RepRap and Exponential Growth

10 views
Skip to first unread message

Erik

unread,
Jan 6, 2009, 3:10:32 AM1/6/09
to Open Manufacturing
Below is an entry from my blog. But first I'd wish to share some
background information.

I'm a student, entrepreneur and builder & promoter of RepRap. The
RepRap project aims to create a 3D printer that can create objects of
arbitrary shapes through a layered fabrication process. This
particular printer is capable of replicating most of its own parts and
is developed worldwide through the open source process. The
replication and open source aspects serve as enablers for a unique
form of distributed innovation that was previously only seen in
software and (virtual) open content. RepRap has the potential to give
the power to manifest ideas into tangible objects to anyone with a
creative mind (or just an internet connection). There is virtually no
cost in sharing the digital models of physical objects that are a
solution to a problem.

Convergence of the virtual and physical world, that's the
revolutionary change that RepRap is part of.

My activities are: giving lectures, presentations and hands-on
workshops around personal fabrication. Furthermore I develop and
optimizing software for the RepRap (mostly firmware and driving
software, some electronics as well).

Here's the article:

A thousand RepRappers!

<http://blog.erikdebruijn.nl/archives/76-A-thousand-RepRappers
%21.html>

There are now an estimated 1000+ RepRappers
<http://www.rrrf.org/2008/12/24/rrrf-meeting-minutes-for-dec-19th-2008/
>!
These people are actively building on their own instance of the Open
Source Replicating 3D printer <http://www.reprap.org>.

When I started working on my RepRap in May 2008 (8 months ago). At
that
time there were about 20 people actively involved in the project and
only a handful of people had a /working/ machine. Because I had a lot
of
spare time, about a month later, I was also part of this group. When I
had the proof of concept build and made some 3D prints, I started
changing the machine's electronics. Then, I had vowed to finish my
bachelors degree first, so I had considerably less time for the
RepRap,
except for giving some presentations
<http://www.wikimediaconferentie.nl/hidden/hidden_programme/reprap/>
and
workshops
<http://blog.erikdebruijn.nl/archives/67-Building-RepRaps-in-Utrecht,-
NL%21.html>.
Meanwhile, the rapid pace of growth has not stopped!

I have always given a factor two exponential growth as an example. I
used to say: for each person with a RepRap that enables another person
to make a RepRap by printing the plastic parts for him, you will see a
doubling of the amount of RepRaps. But now it seems like a 50-fold
increase in three quarters of a year. If you assume the growth is
purely
exponential, this would mean that it's an annual growth of 50^(4/3) =
184.2. This is an estimated 18420% increase of RepRappers in a single
year! Of course, the amount of people with a working machine will lag
behind that. It will range from 1 month, probably up to a year. But
even
if 'lead times' were an entire year on average, the amount of
RepRappers
will still be staggering in a couple of years!

If we take a conservative amount of growth, e.g. 1000% growth annually
and one year lead time, there will be a million people with a
production
RepRap within 4 years! This is with the conservative estimate based on
an exponential effect. With the 18420% growth (which is also
unbelievable to me, I must admit), this is 1.3 years before there are
1
million RepRappers. My gut feeling says that an annual growth factor
of
20 is realistic. In this case it would take less than 2 years and 4
months before there are a million RepRappers, while only 6 months
later,
there will be two million (LOG(10,000,000/1000)/LOG(20)=in 3.07
years)!

It is always hard to predict a trend that is in its early development.
Yet the system dynamics <http://en.wikipedia.org/wiki/System_dynamics>
of this particular case are a strong case for exponential growth. The
drivers behind this growth are (to name a few):
R1. Word of mouth, determined by amount of RepRappers
R2. Evolution of the design, determined by the amount of skillful
people
attracted to develop and improve the RepRap
R3. More people with a RepRap, more readily printable objects on the
internet.
R4. More RepRappers, more infrastructure (sites like thingiverse),
sponsors, more buying power of RepRap stores
R5. More RepRappers, more forks (mobile RepRap, large-scale RepRaps)
and
extensions of the concept RepRap to an entire replicating lab (RepLab)
<http://www.replab.org/>. More forks also means that there will be
more
paths towards a RepRap (a.k.a. RepStraps). I've seen RepStrap parts
cut
from steel, acrylic, plywood. I've seen them built from meccano, lego
and fishertechnik. There are now laser-cut acrylic RepRaps, PlyRap's,
LegoRap's, RepScrap's, Fishertechnik RepStraps, and many more to come.
Also, the RepRap is able to do light milling work, extrude metals,
various pastes, chocolate, etc. Also, rudimentary scanning of 3D
objects
has been done!
R6. Maturization of documentation, larger group giving support to new
users.
R7. Open source is extending more and more into the hardware domain.
This
allows for interesting things to come, since it the RepRap can provide
the body of all sorts of machines, while Arduino's and other open
source
electronics provide the brain. This will provide a nice symbiosis.

R1 to R7 are reinforcing loops.

