Feynman Grand Prize comments
Hal
The Foresight Institute is offering a large prize, currently funded at
$250,000 with hopes of an increase to $1,000,000, for development of
nanotech devices. See <URL: http://www.foresight.org/GrandPrize.1.html
>. Basically you have to build a robotic arm capable of .1 nm positional
accuracy which fits with a 100 nm cube, plus an 8-bit binary adder which
fits within a 50 nm cube. You have to build 32 copies of each.
These are ambitious and interesting goals, but I thought there was one
thing wrong with the adder spec. It says:
The prize winner must also design, construct and demonstrate the
performance of a digital computing device that fits into a cube no
larger than 50 nanometers in any dimension. The computing device must
be capable of:
* adding accurately any pair of 8-bit binary numbers, discarding
overflow.
* accepting input signals of specified types (see below).
* producing its output as a pattern of raised nanometer-scale bumps
on an atomically precise and level surface.
The problem is this last point. This would seem to preclude an
electronic device, or at least provides a strong bias towards a
mechanical one. But this may not be at all a desirable bias to install
in nanotech research at this early date. The elaborate cables and
pulleys used to build circuits in Nanosystems are supposed to be proofs
of concept, not predictions of optimal designs. Drexler and co. have
consistently said that they do not necessarily believe that mechnical
devices are going to be the best way to do calculations, merely that they
are offered to demonstrate that non-scale calculations will be possible.
But they have always left open the possibility that electronics will be
a superior technology for this purpose.
I think it is unfortunate that an all-electronic 8-bit adder of the
required size would be disqualified simply because it had to produce
output in the form of bumps rather than electronic signals. (Oddly,
electrical inputs are allowed, but outputs must be mechnical.)
Presumably the device will have to be augmented by a bump-making output
stage which will be clumsy and inappropriate if electronic signals are
used elsewhere.
Hal Finney
ELY001
Hal Finney <hfi...@shell.portal.com (Hal)> wrote:
>>>>>>>>>>>>>>>>
The prize winner must also design, construct and demonstrate the
performance of a digital computing device that fits into a cube no
larger than 50 nanometers in any dimension. The computing device must
be capable of:
* adding accurately any pair of 8-bit binary numbers, discarding
overflow.
* accepting input signals of specified types (see below).
* producing its output as a pattern of raised nanometer-scale bumps
on an atomically precise and level surface.
The problem is this last point. This would seem to preclude an
electronic device, or at least provides a strong bias towards a
mechanical one. But this may not be at all a desirable bias to install
in nanotech research at this early date.
<<<<<<<<<<<<<<<<
This is a bias from the computational perspective, but is justified if one
wishes to use outputs to drive or control a mechanical system, such as an
assembler.
In particular, it would be undesirable to award the Feynman Grand Prize to
developments in electronics which do not advance or provide capabilities
fascilitating mechanosynthesis and/or tystems.
In particular, if an electronic type device fitting the spec were
produced and could eventually be produced cheaply, there would be one
_less_ motivation to develop strong-mnt. (Though there would be greater
computational capacity to model what still could not be built. ;^/ )
Motivation, of course, means a concrete, immediate and financial
interest in seeing that some state of affairs comes to pass, i.e. that
some one or some entity with the resources to do anything meaningful
would put money and energy into efforts towards a particular goal. For
example, if the semiconductor industry were to perceive that it is going
to hit "the wall" soon enough and expects not to reap the advantages it
has enjoyed under Moore's Law (i.e. doesn't see growth or market size
sustained by product specialization and differentiation but perceives
that _business_ will decline as evolutionary miniaturization rates
decline,) it might become motivated to develop strong-mnt if no
alternatives are perceived to exist. Molecular electronics, DNA
computing, holographic processing, and quantum computing are among
nascent perceived alternatives. Further, it is not clear that "the
wall" will occur where it is presently projected--the projection has
slid backwards a number of times.
