Comments on: IKEA, your days are numbered

IKEA, your days are numbered

greatgefilte — Fri, 18 Mar 2005 11:39:46 -0800

Want it? Make it! 3D printers aren't that new -- already there are robots that print houses, inkjet printers that print human tissue, and for you CSI fans, machines that can reconstruct bullets, among other things. What's new, you ask? Machines that can produce anything and self-replicate, too. All under a GNU General Public License.

By: Dr_Octavius

Dr_Octavius — Fri, 18 Mar 2005 11:53:12 -0800

I believe you speak of a Von Neumann Machine.

By: StickyCarpet

StickyCarpet — Fri, 18 Mar 2005 12:02:05 -0800

I don't know why, but stories like this always remind me of the time I ate a tongue sandwich.

By: karuna

karuna — Fri, 18 Mar 2005 12:04:05 -0800

I'm very skeptical. The part of the machine that holds the molten plastic (or metal or whatever) cannot be made of the same material. You'd need something with a higher melting point. The linked project-site was pretty short on details. A truly self-replicating machine would have to be built at the nano-scale (Drexler's assemblers).

By: hellphish

hellphish — Fri, 18 Mar 2005 12:12:51 -0800

Not true, karuna. Many rapid prototyping machines use a material that hardens only under exposure to a laser.

By: solotoro

solotoro — Fri, 18 Mar 2005 12:30:14 -0800

"The machines would not be able to produce glass items...or objects that would work under intense heat..." Looks like lasers are out, too. But diode lasers themselves have gotten pretty cheap...so as long as it relies on a relatively weak laser (both in intensity and in wavelength), then it could still lead to cheaper technology. It wouldn't be quite entirely self-replicating, but that sidesteps karuna's argument, so maybe that's not a bad thing.

By: ScotchLynx

ScotchLynx — Fri, 18 Mar 2005 13:01:37 -0800

A true "Von Neumann Machine" may be a ways away, but the 3D printers that the rest of the post talks about are fascinating, especially the biological one. My question is where they get the biological mass.

By: BlackLeotardFront

BlackLeotardFront — Fri, 18 Mar 2005 13:21:56 -0800

Ab fab!! Someone had to say it.

By: gurple

gurple — Fri, 18 Mar 2005 13:23:08 -0800

I'm holding off until they come out with that machine from _The Fifth Element_ that can assemble me a Milla Jovovich in 5 minutes.

By: sninky-chan

sninky-chan — Fri, 18 Mar 2005 13:34:48 -0800

As for the laser or the high-temp box, those things can be added later (as the article mentions). It would still keep the cost way down to have machine #1 make 99% of the parts for machine #2, even if you have to buy a laser later. And... what gurple said. *Multiple* Milla Jovovich... um, what's the plural of Milla Jovovich?

By: ROU_Xenophobe

ROU_Xenophobe — Fri, 18 Mar 2005 13:44:11 -0800

what's the plural of Milla Jovovich? Jovovichim. The part of the machine that holds the molten plastic (or metal or whatever) cannot be made of the same material. Sure it could be... You'd need something with a higher melting point. Or you could just cool it with water channels and a water pump. Lots of things have combustion tempertures above their melting points.

By: effugas

effugas — Fri, 18 Mar 2005 13:45:49 -0800

Millas Jovovich. A little like Attourney's General, or the Brothers Karamazov.

By: gurple

gurple — Fri, 18 Mar 2005 14:03:44 -0800

The idea of having each machine make copies of itself is super cool, but is it useful? Presumably a special-purpose machine designed _just_ to make these things would do a much more efficient job of it. You know, what factories do. As a gambit to try to force the price of the technology down by creating homegrown competition, snazzy. As a practical way to make the machines, it rings false to me.

By: Robot Johnny

Robot Johnny — Fri, 18 Mar 2005 14:09:01 -0800

A little like Attourney's General, or the Brothers Karamazov. Or my personal favourite: sons of bitches.

By: Foosnark

Foosnark — Fri, 18 Mar 2005 15:21:42 -0800

With a catchy name like "RepRap" it's sure to succeed!

By: orthogonality

orthogonality — Fri, 18 Mar 2005 15:37:36 -0800

gurple writes " The idea of having each machine make copies of itself is super cool, but is it useful? Presumably a special-purpose machine designed _just_ to make these things would do a much more efficient job of it. You know, what factories do.

"As a gambit to try to force the price of the technology down by creating homegrown competition, snazzy. As a practical way to make the machines, it rings false to me." Tell it to the bacteria. Really, it's simple exponentiation versus multiplication. Let's assume, as you do, that a dedicated factory would be more efficient. Let's go so far as to say that a dedicated factory would be one hundred times more efficient -- that is, in a given duration, the factory could make one hundred machines, and the self-replicator would make only one. We'll call a that duration a "day": in one day, a factory makes 100 (non-self-replicating) machines, and a self-replicating machine makes a single additional self-replicating machine.

