NOTE: This is a post about probability estimation, rather than about inferring design. All systems – whether designed or not – have a certain level of specified complexity associated with them. Only if that level exceeds a certain threshold can we reliably infer intelligent design. The definition of a pattern’s specified complexity makes reference to P(T|H), the probability of a pattern T with respect to the chance hypothesis H. In this case, the pattern we see is an observed structure in a meteorite, and there are two competing hypotheses as to how it arose (leaving aside the possibility of contamination). What I’m interested in is how we would calculate the probability of that pattern if it arose abiotically, as opposed to the probability of that pattern if it is a bacterial fossil. It’s this kind of number-crunching which I feel we need to become proficient at. It would definitely be a feather in our caps if the ID movement could develop a readily utilizable metric to assist NASA in evaluating claimed discoveries of life from outer space. – VJT.
Recently, NASA scientist Richard Hoover looked at some slices of three very rare meteorites using an electron microscope technique called Field Emission Scanning Electron Microscopy, and saw what he believes to be tiny fossils of Cyanobacteria. Hoover’s article, Fossils of Cyanobacteria in CI1 Carbonaceous Meteorites has generated a storm of controversy. Physicist Rob Sheldon has recently blogged about Hoover’s findings here and responded to some common criticisms of Hoover’s work here. Alan Boyle’s report on MSNBC is available online here. Science blogger Dan Satterfield has a post about Hoover’s discoveries here, and a review by “Discover” magazine correspondent Phil Plait can be found here. A critical review by microbiologist Rosie Redfield can be found here, while P.Z. Myers’ dismissal of Hoover’s claims is available online here.
I thought this would be an interesting test case for the concept of complex specified information (CSI), which has been getting quite a bit of attention on this blog recently (see for instance Mathgrrl’s post here, and my posts here and here). So without further ado, let’s proceed.
What is Hoover claiming?
Hoover describes his findings as follows:
A number of biominerals and organic chemicals (that are interpreted as biomarkers when found in Earth rocks) have been detected in CI1 carbonaceous meteorites. These include weak biomarkers such as carbonate globules, magnetites, PAH’s, racemic amino acids, sugar alcohols, and short chain alkanes, alkenes and aliphatic and aromatic hydrocarbons that are produced in nature by biological processes but can also be fomed by catalyzed chemical reactions such as Miller-Urey and Fisher-Tropsch synthesis. However, the CI1 meteorites also contain a host of strong biomarkers for which there are no known abiotic production mechanisms. These include magnetites in unusual configurations (framboids and linear chains of magnetosomes), protein amino acids with significant enantiomeric excess, nucleobases (purines and pyrimidines), and diagenetic breakdown products of photosynthetic pigments such as chlorophyll (pristine, phytane, and porphyrins), complex kerogen-like insoluble organic matter and morphological biomarkers with size, size range and recognizable features diagnostic of known orders of Cyanobacteriaceae and other prokaryotic microfossils. (Emphasis mine – VJT.)
At the end of his article, Hoover summarizes his argument for identifying the structures he observed in carbonaceous meteorites as fossilized bacteria from outer space:
It is concluded that the complex filaments found embedded in the CI1 carbonaceous meteorites represent the remains of indigenous microfossils of cyanobacteria and other prokaryotes associated with modern and fossil prokaryotic mats. Many of the Ivuna and Orgueil filaments are isodiametric and others tapered, polarized and exhibit clearly differentiated apical and basal cells. These filaments were found in freshly fractured stones and are observed to be attached to the meteorite rock matrix in the manner of terrestrial assemblages of aquatic benthic, epipelic, and epilithic cyanobacterial communities comprised of species that grow on or in mud or clay sediments. Filamentous cyanobacteria similar in size and detailed morphology with basal heterocysts are well known in benthic cyanobacterial mats, where they attach the filament to the sediment at the interface between the liquid water and the substratum. The size, size range and complex morphological features and characteristics exhibited by these filaments render them recognizable as representatives of the filamentous Cyanobacteriaceae and associated trichomic prokaryotes commonly encountered in cyanobacterial mats. Therefore, the well-preserved mineralized trichomic filaments with carbonaceous sheaths found embedded in freshly fractured interior surfaces of the Alais, Ivuna, and Orgueil CI1 carbonaceous meteorites are interpreted as the fossilized remains of prokaryotic microorganisms that grew in liquid regimes on the parent body of the meteorites before they entered the Earth’s atmosphere. (Emphasis mine – VJT.)
Relevance for CSI
Uncommon Descent readers will recall that Professor William Dembski defines the specified complexity Chi of a pattern T given chance hypothesis H, minus the tilde and context sensitivity, as:
where Phi_s(T) is the number of patterns for which S’s semiotic description of them is at least as simple as S’s semiotic description of T
and P(T|H) is the probability of a pattern T with respect to the chance hypothesis H.
However, if Hoover were right about absence of known abiotic production mechanisms for the strong biomarkers he observed, that would create major problems for the calculation of P(T|H), and hence CSI of the structures he observed.
Are the structures really bacteria?
But is Hoover right in claiming that abiotic processes cannot account for the structures he observed in meteorites? Microbiologist Rosie Redfield begs to disagree:
He spends a lot of text discussing the morpohlogical similarities of these filaments to cyanobacteria, but I don’t regard these similarities as worth anything. Filamentous bacteria are very morphologically diverse, and additional variations in appearance are likely to result from inconsistent preparation for electron microscopy. It’s probably pretty easy to find a bacterial image that resembles any fibrous structure. In the absence of any statistical evidence to the contrary, it’s prudent to assume that such similarities are purely coincidental.
