This site is not maintained. Click here for the new website of Richard Dawkins.


← Huge hidden biomass lives deep beneath the oceans

Teratornis's Avatar Jump to comment 24 by Teratornis

Comment #184398 by Sigmund:

This puts limits for the length of time a single DNA strand can remain intact - probably somewhere in the order of tens of thousands of years.
That is one of the reasons why the Jurassic Park strategy of reviving dinosaurs doesn't now make scientific sense.

There are several techniques for reconstructing an original signal from a corrupted signal. NASA, for example, tends to get noisy signals from its space probes. The ability to clean up such photos might be as impressive as what's in the photos.

Similar problems arise in reconstructing lost ancient texts from surviving copies and excerpts, which usually contain errors. A harder problem is to reconstruct an extinct ancestral language from its surviving linguistic descendents. In the case of extinct only-spoken languages, we don't have any writing fragments to constrain the possibilities, making the reconstruction harder. Plus, the field of computational linguistics isn't very far along yet, especially for spoken languages, so it's hard to compare the descendent languages efficiently.

As long as the surviving data has not degraded into complete randomness, the original data might still be recoverable. That's a general rule in signal processing. But you have to be able to represent the data in some format accessible to computers. Since DNA is essentially an information-storage medium, and we know the code, it might be possible to apply various signal processing algorithms to infer original DNA strands from corrupted samples, given enough samples.

We'd have at least three sources of data from which to reconstruct dinosaur DNA:

1. Fossilized remains of degraded DNA.

2. Fossilized remains of dinosaur morphology.

3. Living DNA of animals more or less closely related to dinosaurs.

Once scientists finish sequencing the DNA of all species relating to dinosaurs, that would constrain the possibilities for dinosaur DNA. By how much, I don't know. There are probably only finitely many possible DNA arrangements for dinosaurs which could reproduce all the existing data in the above three categories. Even if the possible DNA arrangements remained combinatorially vast, at least it would be possible to create animals that are indistinguishable from dinosaurs as we are able to know them. If their actual DNA is wrong, then we would have created convergent dinosaurs.

With enough computing power, it might be possible to simulate morphogenesis from starting DNA. Thus a computer might search for all possible chromosome arrangements which lead to virtual dinosaurs that duplicate everything we know about real dinosaurs. With enough data about real dinosaurs, it might be possible to constrain the virtual chromosomes down to one set for a given species - and that set would have a high probability of matching the extinct real set.

Sun, 25 May 2008 13:45:00 UTC | #175131