Doesn't Natural Selection lead to Specified Complexity?

Specified Complexity is billed as an explanatory filter, and it is suggested that such 'specified' complex function cannot come about by chance.Since Natural Selection can be viewed as a filter over the random mutations that we KNOW accur in biological systems, couldn't Natural Selection be the source of this supposed 'specificity'?Natural Selection, surely, means that what we can observe in the natural world has not, in fact, come about by chance, but has been 'directed' by the environment -- that is only changes that 'work' are carried forward.

Thread copied here from the Doesn't Natural Selection lead to Specified Complexity? thread in the Proposed New Topics forum.

In the broad sense that is pretty much correct, assuming that beneficial mutations that build complexity occur. Natural selection does act as a filter, which can be seen as specification.Using Dembski's special (and misleading) definition, that is not true, since all sources other than design must be ruled out before a thing may be called specified complexity.As a result I see quite a lot of equivocation over the definitions - ID supporters need to use the broad sense whenever they want to claim that specified complexity is found in nature, and Dembski's sense whenever they want to claim that it supports ID.

According to Dembski:The signature of an intelligence in design is an event that is "contingent and therefore not necessary; if it is complex and therefore not easily repeatable by chance; and if it is specified in the sense of exhibiting an independently given pattern" (Dembski in Natural History Magazine).He uses the concept of 'not easily repeatable by chance' ... which seems to be a subjective measure, and so of little use.But, assuming that 'chance' (in some form) is an issue, then the natural process of selection due to environmental pressure means that we are not dealing with 'chance' any more.Although highly complex, Natural Selection must be somewhat deterministic (i.e. given the exact same inputs, the process will provide the exact same outputs).The random nature of mutation tends to make the inputs change, but those inputs are not purely random, just highly chaotic.

Dembski does TRY to be rigorous with his "complexity" measure (which is really an improbability measure) and I think that he does a bit better than you give him credit for (doubtless due to lack of explanation in the article). However, it does still have subjective elements.The real problem is that actually following Dembski's method is impractical in many cases - including all the biological features Dembski would like to use his method on, to support his creationist beliefs.I also don''t like the idea that natural selection should be seen as entirely deterministic. It is the outcome of a statistical process, and while some aspects of evolution may be nearly as inevitable as a casino making a profit many details of the outcome are not.

By deterministic I meant that given the same set of 'genes', the same 'environment', and the same 'mutations' the outcome would be the same.However (as I said) the 'randomness' of the inputs (i.e. the mutations part) would make going back to the beggining and starting over lead to different results (by and large).The variables are vast and have complex interactions so deliberately setting up for a particular outcome would be near impossible ... but since the process is, well, a process, if one could recreate the exact conditions one could repeat the process.

quote:

---

By deterministic I meant that given the same set of 'genes', the same 'environment', and the same 'mutations' the outcome would be the same.

---

Unless you define "the environment" as a complete and fully detailed description of every event that occurs in the relevant time period, you would be wrong. A beneficial mutation will give an overall advantage, but it is perfectly possible to encounter situations where the advantage does not come into play or is inadequate or even where the mutation is a disadvantage. A beneficial mutation is quite vulnerable in the early stages when it is present in only a few individuals and could easily be lost.Equally drift can cause even weakly deleterious mutations to spread, overcoming natural selection. As you would expect with a statistical outcome the effect is strongest when the population is small, and chance can cause proportionally greater variations in the outcome.

That IS how I define 'the environment'.Changes within the environment (even restricting it to the Earth's bio-sphere) are not random, they are the result of interactions of such a vast array of variables that we are unable to accurately predict the changes -- but that's out limitation not a fundamental feature of 'the environment'.The 'process' of natural selection operates at a fairly simple level, but has vast numbers of variables with unclear interactions.It would appear that the necessary precision of the starting point may be lower than 'absolutely the same' or else the evolution-loss-re-evolution of features (at the level of the individual) would not be seen, nor would we see squid-eyes and mammal-eyes sharing so much common structure.the 'process' is detreministic -- there is so much latitude in the variables that it has the appearance of 'randmomness'

quote:

---

That IS how I define 'the environment'.

​

Changes within the environment (even restricting it to the Earth's bio-sphere) are not random, they are the result of interactions of such a vast array of variables that we are unable to accurately predict the changes -- but that's out limitation not a fundamental feature of 'the environment'.

