Nanotech Doubts

Originally from GreyGoo.

Actually, no, cells aren't nanites. They're microscopic bags of protoplasm, not nanoscopic robot factories. Apart from the great difference in scale, there's a real difference in style too; organic cells do their work by osmotically filtering big membranes full of molecules that tend, statistically, to produce what's desired; nanobots are little pick-and-place robots, tiny self-reproducing precision tools that can tear through a cell like a sawmill through a rainforest.

I'm not convinced.

Why, exactly, should nanoscale robots be far more capable than bacteria? We've never built any viable, self-replicating systems on these scales, so there's no reason to think we have an especially good grasp of the engineering involved.

Perhaps osmotically filtering big membranes full of molecules is a clever strategy. Maybe nanoscale design is hard, and poorly modularized. Maybe small-scale self-assembly inevitably includes nasty, mind-numbing math (ProteinFolding?, or something analogous).

If so, we would need to design our nanobots the hard way. We'd have to use modern heuristic techniques, such as SimulatedAnnealing and GeneticAlgorithms. Eventually, we'd wind up with highly optimized designs that no human could really understand.

Oh, wait.

Nature has already run this experiment - an entire ocean of nutrients and energy, twenty-minute generation times, two billion years of trials, plasmid exchange, and hostile competition. That's not, at first glance, a bad search strategy.

Now, we humans are based on an early version of this architecture, and our cushy multicelluar design allows us to hide a lot of really awful CodeSmells in our DNA. But we're just a sideshow from a biomass perspective.

Our far more successful competitors, the bacteria (and their distant single-celled relatives, the phytoplankton), outnumber and outmass us. They live in a smaller, faster, and far more competitive environment, so they need to TravelLight. They RefactorMercilessly, and they don't accumulate much junk DNA.

Outnumber: the normal human is host to about 10^12 bacteria on the skin, ..., and 10^14 in the gastrointestinal tract. The latter number is far in excess of the number of eukaryotic cells in all organs which comprise the human body. -- Kenneth Todar http://www.bact.wisc.edu/Bact330/lecturenf

Actually, there is no such thing as junk DNA. It's an outmoded concept. Introns play lots of vital roles in gene expression, from acting as binding sites for promotion/inhibition, through nucleic structural support, to non-local signalling via various forms of ion exchange. Geez, keep up won't you?

Hmpf. Our genomes are polluted with useless crud, because we can afford to carry it around. So some recent research suggests that a few of our gadzillion self-replicating Alums have been hijacked into doing something productive for a change. That still leaves all the rest of the freeloading Alums, the broken retroviruses, the endless supply of useless genes for producing reverse transcriptase, and all the other crud. YouArentGoingToNeedIt, and all that other Wiki stuff. :-) No self-respecting bacterium would carry around this much legacy code. Poison a bacteria colony slowly, and it will often evolve to become metabolically dependent upon the poison. Now that's ExtremeNormalForm for you. -- DramaticIdentity
- Those 'broken retroviruses' may be part of our strategy for identifying and eliminating working retroviruses. When (if?) a biologist says 'Junk DNA' what he really means is "I don't know what this bit does". 90% Junk DNA is more a reflection on our understanding than anything else!
  - For a datapoint of one, I am a biologist, and when I say "junk DNA" what I really mean is "I don't know what this bit does and probably it just takes up space, which may be useful on an evolutionary level, but probably isn't on the organismal level." You might also notice that most of the "junk" DNA is between the genes themselves, and not just between the exons that make up the genes (introns). --AndyPierce

Hmm. Maybe bacteria will eat the grey goo. I mean, we ourselves probably wouldn't last a week without our brutally sophisticated immune systems. Why should a bunch of poorly-engineered, slow-to-evolve nanobots (without either PlasmidExchange? or SexualReproduction?) win this arms race?

