#208 – An Interview With Nadya Peek - Gallant Gcode Gerontology

Chris Gammell: This is The Amp Hour Podcast, recorded July 21st, 2014. Episode 208, with guest Nadia Peek. Gallant, G-Code, Gerontology.

Dave Jones: Welcome to The Amp Hour. I'm Dave Jones from the EEV blog.

Chris Gammell: And I'm Chris Gammell of Contextual Electronics.

Nadia Peek: Man, I feel bad that I don't have a radio voice, but I'm Nadia Peek. You're doing good. You're doing good. I'm Nadia Peek. I'm a PhD student at the MIT Center for Bits and Atoms.

Chris Gammell: Welcome.

Dave Jones: Oh, okay. So you're a student.

Nadia Peek: Yeah.

Chris Gammell: Yeah, but like, you know. How's that going? Students are like, students are also like in charge. You know, like that kind of thing. PhDs.

Dave Jones: Especially the PhDs. So how long is your PhD taking? Like forever? Are you one of those professional PhD students who just enjoy the lifestyle and just live there for 10 years?

Nadia Peek: You're starting with the really kind questions, aren't you?

Chris Gammell: You know, at MIT, though, I would live there forever, I bet. I don't know.

Nadia Peek: I'm totally going to graduate one day. Okay, one day. All right, let's get that straight. I'm a serious researcher who eventually will no longer be a student.

Chris Gammell: All right.

Dave Jones: But then you'll have to go work in the real world, and that sucks. Will I?

Chris Gammell: Well, maybe.

Dave Jones: Well, I don't know, but yeah.

Chris Gammell: We were talking, Nadia, what was your, we were talking before the show, I asked you about what your criteria eventually of working is. I thought it was a good answer. So what is your eventual criteria for, you know, post-research world?

Nadia Peek: Well, I guess it applies to research world, too. But in general, I would like to work with not assholes on shit that isn't boring.

Dave Jones: That's awesome. Well, that's what the real world is, unfortunately.

Nadia Peek: Yeah, it turns out that it's a much more constrained space than I originally expected.

Dave Jones: Although it's a lot different these days, because there's lots of, you know, there's Google Labs, and there's all sorts of other, you know, weird and wonderful, you know, hardware startups that are working on wacky stuff and all that.

Nadia Peek: Have you met anyone from Google recently?

Dave Jones: No, I've heard grumbling. Well, apparently it's exciting, you know, like, at least you are working on something exciting. As opposed to me, you worked on...

Nadia Peek: They say that it's exciting, but then they also say that they can't tell you what it is.

Dave Jones: Yeah, that sucks, doesn't it? Not being able to, being under NDA. Boo.

Nadia Peek: Yeah, yeah, yeah. I'm not so sure. I'm not so sure about it. I also think that it would be cool to work on technology that has immediacy in its application, as opposed to working on technology that happens to be the whimsical desire of whoever.

Dave Jones: Well, unfortunately, to get that wish, you have to work on the boring stuff. The immediate stuff is usually, not always, but probably in the vast majority of cases, the boring stuff. Unfortunately, it's, you know... Like the impactful stuff, you mean? It's getting to market. Oh, well... Yeah.

Chris Gammell: Yeah, but think, you're in the research space, right? So maybe it's a different definition for you.

Dave Jones: It's, yeah, totally.

Nadia Peek: Of what's not boring?

Chris Gammell: No, of like what's immediate as well, like what's needed. Oh, yeah, yeah, yeah. Yeah, Dave's talking development side, right? And if you're on the research side, there's always a contrast.

Dave Jones: Yeah, I'm talking a product that makes it to the market, be it a consumer market or a niche engineering market or something like that. Yeah, it's usually, you know, a new version of a product or it might be a whole new product, but it's still boring. It's got to go through, you know, production, which is one of the most boring aspects of product design is getting something into production and out there. Is it? I think, well, some parts. Well, some people enjoy it, some people, you know, but generally, yeah, it's pretty boring, got to admit.

Nadia Peek: Okay.

Dave Jones: All right. Sorry to all you production engineers out there who just love it and just can't get enough of a bill of materials and dealing with suppliers and the assembly houses and, you know, tombstoning components and whatnot.

Chris Gammell: So how about this? Let's get a reference point here. Nadia, can you tell us about your work? Because that's the reason why you're here and I think it's actually really interesting stuff. So, like, maybe kind of leading up to it, too, how you got into it.

Nadia Peek: Okay. So I mostly work on the machines that make project at the Center for Bits and Atoms. And so I don't know if you're familiar or I'll briefly introduce the concept. I'll start even earlier. The Center for Bits and Atoms is a kind of a collection of misfits or a salon de refusés, as they like to call it. We're not...

Dave Jones: What? Is that French or something? It sounds like, yeah.

Nadia Peek: And so we're not in the mechanical engineering department. We're not in the electrical engineering department. We're not in physics. Everybody that's there.

Dave Jones: So you're in Willy Wonka's Chocolate Factory? Pretty much. Is that the... Pretty much.

