🚀 go-pugleaf

In article <Jan.19.22.41.18.1993.23082@planchet.rutgers.edu> cuhes@csv.warwick.ac.uk (Malcolm McMahon) writes <in reply to me>:

>>I doubt that any (contemporary) computer system can be programmed to be
>>sufficiently 'intelligent' as to perform this task.  Aside from that, the
>>task would be monumentally even for a human to administrate:
>>
>>What is an explosive ?
>
>I  don't think I can have explained this very clearly since you are  the 
>second person to make this point.  I am saying that the replicator  only 
>knows  how  to  make a limited number  of  molecular  architectures.  It 
>doesn't  have to say "that's TNT,  that's a no no" but simply "What  the 
>hell's TNT?".
>
>The  user  code doesn't describe the molecular architecture  but  simply 
>selects  it from a list of structures that the system CPU knows  how  to 
>make.
>
>Essentially  you build your artifact out of a fixed variety of  building 
>blocks  and  other  components.  As new  materials  are  discovered  new 
>generations of this nano will have them added to the list.

Ah!  Got you.  Well it still doesn't quite work.  Chemistry is just to
complex a subject that any compound/structure could be classified as
entirely safe.  However, your approach certainly makes the job of the
hypothetical psychopath much harder.

eg) He could make a steel sphere 6" thick containing (insert arbitrarily
    high number) Pascals of Chlorine (or other gas which won't liquefy
    at room temperature) and packed around with nails.

(this is a poor example but it only took me 30 seconds to devise)

>>The question of dissassemblers is equally complex.  What should the
>>CPU forbid to be dissassembled ?  People (obviously) but at the nano-
>>level a person does not look that different to the remains of a TV-
>>dinner (insult your entire audience with one phrase).
>
>I've been giving some though to disassembly,  obviously sooner or  later 
>we are going to want to get rid of each of these new toys (quite  appart 
>from the dissassembly of feedstocks).  What I would suggest is  this:  A 
>frequent  intervals  when constructing a molecular  structure  the  nano 
>prints  a  serial  number into it in some form (say  every  few  hundred 

This immediately puts a huge limit on the usefulness of the technology
because the serial numbers would be larger than some of the things that you
will want to make, nano-componants for example.

>nanometers) you can then have a variety of nano which will  dissassemble 
>only material that bears the serial number it's been fed.
>
>     You  could  make  it a general rule  that  dissamblers  will  touch 
>nothing close to body temperature.

Good. but how does a nano-scale device assess the temperature of an object
1,000,000,000 times larger that it is ?

>>Better ideas ?  Well I suspect that an entirely general assembler is
>>a vast distance in the future as yet.  The problem is far simpler for
>>the more speciallized classes of nano-machine that must certainly come first.
>>
>
>I don't think so. Actually by the time you have a replicator you are most 
>of the way. You just have to solve docking, comunications, problems of 
>cooperation and what to do about nano that dies on you.

Well obviously this implies that replicators are a long way in the future as
well.  (A rreplicator is a special case of an assembler, no ?)

>>One simple approach is the hot-house nanotechnology which is frequently
>>mentioned.
>
>I  don't find the hot-house idea very credible.  What  conditions  would 
>nano  be  likely  to  require?   Not  high  tempartures  and  pressures, 
>certainly,  in fact thermal motion is an important problem.  More likely 
>vacuum  and cold.  Vacuum is easy enough to do at home.  Let's  look  at 
>cold.  Suppose,  to  take the extreme case you had nano that could  only 
>operate continuously in surroundings at a few degrees  absolute,  liquid 
>helium temperatures.  The operating temperature would have to be a  good 
>deal higher (to create the thermal gradient and dissipate the heat).  So 
>nano  that  would operate continously with  liquid  helium  temperatures 
>would  almost certainly operate briefly at liquid nitrogen  temperatures 
>that are readily available.

I wasn't thinking of special conditions of physics, but rather special chemical
conditions.  It seems fairly obvious to me (whatever that means :-) that all
early nano-constructors (where constructor is any form of catalysis/assembly
short of a general assembler) will operate on special chemical substrates.
Simply because that will be easier than going straight for a "grow houses from
sewage" approach.  These special chemical conditions will certainly require
the presence of "exotic" chemicals and further, they might require these 
chemicals to be kept within rigid concentration limits by assay/inject feedback
systems.

Its entirely possible that the 'life-support' for a vat of nanomachines will
weigh several tons.  (have you ever seen commercial bioreactors/fermentors ?)

