Moore's Law and the Origin of Life 272
DoctorBit writes "MIT Technology Review is running a story about an arXiv paper in which geneticists Alexei A. Sharov and Richard Gordon propose that life as we know it originated 9.7 billion years ago. The researchers estimated the genetic complexity of phyla in the paleontological record by counting the number of non-redundant functional nucleotides in typical genomes of modern day descendants of each phylum. When plotting genetic complexity against time, the researchers found that genetic complexity increases exponentially, just as with Moore's law, but with a doubling rate of about once every 376 million years. Extrapolating backwards, the researchers estimate that life began about 4 billion years after the universe formed and evolved the first bacteria just before the Earth was formed. One might image that the supernova debris that formed the early solar system could have included bacteria-bearing chunks of rock from doomed planets circling supernova progenitor stars. If true, this retro-prediction has some interesting consequences in partly resolving the Fermi Paradox. Another interesting consequence for those attempting to recreate life's origins in a lab: bacteria may have evolved under conditions very different from those on earth."
Non-peer reviewed... (Score:5, Insightful)
This is a fine example of how not to use arXiv as a news source. This old yarn has been trotted out before, and it is based on bad assumptions about complexity and offers a handy False Dilemma Fallacy.
Either
1+1=6 or
1+1=8.
1+1=6 is disproved, so 1+1 =8!
Or your math is wrong.
Complexity != genome size.
See c-value enigma.
Re:Or... maybe your assumption is wrong (Score:4, Insightful)
You said it better than I was going to say it.
The way I see it, they:
a.) Plotted some data
b.) Extrapolated a simple trend from that data
c.) Forecasted, using the trend function, before the point of data collection
d.) Came up with some wild conclusions from that forecast (or "beforecast"?) that rely heavily on the validity of the simple trend.
It kind of smells like bad science...or at least risky science.
Re:Looks like creationism... (Score:5, Insightful)
It does, however, use a metric pretty much meaningless to biology and comes with an answer that will get it some attention from the tragically retarded known as scientific journalism (and by extension, Slashdot editors).
Cambrian Explosion (Score:5, Insightful)
The assumptions in the article are especially suspect, given the large number of quite well documented "explosions" of genetic diversity in Earth's history (see, e.g., the Cambrian Explosion [wikipedia.org] for the biggest example, though there are plenty of lesser events), where gigantic leaps in genetic diversity appeared over (geologically) short timescales. An extrapolation assuming a generally smooth growth rate is simply untenable.
Re:Looks like creationism... (Score:3, Insightful)
The problem with the kind of creationism some people are advertising is that they insist that it happened around 6000 years ago. A lot of scientists would be ok with the idea of creationism -- if you allow it to happen billions of years ago as the spark that created life, but then let life evolve independently. But of course then humanity is not special -- unless the creator helped things happen this way for the purpose to create intelligent life.
So creationism/intelligent design is OK, and a higher being managing/guiding the universe is OK; it just doesn't make sense for it to have happened 6000 years ago.
No the problem with creationism is that it's a crappy scientific theory. It doesn't add any predictive power, doesn't resolve the actual question of how life was created, and it fails Occam's Razor. It's exactly as useful as "a wizard did it".
Extrapolation! (Score:5, Insightful)
what could possibly go wrong, particularly when you extrapolate twice as far as you actually have data for.
Re: Looks like creationism... (Score:4, Insightful)
Re:Or... maybe your assumption is wrong (Score:2, Insightful)
Here's the problem: reasonable functions exist which behave completely differently from what goes on inside the data set.
Let's take the article before us as an example. If x(t) is the average number of functional base pairs in an organism as a function of time, the article's authors are asserting that x'(t) = kx(t), or rather that the rate at which more base pairs are generated is proportional to the existing amount of base pairs.
But what if we change this just a tiny bit? What if x'(t) = kx(t) + C, where C is tiny (think 10e-5 base pairs / year)? The graph they show would look exactly the same in the range of known organisms, but now it would only take between 4 and 5 billion years for mammalian-level complexity to emerge, and at that point we're in margin of error. (Don't believe me? Here's a back-of-the-envelope fit to the data that I made with this model: http://www.wolframalpha.com/input/?i=plot+ln%28%284349%29*exp%28%282.3e-9%29*t%29+-+4348%29%2Fln%2810%29+from+0+to+10%5E10).
Note how the log-plot of this function (again, see the link above) has a "hockey-stick" shape located outside of the region where we have data. A more precise fit will provide no more or less error for the data on hand compared to an exponential curve, but it behaves radically differently outside of that range.
Re:Looks like creationism... (Score:4, Insightful)
I see no reason to privilege math over English in this regard.
But you certainly see much reason to privilege reason, i.e., logic and all it implies.
There's no running around the fact that if you refuse the framework you're left with no knowledge at all. Either you accept some kind of basic realism or you give up and go with the methodological anarchism of a Feyerabend, who sees no difference at all between modern Physics and Astrology, or some kind of skepticism, be it classical skepticism, which affirms no possibility of knowledge of anything at all, or the Kantian alternative, which says science can be at best a very precise knowledge of our sensory input, but incapable of saying anything at all about this maybe existing thing that maybe multiple humans (supposing there are more than one) perceive as "the external world".
I tend to switch between realism and kantism, but I concede the later is more rigorous. Too bad it causes everything we say about anything to necessarily become surrounded by double quotes.