I argue that besides just Metcalfe's law that describes network
effects
<http://en.wikipedia.org/wiki/Metcalfe%27s_law>, there are many more
drivers behind this growth, but it is also much more complex. But most
importantly, the strength of many drivers are linked to the scale of
the
RepRap user-base. This means that the rate of growth is determined by
the scale of the user-base, which is increasing. There are of course
diseconomies of scale, market saturation points, etc. Diseconomies,
such
as that the design becomes more complex (in order to produce better
quality), fewer people can still contribute to the design. I think the
core technology will become privileged to a group of specialists, but
there is so much more to a project than its core technology. People
can
do a variety of things to help such a project prosper. And each person
himself knows what he is best at (far better than any HRM department).
About market saturation: I argue that as the machine evolves to become
easier to assemble and still more powerful, new and wider markets will
start adopting this innovation. Also, if there's more infrastructure
supporting the growth of RepRap, it will become easier to acquire
production-ready sub-assemblies and eventually entire machines.

Each of these effects can be seen as positive feedback loops (also
know
as 'snowball effect'). Besides that there are some trends that are
exogenous but still are drivers growing in power:
1. Growing popularity and Open Source and Creative Commons are doing
well and becoming better-known!
2. DIY trend (instructables, Make Magazine very popular)
3. Long tail: Demand for mass customized products. More and more means
to reach the long tail market with your own objects.
4. A need and desire for sustainable living. RepRap provides a big
opportunity there (linked article in Dutch)
<http://blog.erikdebruijn.nl/archives/59-RepRap-NL-Energie-en-
grondstoffenefficintie.html>.

I wonder where we will be next year. I'll surely make another post
about
it around 1 january 2010!

--

Please, cross post your comments to my blog as well, the discussion
might be interesting for my readers as well. Thanks!

Kind regards,

Erik de Bruijn

marc fawzi

unread,
Jan 6, 2009, 3:57:50 AM1/6/09
to openmanu...@googlegroups.com
Hey Erik

I was talking about RepRap a few hours ago with the guys on Bryan's IRC channel

Bryan's first question was if I have an "ulterior motive" (read: commercial purpose) in being interested in this stuff but I don't think the technology is ready for commercial use even after 20 years of being out there. It's still for prototyping not commercial production, although I'm sure the high end machines ($100K and up) are designed for industrial prototyping so they maybe a lot faster and may produce much higher quality parts.

First time I laid my eyes on a rapid prototyper was at MIT in I think 1990 or 91 (was watching friends run their experiments), and it used two UV lasers and a UV curable monomer. It cost a load of money (at least the lasers did) and it couldn't produce anything better than what RepRap can. 

Then I got into an argument with one of the guys on Bryan's channel about whether or not you can have scaled economics with decentralized peer production, to which I argued that you can if the production infrastructure was consistent across the network [and, now that I think of it, if such decentralized production network would also act as the product distribution network, which could lead to better efficiencies than centralization.]

My dream is to be able to build a machine that can build cinder-block-sized Lego pieces on demand and reasonably quickly (from CAD description) that I can then use to build shade translucent (or even opaque/colored?) housing structures ...

Homemade rapid prototyping technology will hopefully enable that kind of project someday, whether on commercial scale or for limited personal use.

sm...@anarchism.is

unread,
Jan 6, 2009, 6:14:16 AM1/6/09
to openmanu...@googlegroups.com

I plan to try and have my students build two rep-raps this semester as part
of their coursework. The Rep-Rap, and FDM in general, isn't very good
technology, but it is very cheap technology and it packs a lot of bang for
the buck. I don't think the rep-rap or FDM is going to change the world of
manufacturing, but it has oodles of potential as a stepping stone, a
bootstrapping device, and an educational gimmick.

- Smári



On Tue, 6 Jan 2009 00:57:50 -0800, "marc fawzi" <marc....@gmail.com>
wrote:
%21.html<http://blog.erikdebruijn.nl/archives/76-A-thousand-RepRappers%21.html>
>> >
>>
>> There are now an estimated 1000+ RepRappers
>> <http://www.rrrf.org/2008/12/24/rrrf-meeting-minutes-for-dec-19th-2008/
>> >!
>> These people are actively building on their own instance of the Open
>> Source Replicating 3D printer <http://www.reprap.org>.
>>
>> When I started working on my RepRap in May 2008 (8 months ago). At
>> that
>> time there were about 20 people actively involved in the project and
>> only a handful of people had a /working/ machine. Because I had a lot
>> of
>> spare time, about a month later, I was also part of this group. When I
>> had the proof of concept build and made some 3D prints, I started
>> changing the machine's electronics. Then, I had vowed to finish my
>> bachelors degree first, so I had considerably less time for the
>> RepRap,
>> except for giving some presentations
>> <http://www.wikimediaconferentie.nl/hidden/hidden_programme/reprap/>
>> and
>> workshops
>> <http://blog.erikdebruijn.nl/archives/67-Building-RepRaps-in-Utrecht,-
>>
NL%21.html<http://blog.erikdebruijn.nl/archives/67-Building-RepRaps-in-Utrecht,-NL%21.html>
grondstoffenefficintie.html<http://blog.erikdebruijn.nl/archives/59-RepRap-NL-Energie-en-grondstoffenefficintie.html>

Erik

unread,
Jan 6, 2009, 10:11:12 AM1/6/09
to Open Manufacturing
It's important to realize that we're talking about a 500 euro machine
that can upgrade itself and replicate many of its own parts, versus a
commercial machine of 20.000 euro and up that is 'static' and prints
from expensive 'commercial cartridges'. The manufacturer's protects
its distribution channel, just like with 2D printers' ink.

