Responding to the last post in "Quantum Mechanics and Self-Reproducing Machine
Tools", now off list in my reader:

Bruce wrote on the 9th of September:

-------------------------------------------quote
Doug,

This is a kool idea, as many of your posts are (even drill press abuse). I
have two comments - 1. you have not shown proof that "two of everything" or
even "two of anything" is the smallest number required and 2. I think
replication without evolution is undesirable.

Early man used sticks and stones as tools. Everything we currently have
(good and bad) has evolved from those simple beginnings. So I suggest that
the minimum number of anything is very close to zero. Send McGiver to Mars
with a Swiss Army Knife (and an unlimited supply of bicycle spokes and boot
laces) and by the time you get there, you'll have to hunt for a place to
park your shuttle between the Bridgeports.

Now project that evolutionary capacity to a capability that may be available
in ideal environments and simple replication of current technology may even
be undesirable. Grow your lathe ways from a single diamond crystal - now
your iron machine tools are as desirable as the barber's leeches. Evolve
some more and now you can generate any product by selectively ADDING
molecules of the appropriate type to the proper location - why machine by
chip REMOVAL? When your razor gets dull, it can be sharpened by replacing
the displaced molecules not by abrading until a fresh edge is exposed.

Machine tool self-replication is an excellent philosophy exercise - and that
brings me to the next level of evolution. Perhaps we don't really need
THINGS as instances of matter, we can just THINK them and therefore don't
actually need to perform an actual task or build a particular device to know
the cosmic truth. We think through the problems, then we know that we can
build that 42-shot-simiautomattic-revolver with each part perfectly heat
treated and accurate to a couple of milliangstroms. Now that we know, we
don't have to actually build it to prove our knowledge or skill. (Made you
think about some old westerns, didn't it?)

Just some ideas to think about.

Bruce
---------------------------------------close quote

"Machine tool self-replication is an excellent philosophy exercise - and that
brings me to the next level of evolution. Perhaps we don't really need
THINGS as instances of matter, we can just THINK them"

Oh, heck no. Von Neumann, who proved machine self-replication is possible,
Wigner, who proved it isn't, and Drexler, who missed the point, are all
philosophy exercises. They never provided THING. All they did was THINK.

I plan on actually accomplishing self-reproduction of a minimal complement of
pairs of machine tools, hopefully all from one manufacturer. That manufacturer
may support the project financially for advertising or patent rights.

Pairs are required if the technique of "copying" is to remove a part from a
machine and, if the part is within capacity, installing it in the machine that
does the final operation as a simulated workpiece, then moving backwards one
operation at a time, to the first, at which point the stock is installed. As
one machine _may_ be of the same class, pairs are required.

The final operation need not always be on the same class machine, but if,
working backwards from the finished part, even _one_ operation on _one_ part
requires a machine of the same class as the downed, part-removed machine, we
can reduce the problem to pairs but no farther. And this would be a pair of
bench grinders, a pair of hex key sets, pairs of all collets required, but not
necessarily a full set, a pair of collet closers, etc.

I assert you'll always find at least one milled part in a mill, at least one
lathed part in a lathe, at least one surface ground part in a surface grinder,
and at least one cylindrically ground part in a cylindrical grinder, as the
self-reproduction potential of all machine tools is well recognized. In fact, I
am fairly but not sure that any machine tool must contain at least one part
made with "itself", that is a machine of the same class.

"Evolve some more and now you can generate any product by selectively ADDING
molecules of the appropriate type to the proper location - why machine by chip
REMOVAL? When your razor gets dull, it can be sharpened by replacing the
displaced molecules not by abrading until a fresh edge is xposed. "

With six billion potential partners, I'd rather screw than jack off. You said
so yourself above, and I in my answer. I am working ONLY with what we have and
know. Not with grey goo and magic atom-by-atom assembly. Talk about wasting a
cup of gas to buy a gallon of milk. How about 150 KWH to sharpen your razor?

"Now that we know, we
don't have to actually build it to prove our knowledge or skill. (Made you
think about some old westerns, didn't it?)"

No, I insist that we do it, proving it can be done. Once done, it's addictive.
I don't remember my old westerns.