Nadia Peek: Right. Okay. And so way back when the Center for Bits and Atoms first started, one of the things that they were interested in is reinventing manufacturing across length scales. So from the Armstrong to building size fabrication and also rethinking computer and processor design. So these are a bunch of the people that back in the day worked on quantum computing and things like that. And so to be able to fabricate the experimental test beds that they needed, they wanted a lot of equipment. And so they applied for funding from the National Science Foundation. And the National Science Foundation was so kind as to award them, let's just call it a lot of money.

Chris Gammell: Metric butt guns? Yeah.

Nadia Peek: And so when you receive a large quantity of taxpayer money for the purpose of research, generally the NSF or people like the NSF want you to do some kind of outreach so that other people understand what it is that you're doing and how you're doing it. And most people do that in the form of like a pamphlet or a white paper or they say, oh, this is what we do, or they'll do a magazine. And so at CBA they thought that sounded kind of boring. And so we had... Well, it is. Right. And so we had, because of this grant, there was a whole fabrication center that was started that had water jet cutters, but like scanning electron microscopes, like that range of length scales. I don't know all of the things that were bought in the beginning, but I would say that now we have, I don't know, several million dollars worth of fabrication equipment, focus ion beam up to like wire EDM. And so what they decided... Sweet. Yeah, it's pretty fun. And so what they decided that they were going to do was to create a kind of analogous lab with some of the more accessible and cheaper digital fabrication machines in kind of inner city Boston so that anyone who wanted to use them would have access to them, not just researchers at the Center for Bits and Atoms. So they put into that space that they called a fabrication laboratory or Fab Lab. They put a laser cutter, like a 35-watt laser cutter, a small format milling machine, a larger milling machine for doing kind of four foot by eight foot sheets of plywood, and a full electronics workbench with a bunch of components that you could do some basic circuit design with. And they kind of just looked at how things would grow from there. And it turns out if you give people access to digital fabrication equipment, they like making all kinds of things that you wouldn't otherwise expect them to make. I think some of the popular things that were made at the original incarnation of that Fab Lab were like a browser for a parrot so that the parrot wouldn't get bored. And I don't know, like alarm clocks that were unconventional.

Chris Gammell: The ones that ran away or something?

Nadia Peek: Yeah, that one came actually, the one that ran away actually came out of a class that was taught by my advisor, Neil Gershenfeld, at MIT. And so people were making all these kind of funny stuff. But it turns out that digital fabrication equipment is still kind of annoying to use because there are these interfaces that are kind of designed for industrial process. Or like, it's not necessarily something where you're like, oh, I'm going to use this laser cutter 10 times. So I want the interface to be something that I don't necessarily have to do any training to learn how to use. It's not really quite like that. And so because of that, we have some projects ongoing at CBA where we redesign both the design software and the CAM software. So how you control the machines is kind of one unified open source interface. So you don't have to learn all of the different whatever nonsense machine format streaming software that goes with each individual machine. It's kind of the last thing that they think about where they're like, oh, we made this machine and now we have to make a driver for it.

Chris Gammell: Yeah. You know you love Mach 3. Just admit it. Everybody loves Mach 3. I'm kidding. No one likes Mach 3.

Dave Jones: Yeah, exactly. Exactly. And I'm going through the page of all your tools here, all the tools that you have in the lab. And like, I'm just, you know, I'm drooling. Right. And then I get down and it goes, there's a sewing machine.

Chris Gammell: Of course.

Dave Jones: It's like, of course. Okay. It's like all these, you know, there's got diamond laser saws and cutters.

Nadia Peek: We have two sewing machines. One of them is CNC controlled and the other one requires three faves power. So both of them, I would say, are pretty badass sewing machines. I don't know if you're trying to be dismissive of sewing machines, but sewing machines are super awesome machines. I think you should take that back.

Dave Jones: Well, it's just a picture of like a tiny little desktop thing here. You know, just like something that you do for your kids' clothes or something.

Nadia Peek: Which one are you looking at?

Dave Jones: I'm looking at the tools page. Here we go. I'm looking at the tools page and, you know, like halfway down, there's just this little sewing machine in between all these, you know, high power laser plasma cutters, diamond saws and wire bonding machines.

Chris Gammell: Oh, this is the programmable one. Yeah. So this can do embroidery and stuff like that. That is the, I think this is the CNC one. Yeah. Oh, okay. So I was going to say with this, this sounds all very familiar. People have read the book Fab by Neil Gershon Feldraub. Yeah. So that's like that.

Nadia Peek: I haven't gotten to the part yet where I tell you what I actually do.

Dave Jones: Oh, yes. Yeah. That helps. That's why she's here. So this is part of your PA. This is what your PhD thesis is on.

Nadia Peek: No, I haven't gotten to that part yet either. Okay. Okay. Wow.

Chris Gammell: We'll, we'll, we'll, we'll quiet down a little bit, I guess. Okay. Okay.