Another effect is that due to the availability of a substitute (see
Porter's 5 forces model) this will force down prices of other rapid
prototyping technologies. This is not necessarily bad for the
commercial companies that are now exploiting similar technology,
because more people are creating 3D printable content (often copyleft)
AND more people people/companies will adopt these technologies.
There's a significant business trend towards mass customization. Using
rapid prototyping (in this case actually called Rapid Mfg., RM) for
some manufacturing steps or small-series parts will become more
common. This is also fueled by the shortening of product life cycles.

> My dream is to be able to build a machine that can build cinder-block-sized
> Lego pieces on demand and reasonably quickly (from CAD description) that I
> can then use to build shade translucent (or even opaque/colored?) housing
> structures ...

Google for 'contour crafting'. It's essentially 3D printing entire
houses. 3D printing can applied at different scales. For a RepRap it
is simple to extend its scale. But it currently has this size so it
conveniently sits on your desk.

> The Rep-Rap, and FDM in general, isn't very good
> technology, but it is very cheap technology and it packs a lot of bang for
> the buck.

Just for fun, replace the word RepRap with "Inkjet printer". This is
exactly what laser printer manufacturers thought: it is inferior
quality, customers will not buy it. Yet it made printing affordable
for everyone. Now almost everyone has a 2D printer at home, and a good
one too! I'm not saying 'superior printers' will be replaced, but home
fabricators will be a valuable addition, not to mention a tool of
personal empowerment. Compared to 3D, printing 2D (mostly information)
isn't that valuable when you can view it on a screen (or from eReaders
and e-Paper within a decade). Still we've spent a significant amount
on new models of such a 2D printer (since they have the nasty property
that you cannot upgrade them!).
FDM can still be very high-res in the X/Y plane. In the Z-plane it can
be a minimum of (0.3 mm or about 0.01 inch). I agree that it's not a
perfect finish, but it is suitable for many applications. Besides an
FDM process you could bootstrap 3D printers that utilize other
processes with superior surface finish.

> I don't think the rep-rap or FDM is going to change the world of
> manufacturing, but it has oodles of potential as a stepping stone, a
> bootstrapping device, and an educational gimmick.

What you're saying is somewhat contradictory. If it is a potentially
great stepping stone, then it's an enabler of something else. I think
that RepRap definitely enables innovation of other objects (machines)
that can be developed over the internet in a distributed (scalable)
way. If these machines together outstrip the importance of RepRap, it
is still the RepRap that has initiated and enabled this change. So I
would argue that a bootstrapping device has a major impact!
An educational gimmick could be very significant (and profitable).
There are many schools around the world. And many teachers I spoke
with (including yourself, it seems) are very interested, from primary
education to professional education (even a management school). A
school with a RepRap could extend its education greatly. Kids will
grow up in a world where 3D printing is common. It might be good to
prepare them for this. I imagine that businesses eventually want to
make RepRap kits for schools, or teachers that want to build it with
their class as a school project. Actually, there are several classes
currently building RepRaps, and many more planning to do this. Tom
Easton is doing a class on the impact of this disruptive technology.
FabLabs are interested. We're actually building 5 more machines with
group in a FabLab in the Netherlands. More people have signed up
already for a next sessions. At FabLab Utrecht they say that a single
RepRap could greatly extend the capabilities of the FabLab. Right now
they have a full colour 3D printer, but it cannot produce the strong,
usable parts that a RepRap can. And the full colour 3D printer cost
60.000 euros and is quite expensive to print with (per cm^3) compared
to RepRaps plastics.

I've heard more people say "RepRap cannot even produce all if its own
parts, it's not really a replicator, it' stupid. Boo..", or "It's a
nice gadget but cannot really do professional things".

"When the first “late beta” version of RepRap -the “replicating rapid-
prototyper"- is released in early 2008, critics have a field day. It’s
slow. It’s clumsy-looking. It can’t actually replicate itself without
adding a few key commercial parts. But where critics see an ugly
duckling, design students, DIY hackers, and open source enthusiasts
see a swan-in-the-making. By the summer, dozens of novel fabber
projects emerge (some forked from RepRap, but most based on original
designs), and by the fall, some have actually produced devices that an
adventurous home user could play with. Forward-looking strategists at
mega-retailers and mass manufacturers feel a distinct chill run up
their collective spine. The open fabber era had begun, and through the
end of the decade, free and open source software hackers around the
world turn their attention to hardware… By the time molecular
manufacturing applications do mature at the nanoscale, Openfabs are a
ubiquitous fact of global life. It’s not surprising, then, that the
first atomically-precise devices are designed with Openfab-standard
interconnects for integration into the existing open world standard
for human-scale production infrastructures”. "

Source: Scenario 2 of the Center for Responsible Nanotechnology
Working Group

Others say that it can only produce plastic parts, so it's limited.
Well, not that much. But we're working on metal. And you can print a
mould or a positive to cast with. A friend of mine cast an aluminium
part for his motorcycle that was made from a 3D printed positive, he's
now happily riding his motorcycle with that part! But even for
plastics objects alone there are more than enough applications for
such a printer to be valuable.