"Early man used sticks and stones as tools. Everything we currently have
(good and bad) has evolved from those simple beginnings. So I suggest that
the minimum number of anything is very close to zero. Send McGiver to Mars
with a Swiss Army Knife (and an unlimited supply of bicycle spokes and boot
laces) and by the time you get there, you'll have to hunt for a place to
park your shuttle between the Bridgeports."

My middle name is McGiver and a machine tool is a bicycle spoke. I have my own
boot laces, thank you. I also use chewing gum as an adhesive because it
linearizes the connection, eliminating the dead band in the mechanical fastener
that supports it, simplifiying analysis and function."

The minimum number of anything is indeed close to zero. It is one in the case
of certain weapons. For much of the rest, you start with two hands and use
both. Thus, the minimum is two, and above I showed another reason why it must
be two in this case. It's an exclusion principle. If one machine is down, you
need a spare. I build almost everything by making the intermediate products
self-upgrade, starting with a brick or a stick.

But such is stock, not a tool, and you only need one piece of stock to make
something. But to make all your tools, using one of each as a model, the spare
will usually be not just handy, but required. One down, the other up. And when
we have two of everything, we can certainly make one of anything. Sometimes we
don't need two, but the only case in which we never need two is stock, and
stock is not self-replicating.

"I think replication without evolution is undesirable."

So do I and that's why there's no grey goo or unassisted replication in my
thesis. God guided our evolution through his random mutation and selection. We
guide the evolution of self-replicating machine tools after we construct the
first one the hard way. They do not do it on their own, since there are too
many scale discontinuities between Pentium and Clausing Colchester. Within
Pentium, it just doesn't happen. Within Clausing, it could happen but we have
to make it happen.



Yours,

Doug Goncz (at aol dot com)
Replikon Research

Read the RIAA Clean Slate Program Affidavit and Description at
http://www.riaa.org/
I will be signing an amended Affidavit soon.

Doug Goncz scribed in
:

"I think replication without evolution is undesirable."

So do I and that's why there's no grey goo or unassisted
replication in my thesis. God guided our evolution through his
random mutation and selection. We guide the evolution of self
-replicating machine tools after we construct the first one the
hard way. They do not do it on their own, since there are too
many scale discontinuities between Pentium and Clausing
Colchester. Within Pentium, it just doesn't happen. Within
Clausing, it could happen but we have to make it happen.

I must say that this is all very intriging Doug. I'd be happy to
spend year or two with a selection of machines and stock, making
duplicates, or even upgrades. Anyone with funding?

Read the RIAA Clean Slate Program Affidavit and Description at
http://www.riaa.org/
I will be signing an amended Affidavit soon.

why

swarf, steam and wind

--
David Forsyth -:- the email address is real /"\
http://terrapin.ru.ac.za/~iwdf/welcome.html \ /
ASCII Ribbon campaign against HTML E-Mail - - - - - - - X
If you receive email saying "Send this to everyone you know," / \
PLEASE pretend you don't know me.

From: DejaVU

I must say that this is all very intriging Doug.

I'm glad because until I get a reply, I don't even know if the post was
readable, not to mention worth writing.

I'd be happy to
spend year or two with a selection of machines and stock, making
duplicates, or even upgrades.

In theory some pairs of machines and one foundry are all that is required. But
to replicate the jig that holds the stock that end up as a machine part, you
sometimes need a second jig. That's what I've discovered in theory, a sort of
parity law or exclusion principle. Goncz's Postulate: You Need Two of
Everything to Be Able to Make One of Anything.

Anyone with funding?

It has recently become clear that the best funding is available from a machine
tool manufacturer. Once their brand is selected as representing an adequate
range of machines and we select a mill model, lathe model, etc,, we write an
order and ask that it be delivered for research free of charge.

No DARPA funding is required, and no university affiliation for anyone but me,
as this is one of three B.S. projects, either a thesis.

And yes, each pair can be in a different location and may or may not include a
foundry. Ideally, all pairs and a foundry, a stock supply, and a coffe pot in
several thousand square feet of comfortable indoor shop, heated, cooled, and
always dry.