Nadia Peek: So machine tools are annoying to use. They're, and, uh, and the interfaces are annoying. And then also they're kind of expensive, like not in general, they're expensive. And they're also, sometimes they're just unjustifiably expensive, I guess, because there's not a whole lot of demand for them. Um, and, uh, and the manufacturers often have kind of a stranglehold on the market. Um, and because you have access to precision tools, when you have digital fabrication tools, um, you can use those tools to make stuff like parrot browsers, or, um, you could make more digital fabrication tools. So my, uh, I would say hobby and research is using digital fabrication tools to make more digital fabrication tools. Um, not necessarily that make themselves. People always ask about that. So I'm going to preempt your, are they self-replicating with, no, they're not.

Chris Gammell: Yeah. Neither is RepRap. Let's, let's not get ourselves here, folks. I mean, I guess they could print parts, but come on. It's not like they're actually making themselves.

Nadia Peek: Yeah, well, I guess the RepRap project is fun because people, people get really engaged with machine design because of that. And I think it's a, it's a, it's kind of an underappreciated aspect of, um, creating automated tools. Um, so I work on the software for controlling machines. I work on electronics for controlling machines and I work on machine design. Um, and, um, sometimes I make machines that are just kind of smaller, cheaper versions of machines that already exist. And sometimes I make purpose-built machines that are only for one thing. But because making machines is a fast thing now as opposed to a slow thing, making a purpose-built machine is not a problem. And, uh, sometimes I work on really silly machines. Um, so, so yeah, that's, uh, that's my research. And, uh, my PhD topic is specifically, uh, if you look at the different parts that are required in machine design, you know, there's a, there's a mechanical part and an electrical part and a software part. Um, each of those are kind of pretty established fields on their own, but the integration between them is all, is all. So let's, let's call it a little bit painful.

Speaker ?: Um.

Chris Gammell: Yeah. I love going through four programs to get something cut on my CNC machine. I don't know what you're talking about.

Dave Jones: So, so your thesis is about making that easier. Yeah, sure. Or making them, making them become one.

Nadia Peek: So my thesis is about kind of using the same paradigm for, like, if you take the spatial configuration of the machine and the machine also has a network of nodes that represent, um, that represent, um, that represent. Like the motors and the sensors and the actuators that the motors are talking to. Uh, uh, then how do you get all of those? How do you get all of those to fit kind of structurally within the same framework so that as a machine designer slash machine user slash, um, machine part designer, you're, you're always looking at something that kind of has the same layout.

Chris Gammell: Mm. So like abstracting to directions versus an actual, uh, instead of like talking about a motor, you talk about what direction it moves in software. Is that kind of the idea?

Nadia Peek: Yeah. Yeah.

Chris Gammell: I remember you showed some of this stuff at solid con, but it's, it's a little fuzzy now. Yeah. Yeah.

Nadia Peek: So, yeah, exactly. And then, and then, so the not, so that's, that's like the, the mechanical layer. Um, and then also the, uh, the control layer, uh, what you're, okay. So, so I don't, I don't know how, how detailed you want to get about this, but so the.

Chris Gammell: Oh, let's go, let's go deep. Let's go deep.

Nadia Peek: Yep.

Chris Gammell: Our audience can take it. I think they're, they're, they're, they'll at least go look it up afterwards if they need to.

Speaker ?: Okay.

Nadia Peek: So the way, so the way that a lot of machines currently are controlled is you have, um, you have like some cam program, uh, you have some CAD program and then you take that CAD design and use some cam program to generate G code that you use a post for that is specific to your machine. Then you stream or you click the awesome button, for example, on a Herco or Haas mailing machine that says load tape.

Chris Gammell: Really? Is that what it does?

Nadia Peek: I mean, no, it's USB.

Chris Gammell: From back when it used tape, right? Yeah, exactly. Like the actual.

Nadia Peek: But the button, the button is still called load tape, but now it's a USB stick. And so you put in generally like the DRM USB stick, um, to load G code, which is not a Turing complete language, uh, commands onto your milling machine, which it then executes. And if something goes wrong, you have to kind of guess where in the G code streaming buffer you are. And you don't necessarily know anything about the position of your machine, which was great 50 years ago before, you know. So your, um, control clock cycle would be fast enough to necessarily stop on time if something went terribly wrong. Right.

Chris Gammell: But, um, there's no, there's no real time machine necessary because there's nothing but real time, right? Yeah. Because there's clunking through instruction at a time on a, on a punch card.

Nadia Peek: And so, and so instead of using G code, um, I use something called a virtual machine. So this is, this is all kind of been developed a lot at CBA going back before my time too, with people like David Kopp and also Alon Moyer, who I work with a lot. Um, but virtual machine control where you have a machine where all aspects of the machine, so say all axes, or if you had, um, say it's a 3D printer and you have like a, a thermistor on it or something. All of those are different nodes on a machine network that are represented in your virtual machine. And so then to be able to move your machine around instead of streaming G code to it and having some PLC or something do interpretation of the G code and turn that into the acceleration profiles for your motors. Um, you're calling functions directly on nodes on a network that are the machine.