I think it will change manufacturing drastically. It lowers the
barrier for anyone to become a local producer. Just like the internet
lowered the barrier for individuals to become knowledge workers (Tip:
read the world is flat by T. Friedman). These can often compete with
remote producers (whore produce centralized and on a large scale)
because of:
+ speed (lead time is shorter, no more waiting shipping)
+ customization (it is Make/Engineer To Order vs. Make to Stock). They
can produce what you want, and not just 'what is available on the
market'.
+ efficiency: Make to Order implies that you don't have obsolete
stock. IF you have over-produces because you over-stocked in
anticipation of sales, you just recycle the objects (put them in a
blender and re-use it in you're RepRap). There are major ecological
benefits too.
+ choice: open object repositories / crowdsourcing and co-creation.
There are website's with 3D printable objects. One of them originated
from the RepRap project: http://www.thingiverse.com Just like
Wikipedia has a lot of knowledge for everyone to use, contributed by
the masses, these web-sites will grow with the commodification of 3D
printing. Besides this, commercial companies like Shapeways and Ponoko
increase the amount of content that is out there through their co-
creation strategy.
+ sustainability: using local raw-materials / resources (from
potatoes, corn, even banana's) you could cheaply make bio-degradable
plastics locally (from starch).

Plug-in compatibility is what makes your electronics cheap. It allows
a CAD/CAM designs to be used for production by Electronics/
manufacturing Service Providers like Flextronics/Solectron. They have
manufacturing facilities distributed around the world. This is all
possible because of digital designs. If there are more open source
digital designs, and enough people have 3D printers (e.g. at first,
one in each neighborhood), this will surely change manufacturing!
Perhaps the design can be done in lower-wage countries, and we will
manufacture over here. If this sounds silly and counter intuitive,
it's already happening:
"When you negotiate with the customer in Japan [a 'western'-country]
for building a house, you would sketch out a floor plan [..] Hand
drawn plans are sent electronically to China, where they are converted
into digital designs, which then are e-mailed back to the Japanese
building firm, which turns them into manufacturing blueprints."
By: Thomas L. Friedman (3 times Pulizer prize winner). "The world is
flat"

It illustrates that if it work can be digitized, the work will go
where it can be done best (depending on the criteria: the cheapest, to
a specialist, near you, etc). Remember that 3D printed objects require
hardly any physical labor, it just occupies a machine which isn't that
expensive. Hence, it can be routed to a home-manufacturer near you, so
you can pick it up on your way to the groceries store. If you're a
frequent buyer, why not get yourself a RepRap? You can save some, and
even earn back your investment by providing printing services. And now
you can even produce things on-demand when the shops are closed.

--
Erik
http://blog.erikdebruijn.nl/

"RepRap is taking over the world, one piece of plastic at a time" :)

Paul D. Fernhout

unread,
Jan 6, 2009, 12:16:16 PM1/6/09
to openmanu...@googlegroups.com
An instantiation of RepRap is an idea about abundance that people can hold
in their hands (like a Cornucopia, but one that really works). It helps
people move beyond a failure of the imagination to the point where they say,
"Oh yeah, I've seen one and it works (sort of)." That is its biggest value.
It is an idea (3D printing) made physical.

Of course, RepRap itself is also an idea about printing ideas, so it's
recursive. :-)

As you suggest, RepRap will get improved on and eventually obsoleted by
those people who use it -- and that is a good thing.

Sounds like your students are lucky to have you as an advisor. :-) Building
a RepRap does sound like a great engineering class project, and I hope other
educators eventually follow your lead.

--Paul Fernhout

marc fawzi

unread,
Jan 6, 2009, 3:22:24 PM1/6/09
to openmanu...@googlegroups.com
Erik,

Contour Crafting sounds a bit implausible in that the 3D plotter is humongous (I know it can be folded etc) and all you really need to build is the wall so why not just make the walls at a 3D factory and ship them to be put together onsite?

I like the idea of building Lego bricks with RepRap... but not sure how painfully slow it would be to generate a cinder-block-sized lego piece?

Where  do I need to go for information on how to build a RepRap?

Marc

Bryan Bishop

unread,
Jan 6, 2009, 4:01:52 PM1/6/09
to openmanu...@googlegroups.com, kan...@gmail.com
On Tue, Jan 6, 2009 at 9:11 AM, Erik <erikde...@gmail.com> wrote:
> I've heard more people say "RepRap cannot even produce all if its own
> parts, it's not really a replicator, it' stupid. Boo..", or "It's a
> nice gadget but cannot really do professional things".

Hi Erik, thanks for deciding to de-lurk and contribute to the
discussions here. There's a few points that I want to make, but
otherwise I think you're spot on. The replication issue with RepRap is
two-fold. First, there is no evidence that an ad hoc design process
can lead to self-replication. What some of us are doing on this list
is doing a computational analysis of the dependency tree of different
manufacturing processes so that we can computationally *validate* and
"prove" that something self-replicates, in the formal sense laid out
by Freitas in KSRM. Secondly, RepRap at its core is a rapid
prototyper, and meanwhile the reprap website and foundations claim
that it was originally meant to do self-replication .. even to the
point of telling the New York Times that the latest beta release can
"mostly self-replicate", which isn't really true, since that is
ignoring the entire issue of the dependencies that the design is
lacking. Anyway, some people think that replication is an *all or
nothing* issue, because if you can replicate 90% of your parts, but
the other 10% require 2000% relative effort, then what's really going
on here? Etc. To see some more on these discussions, check out:

http://groups.google.com/group/openmanufacturing/msg/e4c375acce772250

AASM, KSRM, Gingery, Freitas, etc.