---

The problem is that by doing so you essentially make natural selection unimportant. All you have is a deterministic system working out it's course and you make no useful distinction between selective events and events which just happen to preserve or eliminate a particular allele.

quote:

---

The 'process' of natural selection operates at a fairly simple level, but has vast numbers of variables with unclear interactions.

​

It would appear that the necessary precision of the starting point may be lower than 'absolutely the same' or else the evolution-loss-re-evolution of features (at the level of the individual) would not be seen, nor would we see squid-eyes and mammal-eyes sharing so much common structure.

---

And this is why other approaches are preferred. We should not count an event as natural selection unless genetic variations actually in the population affect the outcome, and unless the event affects enough individuals to have a significant affect on fitness. ANd since we can't predict everything exactly why not use statistical modeling - which works well - rather than throwing up our hands and giving up ?

Treating the environment as non-random (which it isn't, it's just too intricate for us to model accurately) does not make natural selection unimportant.The environment exists in time, and changes over time. For biological entities to persist they have to have a mechanism for keeping up with the changes.All I am saying is that that process is NOT random. In fact, it is deterministic (in the computing sense) in that setting the same initial conditions and letting the process run will result in the same outcome.Change the initial conditions, or any of the steps along the way and the outcome will be different.Using statistical models to investigate systems which are too intricate for us to model accurately is perhaps the best we can do at present -- so long as we don't confuse our 'best approach' with a 'perfect match'.

Hang on .... so natural selection can be viewed as providing the 'specified' bit ... but there is some question over the 'complex' bit?Even that means that SC cannot be used as a marker of intelligent design.

quote:

---

Treating the environment as non-random (which it isn't, it's just too intricate for us to model accurately) does not make natural selection unimportant.

---

Of course, you are missing the point I am making, which is that your view relies on ignoring the distinction between selection and drift. The events that correspond to drift may be deterministic in an absolute sense, but they still aren't natural selection. (Or they may not be - to the best of my knowledge determinism is still unproven).

quote:

---

All I am saying is that that process is NOT random. In fact, it is deterministic (in the computing sense) in that setting the same initial conditions and letting the process run will result in the same outcome.

---

And since you have to assume complete determinism to get to that conclusion, it is utterly trivial and tells us nothing about natural selection. In a deterministic universe, everything is deterministic - we need not know anything about natural selection to reach your conclusion. Meanwhile, models where natural selection is the outcome of a statistical process are both more practical and more informative.

quote:

---

Hang on .... so natural selection can be viewed as providing the 'specified' bit ... but there is some question over the 'complex' bit?

---

IF you are using "complexity" in the ordinary sense of the word, then natural selection does not directly contribute to complexity. It may (and probably does) work to allow complexity to build up, but the complexity has to come from the processes that produce variation, not the process which whittles it down.If you use Dembski's sense then natural selection has to be included in the probability calculations, so it can't affect the "complexity" at all - in fact it makes things that seem more "complex" than they really are. Just one way in which Dembski's "complexity" differs from the usual sense.

By 'drift' do you mean 'neutral genetic change'?In order to determine the dererminism (or not) of natural selection one has to 'fix' the inputs ... so drift becomes irrelevant.If we narrow the environment to a petry dish, and the gene set to a single (at first) bacterium I'm pretty sure we could 'rig' the environment to get a particular genetic change to fix in out petrie dish population.e.g. one could doubtless manufacture an anti-biotic resistant strain from a non-anti-biotic resistant strain by setting the conditions up (to broadly match what happened in hospitals).The statistical anaysis is necessary because mutations are random, and the environment is difficult/impossible to accurately model -- not because natural selection is in any way random.

I was using a common-or-garden understanding of complexity.Dembski:
The design inference uncovers intelligent causes by isolating the key trademark of intelligent causes: specified events of small probability. Just about anything that happens is highly improbable, but when a highly improbable event is also specified (i.e., conforms to an independently given pattern) undirected natural causes lose their explanatory power.Doesn't that imply that specified complexity just means 'intended outcome' ... and backs round to not helping at all in detecting 'intelligence'?If we observe a pattern after-the-fact in which natural selection 'formed' the pattern or an 'intelligence' formed the pattern what difference would there be from Dembski's view point -- nothing I would suggest.