Having each pair in a different location focuses, perhaps too closely, on
in-class reproduction tasks, which could be useful, while all together, there
are more jobs that can be done per day, selecting each from whatever looks
achievable, always knowing that two of everything is / are available, so that
no matter what is picked, work can proceed without delay.

With this, in contrast to a deadline job, we start with the easy stuff first.
Or the lightest stuff in the mass based model.





Yours,

Doug Goncz (at aol dot com)
Replikon Research

Read the RIAA Clean Slate Program Affidavit and Description at
http://www.riaa.org/
I will be signing an amended Affidavit soon.

On 03 Oct 2003 11:16:44 GMT, ( Doug Goncz )
pixelated:

That's what I've discovered in theory, a sort of
parity law or exclusion principle. Goncz's Postulate: You Need Two of
Everything to Be Able to Make One of Anything.

Ain't dat de trufe?


Anyone with funding?

It has recently become clear that the best funding is available from a machine
tool manufacturer. Once their brand is selected as representing an adequate
range of machines and we select a mill model, lathe model, etc,, we write an
order and ask that it be delivered for research free of charge.

Yes! Grants are good.


And yes, each pair can be in a different location and may or may not include a
foundry. Ideally, all pairs and a foundry, a stock supply, and a coffe pot in
several thousand square feet of comfortable indoor shop, heated, cooled, and
always dry.

You forgot "self-cleaning", Doug. Or do you have foundry-bots
scouring the floors for kibbles and swarf, each one eating
only its type of metal kibble? (preventing unauthorized alloys)


With this, in contrast to a deadline job, we start with the easy stuff first.
Or the lightest stuff in the mass based model.

I like it! Then you can see that you've actually made some
progress, with racks of parts piling up as you go. It's a
self-fulfilling impetus.

From: ( Doug Goncz ) ( Me )



I've discovered in theory, a sort of
parity law or exclusion principle. Goncz's Postulate: You Need Two of
Everything to Be Able to Make One of Anything.


Um, that's more like you need to have two of everything in your shop to make
one of anything in your shop, and it's a good policy for making one of anything
outside your shop.

It's like a set: If A1 is in the set and in use as a model, another A2 is
needed to make A3.

Of course, many machine shops receive as customers inventors with one of what
they've built and who want one or a hundred more. These are not just parts but
assemblies. The blueprints at Aerophysics always represented what was produced,
and usually represented what was to be produced, but we built what was needed
and very little more, except for a few hundred meters of weld nugget (we should
have used rivets and sealants but the drawing said weld) and nobody went home
hurt except Francis that one day. We even got checks once in a while.

I had a ball there.



Yours,

Doug Goncz (at aol dot com)
Replikon Research

Read the RIAA Clean Slate Program Affidavit and Description at
http://www.riaa.org/
I will be signing an amended Affidavit soon.

Doug Goncz scribed in
:

oh yeah, what about combo machines like lathe/mills? still need 2?

also, I'm not sure exactly how many machines you're proposing to work
with to begin your replication. are you saying 'I have 2 lathes so I
can make another, despite not having any other machines'
or are you saying, 'given a range of machine tools
(grinder/drill/mill etc), and 2 lathes, I can make another lathe'

swarf, steam and wind

--
David Forsyth -:- the email address is real /"\
http://terrapin.ru.ac.za/~iwdf/welcome.html \ /
ASCII Ribbon campaign against HTML E-Mail - - - - - - - X
If you receive email saying "Send this to everyone you know," / \
PLEASE pretend you don't know me.

From: DejaVU
Newsgroups: rec.crafts.metalworkin

oh yeah, what about combo machines like lathe/mills? still need 2?