Chris Gammell: Hmm.

Nadia Peek: I don't know.

Chris Gammell: So I was going to ask you what you, what you thought about G, like, what is your feeling about G code? Is it like.

Nadia Peek: I mean, G code, it was great for my grandfather.

Chris Gammell: My grandfather was a machinist too, actually.

Nadia Peek: Oh, my, my grandfather wasn't a machinist, but I mean, just in terms of.

Chris Gammell: Mine was, but he, he, he got put out of work by CNC machines. Uh, you know.

Nadia Peek: Right. Yeah. Hmm. That, that, that's a whole nother story. That's a whole nother story.

Chris Gammell: Right, right, right. Yeah.

Nadia Peek: And I, I think that's, that's a, that's a, I think that CNC and automation, it's a, it's a conversation that is always kind of awkwardly angled into labor politics. And so, um, there's, there's a, there's this, there, there's a, there's, well, take architects, for example, right? Architects make some design and they have kind of, uh, not to generalize too much, but often they have little regard for exactly how.

Chris Gammell: Reality. Yeah. Yeah.

Nadia Peek: Or just how, how, how something might be made, you know? And when, um, when like Frank Gehry designed the, that opera or Disney concert hall and, and tried to specify a bunch of sheet metal to within 16th of an inch tolerances, you know, everyone was like, this is this person, this Frank. But, but, you know, you can kind of, you can kind of catch up with that by building, you know, Gehry Technologies developed a lot of interesting ways of making sure that they were manufacturing things to the specification that they needed. Um, and, and it kind of moved technology forward. And with that, you know, you create a lot of new jobs that are about developing new technology. And so if you say, okay, how do we incorporate educating people in new advanced manufacturing techniques into the design of advanced manufacturable things? That's an interesting conversation. But just to say, hey, you know, CNC is, CNC is putting people out of work. Yes. That's, that's technically true if you look at it in a, in kind of a small zoomed in way. Oh yeah. But, but, but I think that, um, really the obligation that we have is to, um, invent technology that, um, accounts for how people are going to use it, who's going to use it and how people are going to be educated to use it. And in the case of G code and the 17 million different like workarounds that you have to go through to get, uh, to get from one place to another in the like 180 hours worth of training that somebody who like would use a five axis Mori Psyche milling machine would have to do. You know, that's, that's just not feasible. It's not something you can expect. Um, and so I think having a new generation of competent engineers who are able to interface with, um, electromechanical systems is really important and it's not reflected at all in the way that we design things. And that's a problem.

Chris Gammell: Hmm. So I always look at this as like, uh, like one reason this stuff persists is it's kind of a lowest common denominator too.

Nadia Peek: Yeah. Yeah.

Chris Gammell: And so like, like, it's like, you know, people understand, okay, I at least can take G code and put it through a converter and my machine will understand it. And same thing with Gerber's, right? Gerber's are still around. I'm sure Dave has lots of opinions about this. Gerber's are still around because fab houses take Gerber's, right? And they, people just, it's this lowest common denominator. Same thing for PDFs, right? Even for schematics and stuff like that. So I guess my question for you would be, how do you then standardize and not have the splintering that would be necessary? I mean, do you have to do everything custom then from now on? So like, I'm not sure what the actual format of this, of this, this language looks like then, but like, how do you standardize across machines? And do you even bother to?

Nadia Peek: Um, yeah, I think that's a, I don't, so at least the, the way that I control machines, you can still also stream G code to them. It'll just convert G code into, um, it'll just convert G code into move commands. And I think, I think, I think that software is kind of the easiest thing to prototype. You just have to build into your software the, the capability for people to extend it. People love extending software and writing widgets and tools and plugins. Um, and, uh, so the, the way that I write all the move commands is just a, it's just, you, you create a, a, a virtual machine in Python. I don't know. I think it's relatively, I think it's relatively straightforward to extend. Um, and, uh, and, and it's kind of my, I guess my expectation is that people would interface with both the software and the machine. And maybe that's an unfair expectation. Maybe you would say, oh, you know, like a machinist doesn't actually want to write any code. Um, but I don't think that going forward, that's going to be true. And, uh, and so I'm just gonna, I'm just gonna, I'm just gonna, I'm just gonna place myself there in the position. My, my, my position. Plant your feet, huh? Yeah.

Dave Jones: Speaking of, um, software and, uh, Python, what, you know, when you want to do something new, you know, okay, we're going to write this. What, what decisions go into making the choice of which language you're going to do it in? Is it because, oh, that's the favorite one, or I think that language is more suitable to do this particular job or is it?

Nadia Peek: Yeah, I think it is always changing. I think everything that you can do one way, you can also do another way. And I think you just have to accommodate lots of different users and how they might do, uh, and how they might, how they might want to use a tool. Um, I, uh, I'm a big proponent of modularity. And so. But what about software obsolescence?