> "When the first "late beta" version of RepRap -the "replicating rapid-
> prototyper"- is released in early 2008, critics have a field day. It's
> slow. It's clumsy-looking. It can't actually replicate itself without
> adding a few key commercial parts. But where critics see an ugly
> duckling, design students, DIY hackers, and open source enthusiasts
> see a swan-in-the-making. By the summer, dozens of novel fabber
> projects emerge (some forked from RepRap, but most based on original

No, it's not that it's an ugly duckling, it's a beautiful duckling,
but I don't see this as a "one swan takes all" thing going on here,
but rather the progression of all of the different works that those of
us in the DIY/maker/fabber community are contributing to. One of the
projects that we're pursuing on this list is the formal packaging of
open source hardware and manufacturing systems into some format, much
like a dot deb file, which makes up the 20,000 software packages for
debian; these dependency trees (check out debtree for dotty-generated
graphs) show how things depend on one another for it to all work.
That's what's going on here IMHO, not some super magical bullet. But
yes, it's quite awesome what's happening. :-)

> designs), and by the fall, some have actually produced devices that an
> adventurous home user could play with. Forward-looking strategists at
> mega-retailers and mass manufacturers feel a distinct chill run up
> their collective spine. The open fabber era had begun, and through the
> end of the decade, free and open source software hackers around the
> world turn their attention to hardware… By the time molecular
> manufacturing applications do mature at the nanoscale, Openfabs are a
> ubiquitous fact of global life. It's not surprising, then, that the
> first atomically-precise devices are designed with Openfab-standard
> interconnects for integration into the existing open world standard
> for human-scale production infrastructures". "

Heh. Molecular nanotechnology. I like to think about it too. Though I
don't particularly rely on it in my plans. You know, what if Drexler
and Freitas fail in their computational chemistry software that
they've been working on since 2007 at this point? Not that I'm
worried, they'll figure it out. But anyway. :-)

> Source: Scenario 2 of the Center for Responsible Nanotechnology
> Working Group

CRnano is where I picked up the term "super-empowered hopeful
individuals". Their blog is interesting, from time to time.

> Others say that it can only produce plastic parts, so it's limited.
> Well, not that much. But we're working on metal. And you can print a
> mould or a positive to cast with. A friend of mine cast an aluminium

Yep, there's work on metals, that's right, but that's also kind of
what worries me. It's the "oh, it doesn't do this yet, so let's patch
it with xyz", where xyz relies on some more tools that aren't really
incorporated into it; anyway, the software I've been working on will
help this out, so it's not like anybody needs to worry .. just needs
to be aware.

> + choice: open object repositories / crowdsourcing and co-creation.
> There are website's with 3D printable objects. One of them originated
> from the RepRap project: http://www.thingiverse.com Just like
> Wikipedia has a lot of knowledge for everyone to use, contributed by
> the masses, these web-sites will grow with the commodification of 3D
> printing. Besides this, commercial companies like Shapeways and Ponoko
> increase the amount of content that is out there through their co-
> creation strategy.

Yeah, thingiverse is a good example. There's also opencollector,
instructables, and so many others. Sam Rose (on this list) and some
others are interested in doing a website like that for the packaging
format that I've described to you above, so that's something to look
into. Of course, all designs will be open source and freely
downloadable, we'll probably want to set up a mirroring network or
something. Package management tools (like apt-get, etc.) will be made
as well. This way we don't have to write scrapers for thingiverse
(etc.) ;-).

- Bryan
http://heybryan.org/
1 512 203 0507

Erik

unread,
Jan 6, 2009, 5:29:07 PM1/6/09
to Open Manufacturing
> Hi Erik, thanks for deciding to de-lurk and contribute to the
> discussions here. There's a few points that I want to make, but
> otherwise I think you're spot on. The replication issue with RepRap is
> two-fold. First, there is no evidence that an ad hoc design process
> can lead to self-replication. What some of us are doing on this list
> is doing a computational analysis of the dependency tree of different
> manufacturing processes so that we can computationally *validate* and
> "prove" that something self-replicates, in the formal sense laid out
> by Freitas in KSRM.

I may come back to this, but it might even be an advantage to not
achieve FULL replication. The thought to still have a human in the
loop is something that sooths the mind. If the machine depends on a
human constructing it (which should be really), that's really fine. I
did most of the construction of my machine in a week. Debugging the
machine is something that took me quite long, but is exactly the time
which can be almost eliminated by good defaults, more reliable and a
more mature design.

Currently I'm more occupied with market trends, forces and social
transformation than with theory of self replication. I think current
models that describe diffusion of innovations can be helpful in
determining what might happen. I'm trying to model it, but while doing
these things and talking about it, I find that you can actually be in
the driver-seat exploring, instead of doing analyses or planning your
route to an unknown destination. An apt quote in this regard is:
"We can only see a short distance ahead, but we
can see plenty there that needs to be done."
A. Turing, 1950.

(Turing and von Neumann are still competing to be my favorite
scientist)

> Secondly, RepRap at its core is a rapid
> prototyper, and meanwhile the reprap website and foundations claim
> that it was originally meant to do self-replication .. even to the
> point of telling the New York Times that the latest beta release can
> "mostly self-replicate", which isn't really true, since that is
> ignoring the entire issue of the dependencies that the design is
> Anyway, some people think that replication is an *all or
> nothing* issue, because if you can replicate 90% of your parts, but
> the other 10% require 2000% relative effort, then what's really going
> on here? Etc. To see some more on these discussions, check out:
>
> http://groups.google.com/group/openmanufacturing/msg/e4c375acce772250
>
> AASM, KSRM, Gingery, Freitas, etc.