The theory is simplified by assuming two of everything, and one's toolbox
should be adequate to the task before any job, so two of each thing to be
replicated, all accessories, but not necessarily two of each operator's tool
unless it becomes appropriate to validate the universality by producing as
output one of each machinist's tool.

also, I'm not sure exactly how many machines you're proposing to work
with to begin your replication. are you saying 'I have 2 lathes so I
can make another, despite not having any other machines'

No, I'm saying that inherently each machine tool is capable of partical
self-replicating, a lathe capable of replicating all its turned parts, a mill
capable of replicating all its milled parts. But for complete replication of
any machine containing milled, lathed, and ground parts, a mill, lathe, and
grinder are needed, but may not be enough if there's for instance, even one
part made with a shaper.

or are you saying, 'given a range of machine tools
(grinder/drill/mill etc), and 2 lathes, I can make another lathe'


2 lathes, 2 mills, 2 cylindrical grinders, 2 surface grinders, and the ability
to add another pair of machines each time an unsupported part is found in any
machine in the system.

ASCII Ribbon campaign against HTML E-Mail - - - - - - -

HTML email can be turned off in some versions of AOL and I think the server is
the source of the box that says "This email contains a picture. Do you still
want to open it?" Some versions of AOL don't even understand HTML email, just
rich text.

I used to use 4, am using 6 now, and 9 is the latest system hogging crap. I'll
probably go back to 4. It was safer.



Yours,

Doug Goncz (at aol dot com)
Replikon Research

Read the RIAA Clean Slate Program Affidavit and Description at
http://www.riaa.org/
I will be signing an amended Affidavit soon.

Doug, thanks for responding. I thought my post got lost in the noise.

The final operation need not always be on the same class machine, but if,
working backwards from the finished part, even _one_ operation on _one_ part
requires a machine of the same class as the downed, part-removed machine, we
can reduce the problem to pairs but no farther.

I still don't accept your assertion that two of every tool is the minimum
requirement. I accept that two of some things establish a convenient
baseline for replication and repair (a broken lathe may not be able to
manufacture its own repair part). But "two of every tool" can not be a
universal minimum, especially when applied to tools like hex key sets and
collets.

...And this would be a pair of bench grinders, a pair of hex key sets,
pairs of all collets required, but not necessarily a full set, a pair of
collet closers, etc.

As an example, the still functional 1/2" collet does not participate in the
replication of another 1/2" collet to replace the damaged one. The same
situation exists with hex keys.

Another example that illustrates "fewer than two" being sufficient is folks
building the "Gingery lathe" - self generation through successive
approximation. The first spindle is used only until the second (improved)
spindle is made. As an example, you mention that at least one part of a
surface grinder is built using a surface grinder. If that part were to fail
on the solitary surface grinder and make the machine inoperative, one could
build several replacement parts using the lathe and mill. Install the first
crudely made part to restore limited operational capacity to the surface
grinder and grind the second replacement part. Now replace the first
generation crudely built part with the slightly improved part. Repeat this
successive approximation approach until the surface grinder is completely
restored.

If you have two lathes and both are in routine service, there is a possible
although improbable situation where both lathes could fail at the same time
through failure of identical lathe-built parts. In this situation, one would
need to rely on successive approximation or perhaps outside assistance. Two
of every tool is not a perfect guarantee by itself.

Bruce

snip

Doug, thanks for responding. I thought my post got lost in the noise.

Sorry, the thread went off my list at 43, total 48 in Google. I recommended the
thread, including your final post, to sci.physics.research and there was at
least one reader who read and enjoyed the whole thread here, an replied in spr.

The final operation need not always be on the same class machine,

can reduce the problem to pairs but no farther.


It just simplifies the analysis to assume two of everything is available. It's
not an absolute requirement for implementation, although it looks like one of
the very first tasks will be to make a single, say M6x1x20 socket head cap
screw, which, properly equipped, shouldn't take more than a bloody WEEK! It's
not easy.

I still don't accept your assertion that two of every tool is the minimum
requirement.

I only meant to reply that one of everything isn't enough and two of everything
is enough, not that two of everything is required.

I accept that two of some things establish a convenient
baseline for replication and repair (a broken lathe may not be able to
manufacture its own repair part).

Yes, and for the simplest part copying process one machine goes down and one
machine goes to work. They may be different-class for any one and likely most
tasks, but E a task | a pair of identical machine tools are required.

But "two of every tool" can not be a
universal minimum, especially when applied to tools like hex key sets and
collets.

