Saturday, July 16, 2011

To all creationists who think they know math.

You don't.
The next time somebody tries to claim that life is impossible by natural means and then quotes some absurd made-up probability, the indisputable fact is that you don't have a clue what you're talking about.

I recently saw a little letter-to-the-editor in a local free paper that used the crazy numbers mode of argument about the question of teaching creationism in schools.  Here's the letter.
Evolution is a doctrine built entirely on crazy speculation, and the only reason they have it in school is because scientists want to say that God is not science.  Yet, in order to teach kids to hate Jesus as much as the scientists do, they have to pose an impossibility.  The probability of life forming on its own by natural means is an astronomically small probability of 1 in 10 to the 700-billionth power.  That is equivalent to winning the lottery jackpot 100 billion times in a row consecutively.  That's just the straight math, plain and simple.  You can't argue with numbers.  And we teach this impossibility to kids as science?  Instead of the one truth in the lord, our God?  It's high time we stop this foolishness and show our kids the real truths of life and avoid driving our nation further into hellfire.
That was utterly classic.
I especially love that attitude of how you can't argue with numbers.  It's a common one, and one that lies in this sense we have that numbers are absolute and immutable and there is nothing we can say to deny their value.  When we see raw statistics, everything looks so solid that it seems like an undeniable fact.  Well, it doesn't work that way.  Part of good academic process is not just the numbers you get, but how you get them.  The method used by creationists, which largely involves forcible extraction from the place where the sun never shines is not admissible by any standards of rigor.

What follows after the jump is my response to the author of the previous letter.

I wrote both to the paper and to the original author, who, through his pseudonym, revealed himself to be someone who had approached me with an entirely different argument over an email thread some years before.  Before sending the full message, I tried to confirm, of course, whether it was actually the same person I thought it to be, and he too, affirmed my suspicions.  The fellow is actually otherwise well-adjusted, and the last time I had my discussions with him, he seemed to accept that it was something of a lost cause, and this time around, he was curious to see if I, as a rationalist, had any sort of argument against his "pure numbers" points.

His morbid curiosity proved his undoing.
I figured it was you.  I can't imagine a whole lot of people using that name... and... Yes, you're right to presume I have a bunch of holes to poke in the numbers argument.  Using retrospective improbability is a fallacious argument to begin with because it relies entirely on ignoring variables which affect causal probabilities.  Instead, it is taking point samples of beginning and ending states and the causal history is not considered.  The failure of this is that all complex states have very tiny zero probabilities regardless of what you're looking at.  Nonetheless, the individual subcomponents of that state, or the timeline thereof and the dependent probabilities and the hysteresis of effect of any substate may not be unlikely at all.  When you don't take any of that into account, the number you get will always be wrong.  And though I could provide it, I don't need a sample space to tell you that the probability of that wrongness is 100%.

There are a number of problems I have with most all the statistical calculations that people put out in favor of divine providence. There's just no rigor whatsoever in their approach, and whenever you see very obscene odds like the one you mentioned, no matter who it is saying it, it's all but certain that the number is completely made up.  While I'm certain you didn't actually do the figure yourself (the analogy you used is provably false), did your source show how the figure worked out, or did he just say that's the figure he got from some other guy?  Let's see the workings, or my first presumption is that your source is lying.

For instance, whenever I hear "Bible Code" people say the chances of finding the messages they found on an arbitrary tome of any other sort is some absurdly small number, I know they are lying.  I guarantee you that no statistician produced that computation for them, and they just pulled the number out of their asses, and this is only furthered by the fact that not one of them can even agree on the order of magnitude of the probabilities.  Even those who try to sound more reasonable by putting forth more graspable figures like 1 in 50,000 are still lying about ever having done anything resembling a statistical computation.  One of the Bible Code fanatics I challenged to show the workings for how he produced his figures laid his hand out and showed his calculation to be fundamentally flawed.  That computation was based on a non-fixed matrix size of random characters with random code selection and a fixed skip rate.  Unfortunately, none of those apply to so-called Bible Codes.  When you take into account that the original source material actually contained coherent words in the language they're searching in, the general patterns of usage common throughout that source, then account for the fact that the rules of skip code searches are kind of arbitrary in that you're allowed to have different skip rates for different words in the same matrix (some aren't required to skip at all), you're allowed to have it be readable in any direction or angle, the odds actually pull up into very ordinary-sounding figures.   On average, the likelihood of any given "Bible Code" hit is right around 1 in 5.  In fact, if you are working in the Hebrew text of the Old Testament, there are specific books like Leviticus where, just because of the grouping patterns, you can actually hit upon sections where finding a code is more likely than not finding one.  It's actually much harder in English because the alphabet of English works in sound "particles" rather than syllables, but in Hebrew, Arabic, Sanskrit, what have you, finding such codes in any text is much much easier.