Dave Jones: Yeah. What about software, uh, language obsolescence though? Like, you know, it's not like you're going to make a choice. Oh, I think Pascal is the most suitable language to write this in. Right, right. But you're not going to, you know, that would have been great 25, 30 years ago. Right. That would have made it a sensible choice. So, you know, does that come into it? Like, oh no, nobody's using this language anymore, but it's the most suitable. Fourth, fourth is fantastic to do this in. It's so efficient, but. Lisp. You know, yeah. I mean, the list is endless. How do you, does, does that factor come into it that you think, oh, look, 10, 10 years down the track, you know, nobody's going to be using this language. So we, you know, we're writing our software for 10 years down the track. So do you have to make a compromise?

Nadia Peek: I guess maybe, but I think even more so, I think that the way that you architect and implement systems, you have to, you have to acknowledge that somebody is going to re-architect and re-implement it probably within, within a few, five, five years or so. Let's call it that. Okay. And so if you, if you, for example, are epilogue laser cutter and you make a file format that you don't document, then it's going to be.

Chris Gammell: I feel a dig coming on.

Nadia Peek: Then it's going to be really hard for someone to write a new software interface to your machine, because if you don't understand your file format, how is anyone else supposed to? And so I would say that that's the more important aspect of it and choosing what, you know, I'm, I'm not, I'm not, uh, I'm not religious about any kind of software. Um, you know, the, the people on this radio show guys made me reboot into windows.

Chris Gammell: Oh, she knows. She's been rebooting. She had to say it. Yeah. We had some, we had some issues with the Ubuntu build of mumble. Ha! You said I would. I know. I said Dave would do it first. You bit me before the show that I'd mention it. Yeah, I would date five bucks. Yep. Thanks, Nadia. Yeah, thanks.

Nadia Peek: No, I guess like, I am acting sort of disgruntled, but I actually don't really care. It's just kind of funny. Because, you know, at the end of the day, it's like, okay, you guys are trying to record audio, and this is, this was the fastest way for us to get to recording audio. And so if you want to, if you want to write something in JavaScript, because kids like JavaScript these days, then.

Dave Jones: I thought JavaScript was on the way out.

Nadia Peek: Who knows?

Dave Jones: Okay, go, whatever. Okay, sorry, I'm not up with it. I'm not hip and young.

Nadia Peek: You know, go doesn't seem like a very hip language, though. It just kind of seems like refactored C to me, but I don't really. What do I know?

Chris Gammell: Erlang. Let's switch to that. Erlang is not hip. I'm just listing things that I've heard of at this point. I don't even know anymore.

Nadia Peek: I guess it's a philosophical discussion about whether or not you expect your users to be able to modify the tools that they're using or not. And I think that users should be modifying the tools that they own and are using to make them more suitable for themselves.

Dave Jones: Or to try and, you know, for like one program to try and do it all.

Nadia Peek: Well, I think that branching off is actually pretty great. I think that having custom machines for specific purposes is actually a really great outcome. I think that if...

Dave Jones: Well, it's always the best. You know, the right tool for the job is always the best.

Nadia Peek: Yeah, yeah.

Dave Jones: So, yeah.

Nadia Peek: So, I'm not concerned, I guess, about... I'm not concerned about people not wanting to contribute to the grand solution of all machine problems ever.

Dave Jones: Okay.

Nadia Peek: Wrong.

Chris Gammell: One machine to rule them all. Yeah. So, what about the assertion then? Like, so, when I was buying a milling machine, someone advised me, buy a tool, not a project, right? And even the tool I bought, you know, there's no such thing really with CNC machines. I think I found out. But, like, what about that, though? Because, like, I feel like that's part of the underlying thing here. You know, you're talking about people modifying the machines and that's a very good thing. But a lot of times, both the company selling machines and also the people buying machines want to just buy something to use and move on, right?

Nadia Peek: Yeah, yeah. Well, I think it has to do with the maturity of the machine tool that you're dealing with, right? You know, like, people buy kind of kit CNC milling machines and they come with, like, some motor control box that they want you to control from Linux CNC and you have to find a computer with, like, a parallel port.

Chris Gammell: I know those feels.

Nadia Peek: And so, yeah, I think that that has to do with the maturity of that kind of project, CNC, like, CNC machines. And, yeah, I would say that I don't have one of those because I don't have a computer with a parallel port.

Speaker ?: God.

Chris Gammell: I use a USB to parallel converter for the record, folks.

Nadia Peek: I'm sure those level conversions are going great for you.

Chris Gammell: Yeah, it's awesome. It crashes about once every 40 cuts.

Nadia Peek: And I think that that makes it or, like, pretending that that is further along than it really is is a detriment, that kind of machine tool development process. And one place that I think is kind of an interesting crossover is that people seem to have, like, not a very clear distinction between robots and machines anymore. So what is a robot and what constitutes just a machine tool? Because machine tools are getting more sophisticated and robots are getting more accessible. And somebody pointed out to me recently that the way that I control machines is kind of similar to, like, a robot control language. I think that's entirely true because I'm thinking of the way that you might think about different joints on a robot arm. I'm thinking about machine tools in the same way. And so robots are designed or robot arms are designed that you probably would put your own end effector on it and might want to interface your robot control system to also control the end effector. I can't say that. I think that they're doing a particularly good job of that. But it's something that's kind of built in.

