At the root of this idea is overwhelming recent evidence for horizontal gene transfer - in which organisms acquire genetic material "horizontally" from other organisms around them, rather than vertically from their parents or ancestors. The donor organisms may not even be the same species. [...]

They found that the actual code is around one in a million in terms of how good it is at error mitigation. "The actual genetic code," says Goldenfeld, "stands out like a sore thumb as being the best possible." That would seem to demand some evolutionary explanation. Yet, until now, no one has found one. The reason, say Woese and Goldenfeld, is that everyone has been thinking in terms of the wrong kind of evolution. [...]

Starting with a random initial population of codes being used by different organisms - all using the same DNA bases but with different associations of codons and amino acids - they first explored how the code might evolve in ordinary Darwinian evolution. While the ability of the code to withstand errors improves with time, they found that the results were inconsistent with the pattern we actually see in two ways. First, the code never became shared among all organisms - a number of distinct codes remained in use no matter how long the team ran their simulations. Second, in none of their runs did any of the codes evolve to reach the optimal structure of the actual code. "With vertical, Darwinian evolution," says Goldenfeld, "we found that the code evolution gets stuck and does not find the true optimum."

The results were very different when they allowed horizontal gene transfer between different organisms. Now, with advantageous genetic innovations able to flow horizontally across the entire system the code readily discovered the overall optimal structure and came to be universal among all organisms. For the researchers the conclusion is inescapable: the genetic code must have arisen in an earlier evolutionary phase dominated by horizontal gene transfer.