With bizarre numbers for "life forming at random," all those made-up figures are based entirely on a lack of detail in their study.  They are based entirely on comparing a null starting point and the current state of life and think in terms of all the atoms coming together just as they are in a SINGLE STEP.  The absurdity of this sort of setup is beyond words, since no scientist ever claimed that life's formation was a one-step process to get where we are right now... that's in every way the exact polar opposite of what we say.  But that's what the creationists go on because that allows them to create absurd figures which because they're numerical sound like nuggets of fact.  It's purely misleading and it depends expressly on lying about what scientists actually say.  If anything, it's closer to what creationists believe because they say that their god simply made things happen in a single swipe through magic...  but science argues that conditions made certain things possible and when those things happened, other conditions could eventually be met to make other things possible, and so on down a long causal chain.

Secondly, there's a tremendous failure on the part of creationists to take into account any natural processes and naturally available localized conditions that influence the odds.  For them, happening by chance can only mean the right number of atoms just happening to be there all at once and they assemble in exactly the right way and there's no such thing as physics or chemistry at play here.  Dembski is one who is very skilled at hiding this by obfuscating his ignorance with a lot of complex language.  Nonetheless, if you read his work, he at most acknowledges the existence of relevant stochastic processes and side information to at least give the impression that he is taking into account context (which most creationists don't even go far enough to pretend to acknowledge)... but he fails miserably at identifying what those processes and context influences are, and he fails to provide any sort of heuristic for determining the importance of any piece of side information and completely omits the question of the completeness of the context or even the order of dependencies in the probability computation.  His essay where he works off the No Free Lunch principle is one where he ignores the fact that any degree of knowledge of causal history affects the likelihood that a subject would identify something as a chance occurrence.

Take, for instance, a simple theorized precursor to life -- a fatty acid micelle structure forming a stable semipermeable membrane which allows it to store and "consume" nucleotide monomers.  What are the chances of this forming at random?  Well, with no such thing as physics or chemistry, it's pretty darn ridiculous indeed.  With the acknowledgement that there is such a thing as chemistry and we live on a planet that happens to have a pretty ubiquitous solvent in liquid form, the chances are higher but still absurdly small if we just select a random spot somewhere on the Earth.  Now... let's say we happen to pose that question such that we are checking in the vicinity of an undersea thermal vent where heat and minerals and sulfurous compounds from magma beneath the crust is constantly spewing out causing the local temperature rise and pH drop and providing an abundance of chemical ingredients for reactions which might occur?  Well, as it so happens, the chances are nearly 100%... and all it really took to realize that was a lot of relevant information and certain stochastic processes that occur within those types of environments to stack the odds in your favor.  Does that mean life was a certainty?  No...  what it means that the "chance" occurrence is only astronomically small when you consider absolutely nothing.  This is always going to give you a wrong figure because life didn't start from a void; it started under certain conditions which themselves were possible because of where we are.  Now if you want to talk about that being improbable, that's partly true, but probability is only one part of the picture.  Low probability with a large population size can still give you a high expected value.  If you have some statistic  that applies to 1 in a billion people, that still means an expected value of ~7 people worldwide.  The same applies to a planet having conditions like ours.

Regardless, your letter proves most disappointing.  When you had responded my editorial piece about 2 years back, you commented as if you accepted that some of the classical arguments I tore down were anyway invalid, and you spoke as if interested in more solid arguments for or against the existence of your God.  Whether they be rooted in science or in the purported historicity of Jesus is a separate matter.  This letter, however, betrays a total unwillingness to exhibit any form of intellectual honesty.  Instead you'd spoken in absolutes in the same manner as a typical evangelical YEC who has already made up his mind.  Furthermore, you spoke as if not merely to be playing the "both sides" card, but to entirely eliminate the opposition and train students not to grow up and approach things scientifically, but Biblically.  True, you had previously said some inane things about intrinsic powers of scripture and all, but you were otherwise demonstrating some measure of sanity.  Compared to how you argued 2 years ago, the turn since then proves most disgraceful, at least if I'm to take the words of your letter-to-the-editor to wit.

Here's to hoping you were being facetious,
- Parashar K.

Well, he apparently wasn't being facetious, but did admit to writing a bit more angrily than he might normally do.  Still, the fact that he has decided a priori that anything other than the Bible is false is pretty much inexcusable.