Rocco Mancinelli, senior research scientist at Bay Area Environmental Research Institute was also skeptical:
As a microbiologist who has looked at thousands of microbes through a microscope, and done some of my own electron microscopy, I see no convincing evidence that these particles are of biological origin.
“Disover” magazine’s Phil Plait was also underwhelmed, writing that Hoover “is basing a lot of this on the shape of the structures he sees… but looking like a microbe doesn’t make them a microbe!”
P.Z. Myers was even more dismissive in his scathing review of the findings:
The extraterrestrial ‘bacteria’ all look like random mineral squiggles and bumps on a field full of random squiggles and bumps, and apparently, the authors thought some particular squiggle looked sort of like some photo of a bug. This isn’t science, it’s pareidolia…
I’d be more impressed if they’d surveyed the population of weird little lumps in their rocks and found the kind of consistent morphology in a subset that you’d find in a population of bacteria. Instead, it’s a wild collection of one-offs.
If these critics are correct, then the factor P(T|H) which appears in Professor Dembski’s equation for calculating complex specified information (CSI) is actually quite high.
But are they right? Physicist Rob Sheldon contends that some experts in bacteriology concur with Hoover’s claims that the structures he observed are specific to certain species of bacteria:
But even if, like Rosie, you claim that lots of abiotic stuff looks biological, then you had better explain why numerous European academicians who are experts in algae and bacteria have examined Hoover’s photographs and agree with him that these are not just identifiably biological, but identifiable by genus and species.
I look forward to hearing more from Rob Sheldon about the verdicts of these European experts, and if Rob could forward any articles by these experts to me, I’d be immensely grateful.
What about contamination?
The other major concern, of course, is contamination. The risks of it occurring in a study like this are very real, as microbiologist Rosie Redfield points out:
An important concern with this kind of study is contamination with terrestrial organisms before examination. He doesn’t say how the meteorites have been stored before he obtained them, nor how the surfaces of the meteorites were treated before being fractured and examined. He doesn’t say how they were fractured – might they have been cut with a scalpel blade or just pressed on until they crumbled? He says that the tools were flame-sterilized, but not what the tools were or how they were used.
She concludes: “As evidence for life this is pathetic.”
“Disover” magazine’s Phil Plait concurs, writing that “the major problem here is contamination.”
Rocco Mancinelli, senior research scientist at Bay Area Environmental Research Institute, was also very concerned about the possibility of contamination:
The techniques used may not have been appropriate for these types of analyses. It is stated that the implements were flame-sterilized, with no details of how this was performed. Were the implements placed in the flame of a Bunsen burner? If so, sometimes soot can get on them at the microscopic level. The usual procedure for flame sterilization is to dip the implements in ethanol then burn the ethanol off. Yet, these would be inappropriate for this type of analysis. You need to have everything clean and then bake at 550 degrees C overnight. These missing details would cause me to question not just about the photos, but the elemental analyses as well. I am also disturbed about the lack of nitrogen. There should be more. There are many technical flaws in this paper.
However, physicist Rob Sheldon counters that contamination is an enormously unparsimonious hypothesis in this case:
Wouldn’t recent contamination be a much more conservative explanation of these biofossils?
Well it would if we could explain (a) how to make microfossils in the first place; (b) how to make them in, oh, the 12 hours it took to collect the meteorites and put them in storage; (c) how to eliminate 12 of the 20 essential amino acids from the meteorite; (d) how to reduce the nitrogen content of the fossils to below 0.5% when 15,000 year-old mammoth hair shows no loss of nitrogen; (e) how to make fossils from organisms last seen on Earth 400 million years ago; (f) how to make the fossils out of isotopically meteoritic material not found on Earth; (g) how to make fossils out of super soluble salts and then combine them with the meteoritic material in such a way as to make them appear intrinsic; (h) how to make fossils with 10nm “fibrils” exquisitely preserved; (i) how to make recent contamination fossils inside a well-formed “fusion crust” of sterile melted meteoritic material; etc. Well, you should just read the paper rather than my summary. My point is that we don’t know how to do any of these things, so that a “contamination hypothesis” actually raises more questions than it answers. The simplest answer is that these really are indigenous microfossils from ancient extraterrestrial microorganisms.
As I see it, there are two rival hypotheses we should consider here: terrestrial contamination, and the possibility that the structures observed are not bacteria anyway. I think we should investigate the latter hypothesis first. We need to know if the structures really are bacteria before we worry about where they came from.
I’d now like to invite comments from readers with specific suggestions as to how P(T|H) might be calculated for the chance hypothesis that the structures observed by Hoover are non-biological. Here are my own ideas. As a first step, we need to count the number of points of similarity between the structures Hoover photographed and the species of bacteria which most closely resemble them. Second, we need to rank each of these points of similarity in decreasing order of likelihood, with respect to the chance hypothesis that their similarity to bacteria is entirely coincidental. Third, we need to quantify the probabilities that each of these features arose by chance, and finally we need to calculate an overall figure which represents the likelihood that the structures observed, taken as an ensemble, arose by chance in each meteorite.
P.S. If readers have any further comments made on the measurement of CSI, CSI-lite or kairosfocus’ X-metric, they are welcome to make them here as well.