Bending hex stock is a machine shop operation. I can conceive of a situation in
which one might hold a collet closer WITH a collet closer symmetrically, but
yes, you don't use a hex key to bend a hex key and you can't hold a collet
blank with the same range collet unless there's a "splicer" or something like
that.

The order presented to the manufacturer asserts Goncz's Postulate. Reducing the
order when rejected and submitting a revised order is a reasonable compromise.
But implementation is to prove Goncz's Postulate and identify all the items not
required in dupe. I say, get two of everything, open everything as needed, and
when finished, return unopened what may be a lot of stuff, generating an
investment cost figure and rather than an overrun, a list and a return credit
figure, the financial landmarks for the first such project. Simplify!

I also assert that one shouldn't work a problem in one's head. So editing a
dupe order down could concievable cause an error, and thus, an error chasing
project, which is never a good thing.

Funding is effectively unlimited. If they're sending us two of each machine
tool at enormous substantial risk and expense, why not get two of every hand
tool and return one of each non-dupe-required, and two of each
oops-we-didn't-even-need-it? Tools have a long shelf life. Receipts get filed,
and are pulled up easily.

As an example, the still functional 1/2" collet does not participate in the
replication of another 1/2" collet to replace the damaged one.

Yes, this particular possibility is slim.
The same
situation exists with hex keys.

Uh, no. One can slip a bit of stock with a through hole, or some tube, over the
long arm section of some hex stock, and use its buddy as a lever to turn the
stock into a key. This bit of tube can be much smaller than the hey key if it
fits well, and such a selection of adapters can be made on a lathe and a boxed
set presented to the manufacturer on completion.

Yes, of course you could use a long bit of stock with a through hole as a tommy
bar handle, but the whole point is to turn each small problem back on itself in
the way most consistent with the self-replication that is the main focus.

Eight tommy bars in scrap, or a boxed set of eight hex key self-reproduction
adapters? The entire array is supposed to be have universality, that is, it can
make anything in its range, and such a solution becomes a new salable product,
rather than a collection of scrap bits. Or maybe that is the way to go.

Drill a bit of stock through, bend the smallest key stock into a key, enlarge
the hole, and complete the set, ending with one piece of used-up scrap and the
need to write down something that the English language isn't particularly good
for: a process specification or flowchart. Doing that takes us way back to
1956.

Von Neumann showed us that a universal constructor and a unversal computer,
plus a certain kind of program, are capable of reproduction, and provided a
specific design that was never built.

I'm talking about building an archive, a museum display that the curators can
figure out, rather than a pile of paper or a computer disk. When it's over,
there's an interview to identify each bit of evidence.

Another example that illustrates "fewer than two" being sufficient is folks
building the "Gingery lathe" - self generation through successive
approximation.

Yes, I am aware of this technique but it doesn't make a pretty display.
Self-upgrade was one of the first properties I identified in my 1995
experiments. I'm not the first one to notice it.

As an example, you mention that at least one part of a
surface grinder is built using a surface grinder. If that part were to fail
on the solitary surface grinder and make the machine inoperative, one could
build several replacement parts using the lathe and mill.

Yes, one could.

Install the first
crudely made part to restore limited operational capacity to the surface
grinder and grind the second replacement part.

Yes. Takes longer. Don't be cheap. Always request full funding. It actually
increases the chance of acceptance and thereby success, rather than making the
funding source think "This looks incomplete". Ginger has already done it an
left a book behind for us to enjoy.

I'm just doing it in some other way, possibily the only other way. Are you
saying the group should complete Gingery's series before attempting a
reproduction of existing machines? Not a bad idea at all. Proof of capacity to
complete the new project. Not bad at all.

Now replace the first
generation crudely built part with the slightly improved part. Repeat this
successive approximation approach until the surface grinder is completely
restored.

Yes, but we have two grinders, so we mill the ways, grind them, and add them to
grinder #3, and check off one task for that day.

If you have two lathes and both are in routine service, there is a possible
although improbable situation where both lathes could fail at the same time
through failure of identical lathe-built parts.

As long as we admit this remote possibility, we are ready to get on with it. I
originally saw several pairs of machines and a single non-multi-tasking
operator. Probabilty zero in that case.