Day	Factory Self-replication
        Cumulative
1	100		1
2	200		2
3	300		4
4	400		8
5	500		16
6	600		32
7	700		64
8	800		128
9	900		256
10	1000		512
11	1100		1024
12	1200		2048

By day 11, the factory has produced 1100 machines, the self-replicators "a-doublin and doublin'" to quote Pete Seeger, have produced 1024. On the next day, the self-replicators in a single day, produce nearly as many machines as the factory has produced the entire previous eleven days. By day 17, the self-replicators have produced 20.48 times the machines produced by the factory to date. By the end of the month, the self-replicators have produced over machines totalling 178956 times the monthly production of the factory. And recall, we started with only one self-replicating machine. Now let's assume that you had an incredibly efficient factory that produced a million machines in the time a self-replicator produces one. Even in that case, by day 26, the factory would have produced 26 million machines, but the self-replicators over 33 million. What's the real lesson here? It's not that you're stupid or short-sighted. It's that these things ring false because they are not part of the human experience that evolution adapted us to deal with. Our brains are not designed to think in millions or about exponentiation or even about factories or statistics or laws of large numbers or even great spans of time., and it shows when we use our intuition to judge situations that evolution never sharped our intuition on.

By: Galvatron

Galvatron — Fri, 18 Mar 2005 15:52:16 -0800

Pessimist that I am, if this thing ever gets off the ground then I wonder whether the plastic and metal construction materials will suddenly skyrocket in price as manufacturers find themselves hopelessly underprepared for demand. But I hope not. Cheap rapid prototyping would be way cool. [this is good]

By: syzygy

syzygy — Fri, 18 Mar 2005 16:26:46 -0800

orthogonality: these things ring false because they are not part of the human experience that evolution adapted us to deal with. From first link: The machines would be about the size of a refrigerator, and would self-reproduce by making a copy of themselves, part by part. These parts would then have to be assembled manually by their owners. We are still talking about assembly by humans. Growth will be nowhere near exonential until the machines are able to completely replicate themselves, and humans are taken out of the production equation.

By: WolfDaddy

WolfDaddy — Fri, 18 Mar 2005 16:27:09 -0800

First chance I get I'm replicating, in no particular order of preference, Avery Brooks, Shaun Cassidy, Orlando Bloom (with black hair dammit), Kiefer Sutherland, Victor Garber and perhaps my childhood calico kitty.

By: orthogonality

orthogonality — Fri, 18 Mar 2005 17:03:16 -0800

syzygy writes "orthogonality: From first link: The machines would be about the size of a refrigerator, and would self-reproduce by making a copy of themselves, part by part. These parts would then have to be assembled manually by their owners.

"We are still talking about assembly by humans. Growth will be nowhere near exonential [sic] until the machines are able to completely replicate themselves, and humans are taken out of the production equation." Digger wasps.

By: greatgefilte

greatgefilte — Fri, 18 Mar 2005 17:11:57 -0800

Pessimist that I am, if this thing ever gets off the ground then I wonder whether the plastic and metal construction materials will suddenly skyrocket in price as manufacturers find themselves hopelessly underprepared for demand. No problem, we'll just get the replicators to crank out some of these babies.

By: sophist

sophist — Fri, 18 Mar 2005 22:38:43 -0800

Growth is still exponential. Just because humans are needed to put them together does not mean that the growth is linear, it is merely slower than if the machines could assemble themselves. I think the main difference in these machines and living things is that these do not exist primarily to replicate themselves. He is talking about making replicators from other replicators, but once every person in the world has a replicator, there will be no need for them to continue making them. He sees the machines existing primarily as small maufacturing plants, creating household items such as plates, clocks, cameras, lamps, etc.

By: syzygy

syzygy — Sat, 19 Mar 2005 02:40:05 -0800

orthogonality: Digger wasps And?

By: syzygy

syzygy — Sat, 19 Mar 2005 02:55:35 -0800

sophist: Growth is still exponential. Just because humans are needed to put them together does not mean that the growth is linear Wrong. The only growth that is exponential is the growth in manufacture of the parts needed to build the machines. Unless you find a process of assembling the machines which can handle an exponentially growing number of inputs (machine parts), you don't have exponential growth in the number of useful machines manufactured, you have only exponential growth of the (unusable- until- assembled- by- processes- that- are- most- likely- not- equipped- to- handle- exponential- growth- of- inputs) individual parts of the machine. Human assembly speed can only grow linearly, and if any necessary part of the production process can only advance linearly, the whole process is restricted by that weak link. Call me back when the machine can actually replicate itself, not when it can almost replicate itself. Almost self-replication isn't all that interesting. Full self-replication is.

By: sonofsamiam

sonofsamiam — Sat, 19 Mar 2005 05:19:59 -0800

To you. Besides, the point is to make all the custom parts yourself, and use commodity parts for everything you can't. Still a big improvement, if you ask me.

By: hattifattener

hattifattener — Sat, 19 Mar 2005 18:57:33 -0800

Uh, syzygy, and everyone, you are maybe forgetting that humans also increase in number exponentially? In practice, what exponential growth usually means is that growth rate isn't what limits the number of copies there are of an object. It'll hit some other resource limit. The most obvious resource limit is that any given human being only wants a certain number of these things, and so will only take the time to assemble that number of them. It depends on what the difficulty of assembly is, of course, and whether any of the raw materials or non-self-produced parts are expensive, or whether the machines are outlawed by the ?IAA.