Police Nanites and Nanorobot Population Limits
Robert Freitas
Police Nanites and Nanorobot Population Limits
I. Can We Detect and Monitor the Activities of Grey Goo?
John Mark Michelsen <jmic...@rigel.oac.uci.edu> wrote:
> These could be used to monitor the heat dissipation (infrared
> radiation) of the area surrounding them, and to broadcast that
> information. This could allow very fine grained information to
> be obtained of the environment. Large amounts of work going on
> (heat dissipation) would then draw attention.
Michelsen is probably correct that free-range gray goo may be readily
detectable from its energy signature.
Consider, for example, a colony of self-reproducing nanorobots "making
hay" inside an automobile tire, as has been proposed in previous
postings in this newsgroup. The average covalent bond in diamondoid
materials requires ~200 kT to form; a 1 micron^3 self-reproducing
nanorobot may contain ~20 billion structural atoms; hence total
replication energy starting from atoms or small-molecule feedstock is of
the order ~20 nanojoules, requiring ~1000 sec per replication
(Nanosystems:Chapter 14) for a 20 pico watt device, giving a relatively
mild power density P = 0.02 gigawatts/m^3. This is very roughly
comparable to the performance achieved by the fastest-breeding microbes
known.
Even at this modest power density, a ball of replicating nanites of
radius R = 3 sigma T^4 / P > 0.3 millimeter has a black-body temperature
T > 450 oK, exceeding the melting point of rubber (sigma = 5.67 x 10^-8
watts/m^2-kelvin^4) -- so even extremely tiny colonies of gray goo
replicators quickly melt holes in the tire surface, causing a blowout.
Alternatively, a 1 m^2 tire surface uniformly coated H = sigma T^4 / P =
116 microns thick with 1 micron^3 replicating nanorobots (~10^14
devices) radiates at 450 oK, emitting a total of 2000 watts; using tire
organics and atmosphere oxygen as the power source, an entire 10 kg
tire is consumed (solely in providing energy) in ~50,000 sec, or ~0.5
day. Undoubtedly the automobile owner will notice.
In the most optimistic scenario, one entire 10 kg tire provides
sufficient mass and energy to replicate ~2 x 10^15 devices of
~diamondoid density, a (13 cm)^3 cube of nanites. However, one problem
with rubber-tire replicators (and other replicators posit ioned in
equally "atomically-challenged" venues) is that these nanorobots may not
have access to atoms from the full range of chemical elements needed to
construct such high replicative-complexity nanodevices. Thus the
replication speed or mass conversion efficiency of "rubber tire" based
replicators may be far less than the optimum figures given above.
II. Mirandizing the Blue Goo
A far more insidious detection and monitoring proposal (see recent
postings in this newsgroup) is the suggestion that blue goo "police"
nanorobots should be unilaterally empowered to examine, and possibly
deconstruct, potentially grey goo infestations at will.
An excellent medium for the growth of a gray goo nanovirus colony is the
human body, which offers ready access to at least 32 of the natural
chemical elements plus abundant energy supplies. If we hope to maintain
any semblance of personal privacy, clearly we cannot permit blue goo to
make unauthorized inspections of our bodies, which would essentially
make both genome and neurome generally available for public viewing. At
the very least, a court order should be required before such inspections
are allowed (courts may be able to act very quickly in the nanoera), or
the intended inspectee should be read his Miranda rights, given the
right to pre-inspect or edit the results, etc.
On the other hand, neither may we permit gray goo replicators to enter
our bodies and digest us to manufacture more gray goo.
If, as I hope, we choose not to permit gumshoe blue goo ("we're from the
government and we're here to help you") to enter the human body
uninvited, various anti-gray-goo prophylactic measures are available to
the individual. For example:
(1) Pink Goo -- Constant patrol of the body by personalized blue
goo-like devices restricted to operation only within the user's own body
and under the sole command of the user, and programmed to continuously
inspect the body for the entry of nanomechanic al invaders of known
configuration as recorded in an onboard nanopathogen library (with
regular informational updates from the societal blue goo police force
describing the latest versions, upgrades, and new species of nanoviruses
that are afoot). (E.g., an artificial immune system for nanopathogens.)
(2) One-A-Day Goo -- similar in function to Pink Goo, the One-A-Day Goo
is periodically admitted into the body (say, once a day) and quickly
performs a sweep of the premises searching for intruders, then, if
nothing is found, promptly eliminates itself from the body.
(3) Think of other alternatives and post them -- I'd be interested in your ideas!
III. Global Nanodevice Population Limit
A distribution of one blue goo police nanorobot per cubic millimeter
over the entire surface of the Earth to a total thickness (altitude plus
depth) of 2 kilometers (maximum biosphere thickness) requires ~10^27
nanorobots. Assuming 100 picowatt per blue goo police nanite (1
micron^3 x 0.1 gigawatt/m^3 power density), total blue goo fleet power
dissipation is ~10^17 watts ~total solar insolation at the Earth's
surface.
This calculation suggests not merely that the number of deployable
police nanorobots is fundamentally limited, but more broadly that there
may exist a worldwide "population limit" of something on the order of
~10^27 active typical 1-micron^3 100-pW nanite s. Planetologists call
this ecological energy release level the "hypsithermal limit" -- release
much more energy than this (10^17 watts) at the Earth's surface, and
clouds thicken, carbonate minerals starts degassing, and Earth's
atmosphere rapidly becomes Venusian and thus uninhabitable by man OR
nanomachine. It is believed that releasing as little as ~10^15 watts
extra energy into the global environment would melt the polar icecaps,
so the 10^27 nanite figure should probably be regarded as an upper
limit. Total human technical civilization currently releases ~10^13
watts into the environment, and local weather (e.g. near major cities)
is already measurably disrupted by these releases.