Chris Gammell: What is the difference between the two? I'm sorry. I don't know the difference, I guess.

Nadia Peek: Well, it's just you wouldn't necessarily stream G-code to a seven-axis robot arm.

Chris Gammell: Oh, watch me do it. I will break some shit.

Nadia Peek: Yeah, I mean, you could probably write an interpreter that would take G-code and, like, do the inverse kinematics and figure out how to map the coordinate system that your G-code is into whatever you decided to set as a coordinate system for the robot arm, I guess. But that wouldn't be common. As opposed to, like, a machinist would say, yes, I'm going to take G-code and load it into this machine and I'm just going to set the coordinate system with G60 or whatever the G-code command is for that.

Chris Gammell: So is the difference just the complexity then and the requisite, like, abstraction layer that's necessary because of all the dynamics you have to do for robots? Is that the difference?

Nadia Peek: I think that the difference is just the people that – the history of how you would program the machines. With robots, it doesn't have necessarily the same historical baggage of how you expect it should be done. And so when people are like, oh, yeah, I'm going to build, like, a rhino-grasshopper interface to moving around the 7-axis robot arm, everyone's like, great, that seems like a great idea. That seems like something we can do. Whereas if you would say something similar and want to hook it up to a CNC milling machine, I think.

Chris Gammell: A Haas VF3?

Nadia Peek: Yeah, I think people might be recalcitrant.

Chris Gammell: Well, you'd have to pay another $20,000 for – I was talking to Nick Pinkston about that. He was talking about, like, buying another 2 megabytes of memory or something like that for – I don't know – or 2 gigabytes for, like, 20 grand or something crazy like that. It was – Yeah, it's people. It was basically because of the market. You know, they were going to the market.

Nadia Peek: And I don't think it's even just the – I don't think it's even, like, people being nefarious. Like, if you take, for example, weaving machines, right, if you take industrial looms, I think that every line that you want to program in a weaving machine is called a pick. And so industrial looms have 80,000 picks. And if you want to have one that has more than that, you have to pay a lot of money. And it's not because it's conceptually a hard thing to do. It's because it was never really designed for that in the first place. And so you're paying someone to, like, hack their own system and figure out how they could make it work. All right. Yeah. And –

Chris Gammell: Front-loading NRE, basically, into the price of the part.

Nadia Peek: So it's a – I think it's, like, a development cycle difference where robotics is something that is used to a quick development cycle. And machine tools are not used to a quick development cycle. And so interfacing with them in different ways has different expected outcomes. It's a semi-articulate argument for why machines and robots are converging.

Chris Gammell: Second PhD right on the line right there. I mean, I'm just saying, you know. Right. So what about some of these machines? So I got to see some of these at SolidCon, like I mentioned. But could you explain just kind of the – I mean, they're kind of discrete pieces. They're modular. You can kind of bolt them together and do interesting things with them. But just what are the pieces, basically?

Nadia Peek: So that's just the machine that I brought to Solid. That's a series of machine tools that I was developing with James Coleman. So before the – I guess the first – okay. So at the Center for Pits and Atoms, we teach four classes. One is, like, a physics class and one is, like, a math class. And then there's one class called How to Make Almost Anything, which you might be familiar with if you read Fab. And then there's another class called How to Make Something That Makes Almost Anything, which is about machine tools. And so in the – not the most recent, but the one before that machine building class, one of the things that came out was we want to do circuit mills. Because the way that people in Fab Lab like making circuit boards is by milling them. You don't get a whole – you can't make very complex circuit boards that way. But it's a really fast process. And the circuit board mill that was being sold in the Fab Lab was the one that we thought was most expensive per unit output.

Chris Gammell: Is this the Modelo? Yeah. Yeah, exactly. Those things are so crazy expensive. It's unbelievable.

Nadia Peek: Yeah. They're made in Japan. They say that the inside of them is actually quite nice. But, yeah, it's $5,000 for not a very fast machine with high precision but not very high precision. So that was kind of the first target, the first target unit. And so we worked on making replacement circuit board mills. And there were a lot of people that worked on replacement circuit board mills. Jonathan Ward, who now works at Other Lab, which also had the spinoff, The Other Machine Company, which is also making circuit board mills still, was one of the first ones. David Carr made a circuit board mill version. And so that was kind of the start of these projects. And controlling those circuit board mills was annoying because that's how we realized that there isn't really a good solution for rapid prototyping of control systems for machine tools. So that's kind of where we started working on virtual machine control. So I don't know if you've seen the MTM snap, but it's a circuit board mill that's made out of cutting board material, HDPE. So you can get in lots of different FDA color code approved colors. Oh, yeah. prototyping machines is that they have no cutting force. So you can have, like, the floppiest, crappiest mechanical machine in it. And it'll still print something. Right.

