’t take the view that that the process is defined by life as it does exist in our history.

Even if every mechanism they have proposed for explaining life development ends up being invalid because of an obstacle, that also is not perceived as a problem. These were just potential mechanisms that could be taken up or discarded, and discarding a mechanism is not perceived as evidence against abiogenesis or evolution. (See Evolution as Fact)

However, there is zero evidence for any kind of life other than what we know. A concept for a type of life other than what is observed is only speculation, and there is no scientific evidence for it. Therefore, such speculation is not scientifically based and it is metaphysical.

In fact due to the massive evidence for fine-tuning of the universe, in my thinking it is more likely that there simply are no other viable chemistries for life other than what we already know. Life depends on many remarkable chemical characteristics of carbon, oxygen, and water and others. These were established at the very initialization of our universe. Therefore, there can be no naturalistic explanation for how these physical characteristics came to be so uniquely ideal for life.

If there is no other life other than what is known, and there can be no other life other than what is known, and life is unexplainable based on what is known of naturalistic processes, then most likely life was specially made here on earth. As the character Sherlock Holmes said, “How often have I said to you that when you have eliminated the impossible, whatever remains, however improbable, must be the truth?”

A better approach is to allow the option of non-naturalistic explanations. Life clearly looks designed. The universe clearly looks optimized for life as we know it. Both of these would suggest a super-intellect that optimized and designed the universe and life.

• Note that the person allowing non-naturalistic explanations still has access to all the naturalistic explanations. They just have a bigger and a better explanatory toolbox than those that claim to only use naturalistic explanations.
• Note also that it is impossible to avoid metaphysical things in science. See Darwinian Metaphysics. Materialistic science depends on a bedrock of immaterial metaphysical assumptions. It is profound ignorance to misunderstand this.

We say that the shape of the crystals is governed by law. Therefore, they have this regularity as a matter of necessity.

The random output of this example is purely by chance. Sequences like this are sometimes called Shannon information. The content has a lot of detail, but it is not complex. They have high "information" content, but very little meaningful information content.

So, any one recorded sequence like this would be very rare. (That is, the sequence is very unlikely to repeat.)

Also, these sequences effectively are collections of noise. Sometimes it is important to get a good example of noise. If noise is what I want, then any one sequence is as good as any other. So, it isn't very valuable. It is rare to get a specific sequence, but every day you get a sequence.

Every day, therefore, we expect a unique sequence because it would be very unlikely to repeat the exact same sequence on successive days. This means we never can expect to get a sequence that fulfills a specification (other than for noise).

Like the previous example, this example is not constrained by necessity. However, it is different this time because it has a complexity that the previous example did not have. The detail is not just random noise; it is not just Shannon information.

This sequence was not just purely chance; it does include considered choices. This is what makes it more meaningful. However, the value of this sequence is limited; it is not a specification for how to live life that could bring someone to a known state.

Salt cubes have their shape as a matter of necessity (due to functions of law). However for any random process there is no necessity (no law) that requires it to be a sequence that reliably makes things work. There just are many more ways to make a process fail than there are to make it work. This kind of sequence that makes things work is a functionally specified sequence.

It is easy to get some types of unique sequences (like the car colors). However, this fact is simply not an argument that therefore it is also easy to randomly get functionally specified sequences.

This last example of unlikeliness is a different category from the others, valuable and very rare.

This type of unlikeliness is associated to functionally specified information.

• Because it not constrained by necessity, it has contingency.
• It is complex, therefore minimally, it is not a product of chance.
• It fulfills specifications, therefore maximally, it is not a product of chance.

This three step method for eliminating chance and necessity was developed by Dr. William Dembski, and is called the Explanatory Filter. By inference, if something successfully passes through the three stages of the filter (is not by necessity, is complex, it fulfills specifications), then reliably the best explanation for it is that it is a product of design. (See Dembski, The Design Revolution, 2004.) Pure information content of this type is consistently a product of a mind. And objects based on this type of information are consistently a product of a mind. Therefore, wherever such things exist, we have reliable certainty that it was produced by an intelligence.

Concluding that a collection of functionally specified information came from a mind is not an argument from ignorance. It is not a conclusion from a gap in our knowledge. It is making a logical inference from what we know very well by long experience. It is an inference from experimentally reproducible data.

Some of the sciences (such as archeology and forensics) are based on this reasoning because it enables us to distinguish between what would happen naturally, and what happens because of the action of some agent.

The scientifically respected SETI project (Search for Extra-Terrestrial Intelligence) is based on this being a good inference. It assumes that specified complex information is an indicator of intelligence and therefore reception of this type of information is evidence of it having come from an intelligent mind.