Note that the nanodevice power density used in this calculation is
driven by the necessity of making or breaking covalent bonds during
mechanochemical assembly operations, and by the minimum physical energy
needed for various sensory, manipulative or loco motive tasks -- NOT by
the computational requirements. Therefore, power density CANNOT be
substantially reduced from the figures given above by simply switching
to more energy-efficient reversible or quantum computational techniques.
Now, 10^27 nanites is 100,000 trillion devices for each of Earth's
~10^10 human inhabitants. This should be sufficient for most reasonable
medical, manufacturing, transportation and other purposes. Still, 10^27
nanites is only ~2 x 10^12 kg of global nanomass, or 200 kg per person.
It appears that energy limits impose a worldwide per-capita nanite
budget. Once the terrestrial nanite population approaches this limit
(10^27 units), acquisition and use of nanites will then become a
zero-sum game. That's because some humans will undoubtedly wish to
command hundreds of tons of active nanites, thus denying poorer or more
ignorant "have-nots" their "fair share" access to the global nanoworks
in order to avoid ecological catastrophe. (God I hate sounding like a
socialist!) Perhaps the new unit of money will become the "teranop"
(trillion nanomachine operations). Operations off-planet are, of
course, not subject to this limit.
In theory a madman could create a gray goo designed to exceed the
hypsithermal limit and thus purposely destroy Earth's ecology. However,
if we assume that even gray goo will be designed to respect this limit
then the maximum global nanomass (~2 x 10^12 kg) falls far short of,
say, the ~2 x 10^15 kg of global forest biomass. If nanodevices can
process on the order of their own mass in a timescale of ~1000 seconds
(e.g. Drexler's exemplar manufacturing system), then the time required
to consume ("gooify" ) the entire forest biomass is ~10^6 seconds (~12
days) even if the entire (2 x 10^12 kg) nanomass were somehow corrupted
to pursue this task, an extremely unlikely event. (The energy release
during such intensive gooification would be the equivalent of lighting
off one 20-megaton fusion bomb every second.) This gives the blue goo
police plenty of time to detect and react to the destructive activity.
IV. Getting a Jump on the Blue Goo?
David Weinstein: david.w...@virgin.net wrote:
> While gray goo, once formed, could replicate just as fast as
> the blue goo, the blue goo has the "jump on it" in the same
> way that the immune system, once primed, can prevent a
> recurrence almost 100% of the time.
To which Chris Phoenix <Chris....@eng.efi.com> responded:
> The blue goo can't replicate at all once it's deployed. Where
> would it get the materials? But the gray goo is allowed to be
> destructive. Suppose an automobile tire suddenly exploded and
> 90% of its mass turned out to be gray goo, now spread across
> the freeway and infecting other tires. The gray goo now has
> an immense advantage over whatever blue goo is in the area.
First of all, even if gray goo CAN replicate and blue goo CANNOT, this
is NOT a significant advantage for the gray goo if we make the sole
assumption that the blue goo can detect the malfeasance within some
reasonable period of time -- say, a few minutes (which seems
conservative) or, at worst, a few hours.
Why is this? Even under the most ideal conditions in which bountiful
energy and a plentiful assortment of just the right feedstock atoms are
available, replication will require at least ~1000 seconds per
generation. On the other hand, once the malfeasance has been detected
an overwhelming mass of blue goo "enforcers" can be rapidly imported
from a central distribution facility. Fresh blue goo may be withdrawn
from massive floor-to-ceiling stockpiles in warehouses, wakened and
programmed, transported to the site, then slathered atop the infected
area as thick as butter on toast, quickly smothering the slowly
replicating grays.
Actually, this would be massive overkill because destruction is so
vastly more efficient than construction. For example, a nanorobot
manipulator arm with appropriate weaponry could fatally lance a
replicating graybot's nanocomputer in ~10^-6 sec, thus de stroying the
gray goo's ability to continue functioning. By comparison, the gray goo
requires >1000 seconds to replicate (and MUCH longer if it is disturbed
or under attack). Hence, to a crude approximation, a single
aggressively armed police nanorobot can disable ~10^9 unarmed
self-reproducing gray-goobots within a single gray goo replication
cycle. A population of 2000 kg of police bluebots could in theory
disable the entire worldwide sub-hypsithermal nanite population (2 x
10^12 kg) in just 1000 sec onds. Obviously the hardware and software
comprising the policebots must be EXTREMELY fail safe and robust, but at
least this is subject to our design control.
Of course this attack/defense system can quickly grow very complex, as
in any evolving ecology. First the grays are outfitted with defenses or
weapons by their madman creators, and are taught to scatter at the first
sign of the cops. Then the blues star t searching out and presuming
hostile any nonpolice nanorobots found outside of human bodies that are
armed with antipolice weaponry, and start using stealthbots and corral
formations to forestall gray flight. Then the grays adopt mimickry to
camouflage themselves as blues. Then the blues are outfitted with IFF
(Identification Friend or Foe) transponders. And so forth. Ultimately,
after a lengthy period of nanotechnogical co-evolution on both sides,
the likely outcome, as in nature, is stalemate. A m inimum amount of
localized violence may occur from time to time, but the system as a
whole remains stable.
Of course, as with police today, there is no way to prevent a determined
criminal from doing a certain amount of local damage before he is caught
and restrained. This is one of the rationales for the personal
defensive nanorobots described in (II) above -- to ensure the maximum
defensibility of human life until outside help can arrive.
Robert A. Freitas Jr.