Chris Gammell: Versus just stopping, like hitting something and stopping. Yeah, right.

Nadia Peek: And so then because we're working on network control systems, we were also thinking about can you use one machine to control lots of different kinds of heads? So can you have a machine that's both a milling machine and a 3D printer and maybe some other things? And so Owen Moyer and I built a briefcase 3D printer, milling circuit board mill, vinyl cutter. Also has, like, a liquid handling head. I don't know if you've seen that. But it was, it's called, we called it the PopFab, pop-up fabrication device.

Chris Gammell: So I am on the MTM, CBA, MIT.edu. And this has all of the awesome, this is, like, so many great little machines on here. I love these things. Yeah. So that was the PopFab one? PopFab, yeah. PopFab, PopFab.

Dave Jones: Looks sweet. Oh, yeah, this one.

Nadia Peek: So that one is all, oops, that one is all water jet cut. And it has this cool, it has this cool timing belt stage design that Alon called the Core XY, which is, it's pretty nice because it means that your motors are stationary. So you're not moving out around a whole lot of mass, so you can move it pretty fast. But it's, because we made it pretty small, it still had enough stiffness to do circuit board milling. That was a cool machine. So then we're like, oh, no, okay, we're changing, we're changing out the heads. We're having, where you have reconfigurable control. We're changing out the heads and we have reconfigurable controls because for each head you need a few different, you need a few different kinds of circuit to, if you switch between, you know, the spindle speed controller for the mill versus the extruder head for the 3D printer. It's slightly different. And so then we're like, oh, okay, so if we're doing reconfigurable software, reconfigurable electronics, maybe we should also do fully reconfigurable hardware. And so Alon, James Coleman, Matt Hirsch, Ben Peters, and I were all kind of like in this machine building. We all kind of like this machine building space and we were, none of us have like the same advisor and we weren't even really in the same department. So we decided to like write a grant proposal to see if we could all just hang out together and work on machines for a while. So, so we did that.

Chris Gammell: Best idea ever.

Nadia Peek: Can we go hang out in somebody else's workshop and work on things that we want to work on without the interference of our advisors? Yeah.

Chris Gammell: Was there a line item for like a beer fridge or anything like that?

Nadia Peek: It's really, really hard to get MIT to pay for alcohol.

Chris Gammell: Darn. Okay. Well, at least you got the stepper motors out of it. So that's, that's good.

Nadia Peek: And, and, and so that, that's where we started prototyping the reconfigurable stages that you saw at solid. Because then we were starting to think, can we, can we abstract machines also just to the degrees of freedom that you're, that you require in, in, in, in a machine. So James made all these super awesome concept drawings of, of how, how the machines might be configured. And, and, and so the, the, the workshop that we're in is run by this guy called Vladimir Kunitsov. And he has a super baller CNC mill with tool changer. So, you know, we could, we milled, we milled, we milled all of the, we milled all of the original stage designs out of HTP in like 20 minutes. It was crazy.

Dave Jones: Yeah.

Nadia Peek: Yeah. That was awesome. It was pretty fun. Sweet. The FlexiCam, the, the FlexiCam is a cool machine. Probably weighs only around 10 tons.

Chris Gammell: Yeah. Or they do like the cast iron like bases. So they're super rigid and super heavy and they never will move.

Nadia Peek: Yeah. No, it's a, it's a nice machine. Um, and so we realized that HTP was kind of heavy for, if you wanted to stack three axes on top of each other, then that's a, yeah, that was kind of untenable. And so then, uh, we started prototyping, we started prototyping the, uh, the stages out of sheet metal. So the ones that you saw at, um, the ones that you saw at solid are made out of bent sheet metal. Um, and, and then they have, uh, kind of a, a standardized whole configuration for bolting to each other. Um, so you can have like an XY stage or an XZ stage or an XYZ stage. So the thing that, um, so the thing that we, we had at SolidCon was a four axis hot wire cutter. So it was four, four stages bolted together in to a configuration where we could control, um, a wire with a spring that was cutting styrofoam. Uh, and so prototyping the electronics for that was just me connecting four nodes together instead of connecting three nodes together. And then the, the virtual machine or writing the virtual machine for that was also really trivial. Um, I made two, two compound nodes, one XY, one UV, and then you can address them as a full XY UV move or just an XY move or just a UV move. Um, and, and at first we were cutting air foils with that because it's the kind of thing that you would cut with a four axis hot wire cutter. But I don't, we didn't think that the, the, we didn't really think the idea was coming across really well. So then we started cutting this super useful things that maybe you saw at Solid that say hell on one side. Yeah, it was like words. And yeah, on the other.

Dave Jones: Yeah.

Chris Gammell: Yeah.

Dave Jones: Well, bugger all this mechanical stuff. When am I going to get a PCB printer that can actually give me a finished PCB and none of this milled rubbish and, uh, none of this, you know, printed carbon tracks. Cause I want a real PCB. I want to push a button on a machine and out pops my PCB.