>     Now  the biological cells evolved in a warm salt ocean and many  of 
>them  require such conditions to operate properly.  They have  developed 
>systems  for  creating  such environments  in  hostile  places;  they're 
>called animals.  A replicators capable of building macroscopic artifacts 
>would  almost  certainly be capable of building it's own  hothouse  (or, 
>more likely fridge).

No organism refigerates, however.

>     I  can  envisage  an "egg" sealed within which was  the  seed  nano 

Who would make such an egg?  Legitimate industry would have no need for
such a complex solution.  It could have a suitable vat waiting for the nano's
when they arrived.  Granted, a government/military might want to be able to
drop nano's on hostile ground but I don't believe that *any* of the things we 
have discussed will control a government.  I don't know how you do that (if
anybody says 'vote' I'll scream !).

>together  with  enough  material  to create a  larger  enclosure  in  an 
>environment  in which the nano could operate long enough to  build  that 
>enclosure.  It  would  burst  out of the egg  as  a  swelling  membrane, 
>equiped with selective transport mechanisms to take in raw materials.

No.  In order to do that nanatechnology will have to have already passed
through the phase we were discussing.   Manufacturing a fridge would be a
*very* advanced topic in the nanotechnology text-book.  We will get the
sort of problems that you first suggested at a far earlier stage than that.

(This seems to be (IMHO) a common problem on this news group.  The assumption
 that what will happen is that a fully developed and functional nanotechnology
 will be suddenly dumped in our laps.  This isn't what will happen!  New
 technologies always arrive by stages.  Nanotech <by virtue of the extreme
 technical difficulties and the shear complexity of the tasks it will enable>
 will probably arrive by more stages than most.)

>> that a deliberate requirement for some
>>suitably rare chemical can be built into a nano-device.
>
>Yes,  I think the key chemical idea will certainly be explored, probably 
>not  so much as a safety feature as as a means by which those that  make 
>and  program  nano can charge hefty fees of those that use  it.  But  if 
>nano  can  recognise  the  key it can  analyse  and  duplicate  it.  The 
>incentive to break such "copyright protection" would be very great.

No, no, no.  That isn't what I meant.  I wasn't suggesting that the
assembler require a 'key' which it would recognise.  Rather I meant that
at some stage of it's life (say reproduction) it would require a raw-
material THAT IT COULD NOT MANUFACTURE.  Obviously it could probably
make a machine (that could make a machine) that could make the (lets call
it a) 'vitamin' but this would be far harder for our potential genocide
to do.

>     I'm not worried about the use of nano to produce illegal drugs.  By 
>the time these things happen I think even the Ammerican government  will 
>have realised the drugs war is unwinable and the DEA and the mafia  will 
>have  moved  their war onto new territory.  There will no  longer  be  a 
>drugs problem.

Yeah!  Let them die! [1/2 :-) It would be a form of evolution in action.]

>Things that do scare me:
>
>1)   The  programmable disease.  It spreads without symptoms for a  year 
>or  so then suddenly,  on a particular date and time it  kills  everyone 
>who,  for  example,  hasn't Praised the Lord aloud in the prior  twenty-
>four  hours  (or  whatever suits the prejudice of  the  megalomaniac  in 
>question.

This is not plausible.  How does the disease know what I've said ?

(I've been meaning to work out some mathematics of macro->nano-scale
 information transfer for some time.  On the face of it it's a very difficult
 problem that I've not seen anybody address)

It could, however, identify a key molecule that would enable it target a
specific race with some success. (cf Frank Herbert (I think) "The White Plague"
in which a biochemist produces an anti-women disease and turns it loose on the
IRA).

>2)   "Ok,  I want you to dissassemble that big lump of pitchblend, weigh 
>each uranium atom,  and build the ones that are the right weight into  a 
>nice hollow sphere."

Another unanswered question.  How does a nano-scale device identify
a single atom (not by weight).

>3)   An  obscure  bug in the software of widely used  symbiotic  medical 
>nano.

This is very freightening, but again, you are assuming a far more advanced 
technology than the stage we were discussing.  The problems with a
medical nano are far and above harder than the problems of a replicator.

>4)   The  mess  that we're going to make of the transition from  a  paid 
>work oriented society.

Hmmmmmm.

i) How do you know what will happen in that much detail ?
ii) Transition to what ?

>5)   The  lack of thought those who are working on AI seem to be  giving 
>to matters of motivation.

Sorry, this went straight over my head.

>Malcolm McMahon