The development of directed evolution methods for nucleic acids was an idea whose time had come, as evidenced by its multiple, independent origins. It can be argued that Oliphant and Struhl were some of the earliest practitioners, having randomized oligonucleotides to identify binding promoters and repressor binding sites in cells, circa 1986–1987 (Oliphant et al. 1989, 1986; Oliphant and Struhl 1987, Oliphant and Struhl 1988). Just as the GCN4 repressor was used to capture binding sites by affinity chromatography (Oliphant et al. 1989), Blackwell and Weintraub (1990) used electrophoretic mobility shift assays to select for differential binding by homo- and heterooligomers of the helix-loop-helix proteins MyoD and E2A. Robertson and Joyce (1990) were amongst the first to recognize that randomization and selection could be applied to functional nucleic acids, such as the Tetrahymena self-splicing ribozyme, although I have always argued that this honor might first go to Sol Spiegelman, whose ideas were far in advance of the technologies of his time. As early as 1970, the Spiegelman lab was able to remarkably select for Qβ ‘mini monster’ variants whose in vitro replication was resistant to ethidium bromide (Saffhill et al. 1970). Just as the Szostak lab of the early 1990s was a remarkable place to be (Doudna, Green, Bartel, Famulok), the Spiegelman lab of the late 1960s and early 1970s must have been extraordinary (Mills, Pace, Kramer, with Orgel as a collaborator).

That said, the discovery of the in vitro selection of aptamers/SELEX is often cited as a milestone in directed evolution. Tuerk and Gold (1990) selected a non-natural protein-binding sequence. The Szostak lab selected a non-natural dye-binding sequence (Ellington and Szostak 1990). The Gold lab used a library of 8 random nucleotides; the Szostak lab used a library of 100 random nucleotides. The Gold lab patented their findings; the Szostak lab did not. The rest, as they say, is history.

A potentially interesting question is: Why was the selection of dye-binding aptamers exciting, given the previous discovery of ethidium bromide not-binding Qβ replicators? A portion of the answer of course has to do with the extraordinary 20 years displacement of the first and second findings. We stand on the shoulders of giants, and then quickly forget their names. The advances in technology in that 20 years period are another rationale: in vitro selection/SELEX became possible because of advances in synthetic DNA technology (making library production possible) and molecular amplification technologies (PCR). The expanse in capability is like the wondrous transformation of highly automated and parallel sequencing into NextGen sequencing. Both are still sequencing, but the scale is now staggering.

But I think the most significant difference between the two findings was the hint that there was an undiscovered country in deep sequence space (to further mix Shakespearian and Star Trek metaphors). The Big Idea that biology is but a subset of a larger set of possibilities is still mind-blowing. Crick (1968) talked about the frozen accident of the genetic code, but the foment at origins that led to something capable of being frozen is still largely unimaginable. That Big Idea got a little bit lost along the way, with directed evolution being used for many biotechnology applications, but often only in passing to describe the possibilities that may have been present at origins. We are slightly closer to the self-replication of ribozymes, the Xeroxase, than we were following the application of a technology hammer to the origins nail, but still not there in a way that would convince your average creationist—not that anything would (Meyer et al. 2012). That said, Joyce remains in the forefront and may yet smoosh chicken and egg together (cf., Robertson and Joyce 2014).

Instead, after another 20 years hiatus the big idea that we could look at the whole landscape to determine the possibilities inherent in our unique biology has begun to boggle us less, and sequence randomization has begun to move from biotechnology back into biology proper. There have been several papers that have begun to use randomization not just to see what is possible, but what is not possible. Deep sequencing has given us access to all of the sequence landscape, which in turn makes possible the description of full functional and fitness landscapes, at least for shortish sequences. Pitt and Ferre D’Amare (2010) have described the fitness landscapes of ribozymes, and Fowler and Fields (2014) have begun to do the same for proteins (fulfilling in some ways the wildest fantasies of Knowles and co-workers (Hermes et al. 1990)). Seattle has become a hotbed of un-biology, of learning what is out there in order to better figure out what is already here. Soon, there will be enough landscapes to surf that we can begin to figure out what evolution already knows: how to evolve to evolve, how to center sequence in a way that allows an organism to effectively mutate no matter what it encounters in an ever-changing (and ever-cycling) environment.

While I think the Big Idea has in some ways been found, lost, and found again, this is only my own perspective. It is fair to say that many others never lost their way. Chief amongst these would be Jack Szostak. There are many reasons that Jack is maddeningly brilliant, but one of the three attributes that make him truly extraordinary is his ability to see the future far more clearly than almost any scientist I have ever met (the other two: a gifted intuition for picking people, and a remorseless addiction to quality). Jack understood that randomization and selection would yield function; he intuited the possibilities. But I think at some deeper level he understood and understands the landscape as a whole, and it is for this reason that he continues to search for origins. It is not that he knows it happened (as do we all), it is that he can taste its coming in a way that many of the rest of us cannot. Jack knows the umami of the prebiotic world.