> No, it's not that it's an ugly duckling, it's a beautiful duckling,
> but I don't see this as a "one swan takes all" thing going on here,
> but rather the progression of all of the different works that those of
> us in the DIY/maker/fabber community are contributing to.

It is very hard to make predictions about RepRap specifically. It
could be a fork or another project that leapfrogs beyond it. It is a
good candidate and example of a technology that might completely
change the world. I think it is the best example if we had to make a
guess. But shold admit that I'm biased since I'm highly involved in
the project. I've got a lot of antenna's out, but it's hard to keep
things in perspective.

> One of the
> projects that we're pursuing on this list is the formal packaging of
> open source hardware and manufacturing systems into some format, much
> like a dot deb file, which makes up the 20,000 software packages for
> debian; these dependency trees (check out debtree for dotty-generated
> graphs) show how things depend on one another for it to all work.

That's exactly what I like about FOSS. A packaging system and a single
namespace software. However while debian can serve as good inspiration
and good practices, you shouldn't introduce legacy from the start. A
CLI might be useful for retrieving software (ideas) but 'things' are
more graphical.

> That's what's going on here IMHO, not some super magical bullet. But
> yes, it's quite awesome what's happening. :-)

I agree!

> Yep, there's work on metals, that's right, but that's also kind of
> what worries me. It's the "oh, it doesn't do this yet, so let's patch
> it with xyz", where xyz relies on some more tools that aren't really
> incorporated into it; anyway, the software I've been working on will
> help this out, so it's not like anybody needs to worry .. just needs
> to be aware.

It might be a patch at first, but it makes the entire platform more
versatile. Things that make it more suitable for other adding
toolheads will be added. This has spillovers for other manufacturing
processes to be developed without needing to develop an XYZ platform
first. A good thing about open source is that you can start focussing
on the core competence. If you use a layered approach, you can later
do further tailoring the XYZ platform to the application.

> something. Package management tools (like apt-get, etc.) will be made
> as well. This way we don't have to write scrapers for thingiverse
> (etc.) ;-).

If you need a dedicated server (a Xen guest instance), my company
could sponsor it. https://www.budgetdedicated.com/concepts

marc fawzi:
> Contour Crafting sounds a bit implausible in that the 3D plotter is
> humongous (I know it can be folded etc) and all you really need to build is
> the wall so why not just make the walls at a 3D factory and ship them to be
> put together onsite?

You have a point, using 'prefab' components is becoming standard
practice in construction work (where I live). I think this practice
will one day be replaced. I think contour crafting has a big plusses,
but there might be a hybrid situation that works best.

Paul D. Fernhout wrote:
> Sounds like your students are lucky to have you as an advisor. :-) Building
> a RepRap does sound like a great engineering class project, and I hope other
> educators eventually follow your lead.

Actually I'm a student myself. But I've been through the process of
building the machine, so that's valuable to them. I'm just making sure
I'm not the only one having fun with these machines, so the entire
fabbing scene can flourish and we can create more sustainable,
creative and fun world to live in.

Bryan Bishop

unread,
Jan 6, 2009, 6:15:50 PM1/6/09
to openmanu...@googlegroups.com, kan...@gmail.com
On Tue, Jan 6, 2009 at 4:29 PM, Erik wrote:
>> Hi Erik, thanks for deciding to de-lurk and contribute to the
>> discussions here. There's a few points that I want to make, but
>> otherwise I think you're spot on. The replication issue with RepRap is
>> two-fold. First, there is no evidence that an ad hoc design process
>> can lead to self-replication. What some of us are doing on this list
>> is doing a computational analysis of the dependency tree of different
>> manufacturing processes so that we can computationally *validate* and
>> "prove" that something self-replicates, in the formal sense laid out
>> by Freitas in KSRM.
>
> I may come back to this, but it might even be an advantage to not
> achieve FULL replication. The thought to still have a human in the

That's true, there might be some big regrets everyone has after
self-replicating hardware (other than the biological (for that, see
the DIYbio groups)) explodes. With a replication period of one day,
within a month you have 8 billion replication units, which is
estimated to be enough for everyone on the planet to have their own
personal fabricators. What would everyone do with so much technology?
Sometimes, on the transhuman discussion lists, people get all worked
up about it because they think it will be molecular nanotech, so only
companies and the government would have access to the technology [for
some reason], and thus it won't exponentially replicate with a usable
human interface, etc. Personally I think that might be a bit wrong- it
takes only one person to program the machines to fabricate themselves
some jets, wings, or tracks to go roll on over to give themselves away
to other people .. ;-)

Anyway, just a minor distraction in the discussion.

> loop is something that sooths the mind. If the machine depends on a
> human constructing it (which should be really), that's really fine. I
> did most of the construction of my machine in a week. Debugging the
> machine is something that took me quite long, but is exactly the time
> which can be almost eliminated by good defaults, more reliable and a
> more mature design.

Congratulations, by the way. :-)

> Currently I'm more occupied with market trends, forces and social
> transformation than with theory of self replication. I think current

Looking for business opportunities, I presume?

> models that describe diffusion of innovations can be helpful in
> determining what might happen. I'm trying to model it, but while doing
> these things and talking about it, I find that you can actually be in
> the driver-seat exploring, instead of doing analyses or planning your
> route to an unknown destination. An apt quote in this regard is:
> "We can only see a short distance ahead, but we
> can see plenty there that needs to be done."
> A. Turing, 1950.