- - -
We should all be clear, however, that prizes are retrospective rather
than prospective. Though they may provide incentives to individual
researchers, they do little to directly advance the channeling of
resources into a particular line of work. At best, the existence of
such a prize objectively manifests to observers that some group or
organization considers some goal important enough to put their money
where their mouths are, to the tune of the prize sum. This may (or may
not, depending on reputation) give the goal a small degree of
credibility to the financial community. It may also get a few more
people thinking about the problem harder with the aim of winning the
prize, but it will do little to enable them to do anything about it. I
have to wonder, though, whether anyone clear on the concept of
strong-mnt is about would really care much about winning the prize
_sum_. I submit that anyone either attracting the resources to do
anything meaningful towards accomplishing the above goals (a) doesn't
need the money, by definition; and (b) won't need the money, unless they
manage to completely loose all IP rights, get no royalty, and also fail
to parlay the accomplishment into significant career opportunities.
E.
John Mark Michelsen
> The prize winner must also design, construct and demonstrate the
> performance of a digital computing device that fits into a cube no
> larger than 50 nanometers in any dimension. The computing device must
> be capable of:
> * adding accurately any pair of 8-bit binary numbers, discarding
> overflow.
How many and, or, not gates, etc. does a device able to "add accurately
any pair of 8-bit binary numbers, disregarding overflow" require?
John Michelsen
[A half-adder (which adds two one-bit numbers producing a two-bit
result) can be built with two inverters, three AND gates, and an OR
gate. A full adder (which adds three one-bit numbers producing a two-bit
result, and thus can be cascaded to build bigger adders) can be built
with two half-adders and an OR gate. You'll need a half adder and seven
full-adders (saving three gates in the last one by discarding overflow)
for a total of 94.
You can save gates if, like rod logic, you can invert the sense of
signals without extra hardware (locks that block on zero are no bigger
than locks that block on 1) or have a hardware XOR gate.
--JoSH]
Hal
ely...@aol.com (ELY001) writes:
>Hal Finney <hfi...@shell.portal.com (Hal)> wrote:
>The problem is this last point. This would seem to preclude an
>electronic device, or at least provides a strong bias towards a
>mechanical one. But this may not be at all a desirable bias to install
>in nanotech research at this early date.
>This is a bias from the computational perspective, but is justified if one
>wishes to use outputs to drive or control a mechanical system, such as an
>assembler.
However, you would not need to convert to mechanics except at the final
output stage of your computer. The convertor should be a specialized
device, not something which gets counted as part of each adder, IMO.
I see a couple of different way of looking at this Prize:
A) Something to motivate people to create devices (robotic arm,
adder) which will be useful in making assemblers.
B) A test to see whether nanotech assembly capabilities have
advanced to the point where complex devices can be built.
The thing is, building a robotic arm 100 nm in size is going to be
really, really hard to do! You are going to have to have the ability to
build virtually any device of that size in order to build that thing. So
you must already have strong nanotech assembly capabilities in order to
win the Prize. In this sense (B) the Prize might equally as usefully have
rewarded the construction of a scale model of the Eiffel Tower.
However the one thing lacking from "full nanotech" that might still allow
winning the Prize is the ability to produce large quantities. Only 32
copies of the devices are required. So if in fact nanotech develops
along a path which lets you build 32 robot arms but not a lot more, then
perhaps these robot arms could be used to build a more flexible
assembler, capable of self replication, which would lead to macroscopic
level molecular manufacturing. This relates to motivation (A) above.
So what I see in the Prize is an assumption about the path that nanotech
will follow: it will be based on probe microscopy, allowing eventually
good control over atomic assembly but at a very limited scale of
replication. At that point the last hurdle that must be met is to build
an assembler system which is small enough to replicate itself.
However if nanotech follows a different path, say one based on advanced
biotech, with protein based assemblers gradually acquiring the ability to
work with sturdier materials, then the Prize becomes almost irrelevant,
serving only as an extra "congratulations" once full nanotech assembly
and replication capability is developed.
I personally feel that the biotech path is interesting because it
offers the possibliity of a range of spinoff products throughout the
development period. Macroscopic quantities are available from the
beginning, and it is the quality of the materials produced which
improves as technology advances. Probe based research will be much more
limited in producing any significant quantity of product even if it has
more precise control initially.
Again, I hope the Prize doesn't end up diverting research down a path
which is not in other respects the most promising.
Hal Finney