Simple, really. DNA machines are an invention of RNA machines. RNA is the self-replicating polymer goop that's been demonstrated to form spontaneously in nature. RNA just loves to hybridize, but it has this problem with stability. But it's easy for it to hybridize with the elements of DNA, and DNA is very bloody stable. Add a little autocatalysis to construct some machinery to make proteins from the RNA that's transcribed from the DNA, and you've got about all you need to make any form of carbon based life you can find in nature. Yeah, yeah, there was bootstrapping all over to make things like ribosomes and nucleic structures, but you get the idea.

The cute thing is that everything else in nature is generated from the proteins that get expressed from this stuff. BUT proteins aren't particularly strong or efficient machines. What they are is tremendously versatile and reflective.

So there's lots of ways we can make better tools for specific purposes. Think of the protein machines as being like a whole bunch of lego blocks. You can make anything you like out of lego ... but it has a minimum size much larger than the smallest machine components you can make if you don't restrict yourself to lego, and all kinds of physical restrictions that other ways of building don't have.

On the other hand, proteins have some deep advantages. They support SelfAssembly quite nicely. You can mass-produce them quite cheaply using standardized conveyer belts and a digital specification. You can get the individual components OffTheSelf?. They're compatible with the IndustryStandard AdenosineTriPhosphate? power sources. And best of all, you can even optimize or redesign them - automatically, in the field - by running a measly ten or twenty thousand generations of a massively parallel GeneticAlgorithm. Who wants to use a bunch of fragile, unproven PickAndPlace? VaporWare that hasn't even demonstrated SelfAssembly or the ability to reconfigure itself in the field? -- DramaticIdentity

As to all the aeons nature took to evolve beasts like us - who cares how long it took? The main thing is we've already demonstrated smarts to build tools nature can't equal. So the burden of proof is on you to explain why in this one domain we shouldn't be able to do so. Particularly when there are already plentiful realistic designs on the books for nanomachines that will. And these things don't have to develop through the slower-than-grass-growing processes of evolution - they can rely on us to develop them. Or once we crack AI - maybe with QuantumComputers - they could do a very handy job of that themselves.

: Nature has built both widely successful nanobots and the human brain. So far, we've built the space shuttle and the Pentium IV. I'd say we're pathetic amateurs impressed by our ability to kill off a bunch of highly specialized, unwieldy MegaFauna (most of whom were scheduled to be killed off in the next MassExtinction?, anyway). The interesting stuff all involves SelfAssembly on the microscale, and so far, we don't have a single success to our credit. (Oh, sure, we could spend many millions of dollars to produce a single, specialized nanomachine that works much better than a specific protein. But if it can't self-assemble itself in an energy-efficient fashion, it'll never survive in the RealWorld.) -- DramaticIdentity
: Nature only barely managed to build the human brain using the resources of the entire galaxy over several billion years. Within a couple thousand years, and with a population of less than 100 billion in total, we've positioned ourselves on the edge of harnessing molecular nanotechnology. Oh yeah, 99.99% of the time humans have spent evolving socially and technologically was sunk into fighting nature and laboriously rectifying its mistakes. Nature sucks.

Or perhaps Nature only found a (very stable) LocalMaximum for efficient molecular self-replication, and spent the last 3.6 billion years of heated competition wandering around the plateau.

Yep. Nature actually does that quite a lot.

And it's the whole weakness of massively parallel genetic algorithms. Sure, they'll find any local maximum (especially one with a large enough attractor basin), but they'll happily overlook some far superior solutions.

On this planet, right? And nanobots seem to be much better suited for cosmic travel than bacteria. So why don't we see them here?

Who says we don't? Please donate to the SearchForIntraTerrestrialIntelligence!

Also, read BloodMusic for a few ideas on what nanomachines could do. (It's a book about organics, but it has some interesting ideas. ;-)

Living cells have mechanisms for repairing damaged DNA and other defects. What do the nanomachines have? Once you start adding such things you are getting back toward a cell again, in terms of complexity. And without it the nanomachines will be destroyed by random forces before too long. -- RobertField

They have error-correcting codes, checksums, crypt-authentication algorithms like RC4, and so on. Anything nature can do, we can do better. We can do anything better than Nature.