In this situation, one would
need to rely on successive approximation or perhaps outside assistance.

Verboten! Heretic! Off with his head!

Successive approximation is fine. Subcontracting ruins the whole thing.

Two
of every tool is not a perfect guarantee by itself.

It's enough, though. Maybe more than enough.



Yours,

Doug Goncz (at aol dot com)
Replikon Research

Read the RIAA Clean Slate Program Affidavit and Description at
http://www.riaa.org/
I will be signing an amended Affidavit soon.

On 03 Oct 2003 11:56:53 GMT, ( Doug Goncz ) wrote:
Yes, and for the simplest part copying process one machine goes down and one
machine goes to work. They may be different-class for any one and likely most
tasks, but E a task | a pair of identical machine tools are required.

You're implicitly assuming that to replicate a part you need an identical
part on hand as a template. But that's not a requirement. All you really
need is a print.

For your postulated Mars machine shop, the mass of prints (most likely
CAD files) is a lot less than the mass of unnecessarily duplicated machines.
Payload mass fraction is *the* critical limit on space flight.

Seems to me you should be trying to discover the fewest number of
machines necessary to replicate a machine shop, given full sets of
plans for all the machines to be produced. I don't think you'd find it
necessary to have identical pairs of any machine tool in order to do
that.

Gary

From: Gary Coffman
Newsgroups: rec.crafts.metalworking

You're implicitly assuming that to replicate a part you need an identical
part on hand as a template. But that's not a requirement. All you really
need is a print.

Of course all you need is a non-self-replicating print, and a
non-self-replicating filing cabinet to keep it in. Do you want to use the lathe
to build a paper mill so you can make the print, and the mill to build a sheet
metal shear so you can file the print, or do you wan't to simplify, simplify,
without cheating.

The only self-replicating thing about a file cabinet full of prints is mildew.

For your postulated Mars machine shop, the mass of prints (most likely
CAD files) is a lot less than the mass of unnecessarily duplicated machines.
Payload mass fraction is *the* critical limit on space flight.


I understand this but explorers are not computers. It takes five minutes to
learn how to repair a bent feed screw and a lifetime to get it really right. It
takes five months just to familarize yourself with the CAD system, the
directory tree, and the contents.

Take out a working or broken part. Clamp it in the obvious way. See what's
needed for the final cut that gives you what's in the chuck right now. Work
backwards. Get stock. Perform first through final operations. Install repaired
part or get on with building new machine.

?

Seems to me you should be trying to discover the fewest number of
machines necessary to replicate a machine shop, given full sets of
plans for all the machines to be produced. I don't think you'd find it
necessary to have identical pairs of any machine tool in order to do
that.


That's a good problem, but I already gave the specification for a universal
machine shop: one that can make useful products, anything you'd like. Except
usually this universality does not include the ability to make each part of
everything in the shop, as one cannot sell such an unauthorized repair part at
full price. So by extending the definition of universality to uneconomic
production, we find among other things that this system is sefl-replicating.

When you're on Mars, you don't hire someone to dig a hole in the ground so you
can take a poop. You cover your own tracks. There's no money planned for Mars,
is there? This is an economy in which productivity is of survival value and
explorers are highly productive people who can surve nearly anything.. Who
cares who pays for what?

You're going to charge your trail buddy hospital rates for fixing his arm when
you're the hospital and he can't make it to the next one without your help? And
then bill him and work out a payment plan? A simple Thank You is enough for me.
Then we move on.

We have about 168 currencies. Even four is too many. One, two, or ideally three
is actually the definition of currency. A traingle is stable. A frame
collapses. So does anything larger.

Now that money isn't backed by gold, we can go to three world currencies any
time. Rather than business, government, and consumer money, we'll probably end
up with Northern Money, Southern Money, Eastern and Western Money, which is
already still too many. We tried for one and ended up with 168. I say stop
trying for one. Try for two, and get 25, each of whom has a one or the other
opinion. Try for three, and get three. Everyone in the industry knows why
there's no point in trying for four. It's not as unstable as 168, but they do
know it isn't stable. Three or fewer is stable.