You're trying to model the diffusion of information ? Have you
considered reading up on long tails and torrent trackers? I see that
you're already on the ball via your blog though-

http://blog.erikdebruijn.nl/index.php?/categories/7-RepRap

So nevermind :-). I also see that you own "replab.org", which is
basically the project that I've somewhat described to you in my
previous email. Maybe we can put the domain to good use soon enough.
:-)

> (Turing and von Neumann are still competing to be my favorite
> scientist)

von Neumann.

>> One of the
>> projects that we're pursuing on this list is the formal packaging of
>> open source hardware and manufacturing systems into some format, much
>> like a dot deb file, which makes up the 20,000 software packages for
>> debian; these dependency trees (check out debtree for dotty-generated
>> graphs) show how things depend on one another for it to all work.
>
> That's exactly what I like about FOSS. A packaging system and a single
> namespace software. However while debian can serve as good inspiration
> and good practices, you shouldn't introduce legacy from the start. A
> CLI might be useful for retrieving software (ideas) but 'things' are
> more graphical.

Could you elaborate more on this 'legacy'? Do you mean about making
stupid restricting decisions? Maybe something else ?

>> something. Package management tools (like apt-get, etc.) will be made
>> as well. This way we don't have to write scrapers for thingiverse
>> (etc.) ;-).
>
> If you need a dedicated server (a Xen guest instance), my company
> could sponsor it. https://www.budgetdedicated.com/concepts

"""
We actively encourage and support an open-source ecosystem. We support
businesses that create revenue around open-source software. We also
encourage adoption of open standards. We supply bandwidth and a server
for the official Wine repository, because we feel it is important to
be able to switch to Linux and still run the few win32 programs you
can't say goodbye to yet. We sponsor the RepRap project with bandwidth
and a server (seeding torrents and hosting ISOs). BudgetDedicated is
comitted to help making the RepRap project a success. We have also
built our own RepRap.
"""

Awesome, thanks for the offer. Also, kudos on running a wine mirror --
or is it the actual official release repository? Double kudos on
serving up the reprap project, could you drop any hints on the
webalizer stats for the server? Surely it's being pulvarized.

Would you be willing to do some commentary on the repositories that we
have passing around here, and the todo list etc.? Maybe we could do
this over IRC when all of the developers are awake to do things the
right way if we wanted to set up the server. I'm presently operating
as a giant SPOF (single point of failure) in some of the development
work because of my slow outgoing connections, etc.

Erik

unread,
Jan 7, 2009, 9:26:13 AM1/7/09
to Open Manufacturing

> human interface, etc. Personally I think that might be a bit wrong- it
> takes only one person to program the machines to fabricate themselves
> some jets, wings, or tracks to go roll on over to give themselves away
> to other people .. ;-)

Technological capabilities aren't evenly spread across countries and
social groups. The internet has done a lot to create a more level
playing field.
People had a limited view of information. It was seen as a scarce
resource and an asset that had to be protected. Now when people see
the emergent value of sharing information and learning in interaction
with others, this is starting to change. Now people in businesses
realize there's an abundance of information, which requires a
different way of looking at people/knowledge assets and how to make
money in a world that is opening up. Without changing this will result
in losses. While understanding the dynamics, everyone benefits and you
can still make money.

My point is that these developments (digital fab. technologies)
eventually also level the playing field of production of physical
things.

>
> Anyway, just a minor distraction in the discussion.
>
> > loop is something that sooths the mind. If the machine depends on a
> > human constructing it (which should be really), that's really fine. I
> > did most of the construction of my machine in a week. Debugging the
> > machine is something that took me quite long, but is exactly the time
> > which can be almost eliminated by good defaults, more reliable and a
> > more mature design.
>
> Congratulations, by the way. :-)
>
> > Currently I'm more occupied with market trends, forces and social
> > transformation than with theory of self replication. I think current
>
> Looking for business opportunities, I presume?

Sure, but mostly to be doing what I want in an individually
sustainable way. Open source / peer production is sustainable by
itself, but as an individual you still need a source of income. I try
to combine a revenue stream with working on what I think is right. It
is interesting to see how the open source ecosystem has changed
software. Since hardware is now up for a change, it's a great thing to
be able to be part of that change!

> You're trying to model the diffusion of information ? Have you
> considered reading up on long tails and torrent trackers? I see that
> you're already on the ball via your blog though-
>
> http://blog.erikdebruijn.nl/index.php?/categories/7-RepRap

What do you suggest with the torrent trackers? My first strategy was
to measure forum activity, RepRap store sales aggregates and do some
polls. Do you mean measuring the amount of download of RepRap
software? (too little centralized and a lot of them are without stats,
such as SF.net's subversion, ad hoc files shared, etc)

> So nevermind :-). I also see that you own "replab.org", which is
> basically the project that I've somewhat described to you in my
> previous email. Maybe we can put the domain to good use soon enough.
> :-)

It's available for good use. We can discuss this, but in a couple of
weeks. Right now I've got my 4 last exams for my masters coming up.

>
> > (Turing and von Neumann are still competing to be my favorite
> > scientist)
> von Neumann.

I would expect that!