No you can't.

Yes we can.

No you can't.

Yes we can.

No you can't!

Yes we can, yes we can, yes we can!!!

Since we're part of Nature, how can we do anything better than it?

(I do love the cut-n-thrust of intellectual debate)

How do nanomachines deal with thermal noise? Proteins tolerate it - they flex and jostle and breathe, and their active sites are in a more-or-less right conformation enough of the time to get their jobs done; they certainly aren't rigid. Nanomachines, on the other hand, are depicted as being like machines only very small; it's hard to square the implied rigidity with the colossal thermal motion observed at atomic scales.

The "thermal noise" issue has been raised so often that Ralph C. Merkle wrote a FAQ about it: http://www.zyvex.com/nanotech/impossible.html http://foresight.org/impact/impossible.html

We could just try PositiveReinforcement? and NegativeReinforcement?.

"Bad nanoprobe! Bad!" --PhlIp

See also, or perhaps instead, NanotechDesignProblems.

The most immediate benefit of Nanotechnology will be improvements in everyday electronics and machines. Like ArtificialIntelligence and VirtualReality, the hyped vision may not come to pass but aspects of the technology will make existing products better, hidden away so you would not even know it is there. For instance really really small components should make possible a thin HandHeld with

WiFi + BlueTooth + GPS + 3G + video + FM Radio + TV + satellite (when really far) + lots of fast memory + Digital wallet + long life rechargeable batteries + + bright hi res screen viewable well outdoors + ...

All of which you can do now but only by juggling CompactFlash? and SDIO cards and quickly draining batteries. The Sony UX50 has video + WiFi + Bluetooth and some of the other features but no SDIO (only memory stick) so you can't add what is not there except bluetooth enabled devices such as some GPS units but then it is an extra thingy to carry. But it is moving that way - I would not be surprised if some of what is in the UX50 could be classified as what was once thought of as Nanotech.

Then you'd be completely dead wrong. In fact, you'd be an idiot who doesn't have any clue what MolecularNanoTechnology is. Let's be really crystal clear here. What "was once thought of as Nanotech" is uniquely MolecularNanoTechnology. And it has nothing to do with 1 nm features. The presence of 1 nm features is just marketing hype by silicon manufacturers.

Ah, I see you've found religion...or are someone who should take the time to remove their foot from their mouth long enough to have a dialog and understand some of the details of modern pda components. Example see

Nano based displays - http://www.japancorp.net/Article.asp?Art_ID=5726 Nano Antennae - http://www.nanoxchange.com/NewsNewsstand.asp?ID=185 Nano based power - http://www.smalltimes.com/print_doc.cfm?doc_id=5719

These are not just further miniaturization of silicon circuits. In any case useful engineering systems are a combination of electrical, mechanical, chemical, fluid, computational etc subsystems to try divorce one set (ie mechanical nano assemblies as "above" 1nm electronic features) from the other is nonsense. They will work in conjunction.

I don't need to follow those links because I know exactly what you're talking about. MEMS. And you know what? Fuck them. MEMS are in no way MOLECULAR, which has always been the meaning and goal of MolecularNanoTechnology.

But but but you're gonna whine "they've got the word "nano" in them". Nobody gives a fuck. Nanometer features was never the goal of nanotechnology. I repeat myself, nanometer features are not, and have never been, and never will be, the goal of nanotechnology. That refers to any features; electronic, optical, mechanical, chemical, thermal, fluid, computational, hell even transwarp assemblies. The "nano" in nanotech is a misnomer.

The whole point of nanotech, as far as all non-marketroids are concerned, is self-replicating machines. Not whatever latest PDAs or cure for hemorrhoids you're peddling. MEMS are not nanotech. PDAs are not nanotech. Nothing in any development shop is nanotech. The only things that are even vaguely related to nanotech are STMs, scanning tunneling microscopes and biochemistry.

Oh, and buckytubes are not nanotech either.

I really hate marketroid time-wasters and LanguageAbusers.

See: LaynesLaw

CategoryNanotechnology