Yours,

Doug Goncz (at aol dot com)
Replikon Research

Read the RIAA Clean Slate Program Affidavit and Description at
http://www.riaa.org/
I will be signing an amended Affidavit soon.

On 05 Oct 2003 09:18:30 GMT, ( Doug Goncz ) wrote:
From: Gary Coffman
Newsgroups: rec.crafts.metalworking

You're implicitly assuming that to replicate a part you need an identical
part on hand as a template. But that's not a requirement. All you really
need is a print.

Of course all you need is a non-self-replicating print, and a
non-self-replicating filing cabinet to keep it in. Do you want to use the lathe
to build a paper mill so you can make the print, and the mill to build a sheet
metal shear so you can file the print, or do you wan't to simplify, simplify,
without cheating.

A print is no more non-self-replicating than a machine tool. Both need
human intervention and stock in order to be replicated. At least the print
only needs a pencil and tracing paper, while your scheme requires a
whole other machine tool being used as nothing more than a template.
The mass penalty difference should be obvious.

The only self-replicating thing about a file cabinet full of prints is mildew.

The only self-replicating thing about a machine shop full of tools is rust.

For your postulated Mars machine shop, the mass of prints (most likely
CAD files) is a lot less than the mass of unnecessarily duplicated machines.
Payload mass fraction is *the* critical limit on space flight.


I understand this but explorers are not computers. It takes five minutes to
learn how to repair a bent feed screw and a lifetime to get it really right. It
takes five months just to familarize yourself with the CAD system, the
directory tree, and the contents.

Huh? Computers were originally people, with pencils and paper. Later,
operations were mechanized, and finally machines that could be programmed
were built. But it is all the same continuum. Any literate and numerate person
can be a computer, or understand and operate a mechanized computer.

It might take an especially slow learner five months to familiarize himself
with a CAD system's directory tree and contents. But a moderately intelligent
person should be able to do so in far less time, certainly less than the lifetime
you're assuming to gain competence making parts from templates with a
machine tool. One would hope that explorers would be of at least moderate
intelligence.

Gary

( Doug Goncz ) wrote:

From: Gary Coffman
Newsgroups: rec.crafts.metalworking

You're implicitly assuming that to replicate a part you need an identical
part on hand as a template. But that's not a requirement. All you really
need is a print.

Of course all you need is a non-self-replicating print, and a
non-self-replicating filing cabinet to keep it in. Do you want to use the lathe
to build a paper mill so you can make the print, and the mill to build a sheet
metal shear so you can file the print, or do you wan't to simplify, simplify,
without cheating.

The only self-replicating thing about a file cabinet full of prints is mildew.

The only thing needed to make it self-replicating under your
definition is a pantograph, which itself can be replicated using the
other machine tools. This assumes that the stock (in this case, paper
and ink) is readily available, which seems to be a general assumption
that you're already making. Of course, if you prefer you could do away
with the paper and ink and have the plans scribed on thin sheet stock.

For that matter, you could do away with the need to replicate the
plans by including the plans as part of the agent package rather than
part of the machine package. Your self-replicating machines cannot
replicate except through the actions of an external agent, and there
is apparently no requirement for that agent and its embodied knowledge
to be replicable; thus, you can significantly impact the difficulty of
the problem by your choice of how to distribute the information, work,
etc., between the machine and the agent, a choice which is essentially
arbitrary.

Consider, for example, that when you pull a part off a machine, you
have no way of knowing what the original dimensions of the part were
(unless the machine has never been used), what tolerances are
allowable, how much torque should be applied to threaded fasteners, or
even what the part's material specifications are (i.e., alloy
composition, heat treat, required hardness and strength, etc.), unless
you have a large suite of (non-destructive) testing equipment. So your
agent must embody that knowledge in one form or another (along with
knowledge of how to use the machines, how to assemble the replicated
parts, etc.), unless your definition of "replica" is "something
kinda-sorta similar" or "something that looks the same and may or may
not provide similar functional performance." And if your agent already
embodies that knowledge, then it doesn't take much of a shift in the
agent/machine boundary to include the complete part specifications
with the agent, so replication of the plans becomes unnecessary.

Bert