3 ms·
The Plausibility of Life by Kirschner & Gerhart OK, this is an oddball one, in that a book on Evo Devo is pretty far from CS, but reading it completely changed
by webmaven 10y ago
The Plausibility of Life by Kirschner & Gerhart
OK, this is an oddball one, in that a book on Evo Devo is pretty far from CS, but reading it completely changed how I look at software architecture, protocols, formats, programming languages, and a whole range of other CS-related topics.
Superficially, it is what it says on the tin: an examination of the mechanisms by which life manages to evolve at all, that keep most mutations from being immediately lethal, and instead have a decent chance of producing (potentially) useful variations.
But under the hood, I've found it extremely interesting how and why so many aspects of the living world are resilient and even antifragile, and there is are lot of useful ideas and concepts to mine beyond relatively superficial biomimicry like genetic algorithms, neural networks, or even "design for failure" approaches for software and services.
Read it, and you won't think about "Worse is Better", "Bus Numbers", or Chaos-Monkeys the same way ever again.
http://yalebooks.com/book/9780300108651/plausibility-life http://yalebooks.com/book/9780300108651/plausibility-life
- div0 10y agoThis book influenced me tremendously. It explains how can random process (e.g. genetic mutation) builds highly non-random structures (e.g. the human eye). Recently I was using American Fuzzy Lop (http://lcamtuf.coredump.cx/afl/ http://lcamtuf.coredump.cx/afl/) to find bugs in my software, and was able to appreciate the surprising effectiveness of the "random" strategy it employs to discover structures in the program. For example, AFL could construct a valid JPEG using randomized input: https://lcamtuf.blogspot.com/2014/11/pulling-jpegs-out-of-thin-air.html https://lcamtuf.blogspot.com/2014/11/pulling-jpegs-out-of-th....
- webmaven 10y agoHey thanks for the link, that jpeg example is really quite cool. One of the things that I gained from the book was a bit better "feel" for getting the right granularity/modularity in order to facilitate evolvability (as in the chunks I might want to swap out for another implementation, while leaving others in place), as a separate concern from maintainability. BTW, I've noticed is that the "order blindly emerging from chaos" theme keeps popping up in fiction by writers like Bruce Sterling, Cory Doctorow, and Charles Stross (all favorites of mine).
- TsukiZombina 10y ago"The Plausibility of Life ends with a brief critique of intelligent design, suggesting that the concept of facilitated variation will provide a solid argument to rebut creationists. I applaud the authors' intention, as it seems to me that more scientists ought to face the realities of public misunderstanding of science. But their presentation is too brief, and a bit too simplistic. The "controversy" about evolution has nothing to do with the soundness of scientific explanations of the history of life: It's not a scientific controversy, but a social, cultural and political one. Creationism is the result of centuries of anti-intellectualism in the United States, coupled with sometimes cynical exploitation of the issue for political gain. In addition, many scientists have no interest in getting out of the ivory tower to talk to the very same public that pays their salaries and funds their precious research grants. The recent defeat of intelligent design at the trial in Dover, Pennsylvania—at the hand of a conservative judge appointed by George W. Bush—will do much more to promote sanity in public education than any theory about facilitated variation, as scientifically sound as the latter may be." http://www.americanscientist.org/bookshelf/pub/have-we-solved-darwins-dilemma http://www.americanscientist.org/bookshelf/pub/have-we-solve...
- webmaven 10y agoThanks, that's an interesting review, though to me that passage you quoted is the least interesting bit about it. ;-) For the viewers following along at home, this passage summarizes why I found the book interesting in the first place: "Organisms are not analogous to human-engineered machines [...] rather, they are characterized by developmental systems that are capable of accommodating quite a bit of disruption—be that from changes in the external environment (phenotypic plasticity) or from mutations in their genetic makeup (genetic homeostasis). This ability to accommodate is in turn made possible by the modular structure of the genetic-developmental system itself, which allows organisms to evolve new phenotypes by rearranging existing components. " To which my emotional response at the time was, "Hey, human-engineered machines are becoming less like human-engineered machines in many of the same ways!" And this bit was the part I found most interesting about this review: " Inherited epigenetic variants can interact with their genetic counterparts to multiply by orders of magnitude the phenotypic variation available to natural selection, thereby expanding the mechanistic bases of evolutionary theoretical explanations and greatly increasing their plausibility as an account of life's diversity." Huh. The analogous mechanisms for software seems like they may be those that don't get propagated by code per-se (but certainly affect it), like the surrounding community. I'll have to think about that some more (and probably eventually read additional Evo Devo books).