h+ Magazine

"If you haven't read about the recent ramp up of neurotechnology research at MIT, please take a moment to read my book review of Zack Lynch's "The Neuro Revolution" (or better yet, take a few days to read his book). My comment about the Neuro and Nano "revolutions" is an attempt to place genomics in the context of these other scientific endeavors that may or may not come to fruition, but have tremendous transformative power. If you feel that "revolution" is too strong a word to apply to the potential of neurotechnology or nanotechnology, I'm not the only one (journalist or scientist) to use it in this context (these fields very much have the paradigm-shattering characteristics described in Kuhn's "The Structure of Scientific Revolutions.")"

While I've not read "Neuro Revolution" - I tend to avoid books purporting to cover scientific topics that have the word "revolution" in the title, just as I avoid emails that have the words "easy riches" in the title - I am fairly familiar with the underlying research literature. Having been a doctoral candidate (on "slow motion" track to completing my dissertation, admittedly) for almost ten years in a closely related field - Systems Science, research topic "Quantitative Consciousness and Neurocomputation Models" - I do my best to keep abreast of the substantive research in the field itself, and rarely allow myself to step back and read the "popular" summaries that are tailored more to selling a particular flavor of "revolution" to the consumer masses.

--------------------------------------------

"Venter may be controversial and optimistic, but I would hardly call his claims "hype." In a study just published in Science, his team has taken a genome from the bacterium Mycoplasma mycoides and transferred it to a yeast cell. After altering the genome in several key ways, they transplanted it into the hollowed out shell of a different bacterial species, Mycoplasma capricolum. The altered genome started functioning and instructed its host bacterium to produce colonies of M. mycoides."

Venter's work, as I said, is profoundly interesting basic science - genomic science. It isn't a computational line of research, nor does it even live in the same neighborhood. Swapping genomes around, as a practical challenge, is non-trivial. However, neither Venter nor any other working geneticist or bioinformatician with whom I'm familiar is making ANY claim to have genuinely modeled the real-time interactive matrix that translates genomic raw data into living wetware.

The beauty of the field of genetics, from a computational perspective, is just how well it highlights our complete failure to construct viable mathematical models of systems such as the genetic strands of DNA/RNA. It challenges the computational folks to dramatically re-envision the linear models to which so many physicists and mathematicians had become so utterly, hopelessly attached. Indeed, a professor of mine (Dr. Zwick) remarks that, in the universe of known systems in the physical world, those that can be viable modeled using linear tools are a fraction of a percent; the rest are non-linear and qualitatively different to explore using mathematical tools.

Conflating Venter's powerful, largely tactical work in the mechanics of swapping genetics elements around in lab microbes with the parallel challenges of making sense of the computational framework within which those strands actually do their real-life work is a logical fallacy. My effort to point that out involves no denigration of Venter's work - it's simply comparing apples and grapes. The charge of hype lies not with Venter, to be clear, but with the original article published here.

----------------------

"The article never makes any claims about "avoiding" evolution, it suggests that genomic engineering and the design of organisms -- as opposed to using only blind trial and error -- involves mathematical modelling and well as genetic sequencing. Mathematical models aren't necessarily tied to the number of transisters on a silcon wafer (Moore's Law) except maybe to run faster; they are dependent upon the sophistication of the software algorithms (combinatorial optimization, evolutionary, etc.) used to create the models. Evolutionary models are just that: models. And they can inform the design process."

Your statement about the "blind" process of natural selection, i.e. evolution, carries a substantively different meaning than this more inclusive re-definition of mathematical models to, in fact, include evolutionary design. Further, by contrasting the "limits" of Moore's Law with the (putatively unlimited) "revolutionary" power of genomic computation, you do in fact suggest that non-silicon mathematical tools are in the pipeline which promise to exceed the developmental curve of silicon-based systems.

Otherwise, why bring Moore's Law into the discussion at all? As a straw-man, it must be something other than that which you are proposing as the "revolutionary" new replacement for it, correct? In fact, I've suggested that an effort to compare Moore's Law with genomic advances (or the "neuro revolution" if that strikes a more exciting note) is comparing apples to, say, Lego. Different in class and in quality.

In the field of algorithmic analysis, we use "models" and "algorithms" somewhat interchangeably; referring to "algorithms" that are used to "create" models is a bit of the snake swallowing his own tail. Whatever creates algorithms - saying that they then create "models" does suggest a fundamental recursion of logic. If the models make the models, then what made the first model? Either we acknowledge that someone (i.e. a biological human) "programs" the underlying algorithm - whether that algorithm then makes "models" and so on from there, or not - and compare that process to "blind" natural selection, or we're engaging in rhetorical obfuscation. The answer cannot simply be "it's turtles. . . all the way down."

Also, Moore's Law does NOT merely refer to increased "chip density." Please check your references before failing to accurately present a central concept such as this.

-------------
"Nor do I make such a claim about genomics in the article. I do think, however, that both neurotechnology and nanotechnology may have the transformative power to become revolutionary."

Again, circularity of logic: if you claim that these two (entirely unrelated) fields - "neuro" and "nano" - are "revolutionary" in scope, then you must be claiming that they are poised to DO something. Right? You aren't referring purely to Khunian paradigm-jumping intellectual "revolution" from the way the article reads. Indeed, in dragging Moore's Law into this you explicitly set up a promise of tactical, practical "revolution" in concrete, techno-economic behavior of human beings.

What are these "revolutions" supposed to entail? The nano stuff - yes, we all know how that plays out, having read Stephenson's "Diamond Age." But the "neuro" revolution? Given that we have little or no real understanding of neuro-anything at this point (not sure how the popular press is spinning this core truth, but in the research itself nobody would question the nascent state of much of the basic research work itself) I question whether it can be hyped as "revolutionary" just yet. Not only is the cart before the horse, but the cart is out on the road, and the mare is still in her stall eating breakfast. A bit premature to be calculating the rate of acceleration of said mare/cart combo.

I really don't understand at all how Venter's work fits into either "neuro" or "nano" anything. He's doing genomics. Different field entirely - and different from computational/algorithmic research, as well.

These are all interesting fields, and much truly wondrous and clever work being done by researchers far more skilled and qualified than I. At best, I can raise a warning flag and suggest that tossing all these diverse concepts and buzzwords into a virtual Cuisinart and producing a melange of mixed-metaphor mushiness does a disservice to all these endeavors, combined.

Then again, I live down in the weeds of at leas a few of these specialized fields, so I'm going to be unnaturally concerned with maintaining the analytic purity of discrete subjects. That said, it still seems to me that Shake 'N Bake models of recombinatorics could result in novel prospective pairings. . . but only at risk of coming perilously close to the "blind" modes of natural selection which are, as you say, about to be left behind by the power of magical "mathematical models" to build models that, in turn, build themselves. Or something.

Fausty | www.cultureghost.org