>
> >> One of the
> >> projects that we're pursuing on this list is the formal packaging of
> >> open source hardware and manufacturing systems into some format, much
> >> like a dot deb file, which makes up the 20,000 software packages for
> >> debian; these dependency trees (check out debtree for dotty-generated
> >> graphs) show how things depend on one another for it to all work.
>
> > That's exactly what I like about FOSS. A packaging system and a single
> > namespace software. However while debian can serve as good inspiration
> > and good practices, you shouldn't introduce legacy from the start. A
> > CLI might be useful for retrieving software (ideas) but 'things' are
> > more graphical.
>
> Could you elaborate more on this 'legacy'? Do you mean about making
> stupid restricting decisions? Maybe something else ?

You have several users of such a system. Some would simply want to
retreive a CAD file and fabricate it. Others are looking to make
something new and wish to reuse components. This is especially what
open CAD content would be valuable for, you can quickly combine ideas
and create something new. There are so many combinations possible, and
if a component lacks, it will probably be designed and shared by
someone.

If you currently want to make a bird house with digital fabrication,
you will find a complete one for download (go to thingiverse.com). But
if you want to modify it, and apply a template for holes to mount it
to your window with suction cups, you can also find that on
thingiverse. I think the process of recombining existing components
into a larger project should be supported. But that makes it a problem
specific piece of software, and probably needs more than a .deb
repository.

For software I would see the CLI as a big advantange since I can judge
what I want by reading. For physical objects you should see it,
multiple view or even a rotatable rendering would be desirable. I'd
argue that it need a much richer interface. A central repository would
be a very, very helpful. So, you've absolutely got an essential thing
right there. Through sporadic ad hoc sharing, you will have a lot of
transactions and not much utility. To get the network effects rolling
information should be easily accessible.

> >> something. Package management tools (like apt-get, etc.) will be made
> >> as well. This way we don't have to write scrapers for thingiverse
> >> (etc.) ;-).

Thingiverse could have an XML api for such things. I can ask Zach who
made and manages the site.

>
> > If you need a dedicated server (a Xen guest instance), my company
> > could sponsor it.https://www.budgetdedicated.com/concepts
>
> """
> We actively encourage and support an open-source ecosystem. We support
> businesses that create revenue around open-source software. We also
> encourage adoption of open standards. We supply bandwidth and a server
> for the official Wine repository, because we feel it is important to
> be able to switch to Linux and still run the few win32 programs you
> can't say goodbye to yet. We sponsor the RepRap project with bandwidth
> and a server (seeding torrents and hosting ISOs). BudgetDedicated is
> comitted to help making the RepRap project a success. We have also
> built our own RepRap.
> """
>
> Awesome, thanks for the offer. Also, kudos on running a wine mirror --
> or is it the actual official release repository?

It's the official debian and ubuntu repo. Not the wine website, though
we've been discussing that. IMHO it should be CodeWeavers job to share
some of the burden, since they reap the benefits.

> Double kudos on
> serving up the reprap project, could you drop any hints on the
> webalizer stats for the server? Surely it's being pulvarized.

I just run a torrent client off a gigabit line and for some RepRap
liveCD ISO's, not the entire project. It would not run so slow if I
did, but I don't get to decide everything ;)
Maybe in a while we will run the site in a more distributed manner
from multiple PoPs.

> Would you be willing to do some commentary on the repositories that we
> have passing around here, and the todo list etc.? Maybe we could do
> this over IRC when all of the developers are awake to do things the
> right way if we wanted to set up the server. I'm presently operating
> as a giant SPOF (single point of failure) in some of the development
> work because of my slow outgoing connections, etc.

I'd love to enter in the discussion. But for now I have some studying
to do!

--
Erik

Bryan Bishop

unread,
Jan 7, 2009, 10:09:49 AM1/7/09
to openmanu...@googlegroups.com, kan...@gmail.com

Yes, overall what we've been hashing out is a "dot skdb" format (just
something new). YAML metadata and then an internal format that
includes dot g (which is like STEP except intercompatible and better,
since the tools already exist). An internal instructions/recipe format
too- see some of the discussion on this idea:

http://groups.google.com/group/openmanufacturing/browse_thread/thread/a8d8ee245aaae97d/1fc4fbbfd4a6fb23?lnk=gst&q=recipe+representation#1fc4fbbfd4a6fb23

> For software I would see the CLI as a big advantange since I can judge
> what I want by reading. For physical objects you should see it,
> multiple view or even a rotatable rendering would be desirable. I'd

That rendering would require downloading the CAD content anyway, so in
the end it will depend on how much bandwidth you have and whether or
not you want to download the entire repository at once.

> argue that it need a much richer interface. A central repository would
> be a very, very helpful. So, you've absolutely got an essential thing
> right there. Through sporadic ad hoc sharing, you will have a lot of
> transactions and not much utility. To get the network effects rolling
> information should be easily accessible.

Yeah, we've talked about "package maintainers" before, since we'll
probably have to have some sort of policy-for-inclusion into the
repositories, i.e. throwing a SLDPRT file into a zip file isn't quite
enough for packaging.

>> >> something. Package management tools (like apt-get, etc.) will be made
>> >> as well. This way we don't have to write scrapers for thingiverse
>> >> (etc.) ;-).
>
> Thingiverse could have an XML api for such things. I can ask Zach who
> made and manages the site.

I'd be willing (and am very eager) to write the code for that, so if
you really know Zach point him in my direction please :-).

Reply all
Reply to author
Forward
0 new messages