Nadia Peek: Well, Dave.

Dave Jones: Double sided, play it through. I can even, I can even live without the silk screen. Even with, even the solder mask, maybe just a, you know, double sided, play it through a hole, push a button, out it comes. Is anyone working on this? Um. And if not, why not? Well, I. I want it, damn it.

Nadia Peek: I think that, I think that is kind of exactly my point where you want something and you're yelling at me and I want you to want something and to just build it. And so I'm trying to make it easy for people like you to build your own machines. Cause there's nothing really that complicated about what you just described. It just, but it's going to require. Well, there. Like you have to, you have to actually. Well, there is. You have to. No, not really. But you're going to have.

Dave Jones: Well, there's lots of manufacturing steps which go into your traditional PCB. That's what I'm saying. You know.

Nadia Peek: Sure. But, you know, none of them are unsolved.

Dave Jones: No, none of them are unsolvable.

Chris Gammell: Dave, she's still calling you out right now. It's awesome.

Dave Jones: No, I just want some novel idea that somebody hasn't come up with. Yeah, every man needs dogs doing a PCB milling machine with a little bit and you can mill off the excess copper on the board. Whoop-de-doo. You know, and, but it's always a separate process where you plate the holes separately and you mill separately or you etch separately or you do. You know, I want someone to come up with some clever manufacturing or materials process that gives me a finished fiberglass-like board with conductive traces that's as durable as a regular PCB.

Chris Gammell: Well, maybe this is a good question, though.

Dave Jones: It's probably more of a material science thing than just, you know, like, you know, whacking some CNC motors on and coming up with some novel way to mill it. It's, you know, I think there's more to it than that.

Chris Gammell: Well, I guess my question would be maybe a better question than what Dave's asking or demanding.

Dave Jones: Demanding, damn it.

Chris Gammell: What do you see? I mean, do you see this as, so, like, you're talking about making things, like, easier to build these. Do you see this as being able to go into a production-type environment ever?

Nadia Peek: The machine tools?

Chris Gammell: Yeah, like the, so, I mean, you were talking about, like, okay, go ahead.

Nadia Peek: Yeah, definitely. I guess the, my interest is in not necessarily mass manufacturing and economies of scale that work in that way, but more, yeah, small-scale manufacturing to the level of personal fabrication. So, like, one of the use cases that I imagine when developing machine tools like this is not, can I make something that already exists, PCB fab, that already exists in really quite a mature form, easy enough for Dave to buy and put in his house. Yeah, that's kind of, that's not necessarily a super exciting problem, in my opinion.

Dave Jones: Well, it's super exciting when, if it's a new materials process or something like that, that's what I'm getting at.

Nadia Peek: All right, well, maybe we can argue about that in a second. Yeah, you know. Sometimes, even though it snows, they won't want to go to the mountain because at four o'clock they have to, like, turn on a laser. All right. I'm talking about you, Drew. And so, you know, yes, it would be really cool if you had a little box on your desk that would make ten-layer, like, blind circuit boards for you in five minutes.

Dave Jones: Yay. I don't care if it takes a day. I'm flicking.

Nadia Peek: But I think that, you know, in terms of possible new kinds of machines that you could build, one that where you're incorporating thermal processes, motion processes, feedback from computer vision, like all of those systems, people aren't really doing that. I guess maybe George Church is doing that. That's why he has so many publications in science.

Chris Gammell: I don't know who that is. Sorry. Is that a biologist?

Nadia Peek: He's a, he's a, he's a, yeah, I would say that he's, like, the biologist. I don't know if he, do you know the Wyss Institute at Harvard? I've heard of that. They make, the professors there include Jennifer Lewis, who is working on new kinds of materials for doing things like 3D circuit design. So Dave might be excited about that. And also the robotic bee. I don't know if you saw that tiny robotic bee.

Dave Jones: Yeah, I see that. Yeah, very cool.

Nadia Peek: So those are all from the Wyss Institute. And the Wyss Institute is directed by George Church. Okay. Who's also, who's also a visiting, or not a visiting, who's also a professor at the Media Lab, which is tangentially related to the Center for Business and Atoms, in that we're in the same building.

Speaker ?: Yeah.

Chris Gammell: Got it. You're in the same cafeteria or something like that?

Nadia Peek: We don't have a cafeteria.

Chris Gammell: Well, have you had lunch with George Church? Is that what I'm hearing, though? Because that'd be pretty cool. Yeah, yeah.

Nadia Peek: Well, I guess this is kind of the cool thing about the Center for Bits and Atoms is that we have lunch with cool people all the time. So George Church, George Whitesides. If George Church is biology, I would say George Whitesides is chemistry. There's a lot of cool people called George, except for one. No, two. There's two specific.

Chris Gammell: Yeah, we're talking to you, George Foreman. We don't need your grill.

Nadia Peek: It's poorly controlled anyway. Temperature curves.

Dave Jones: Yeah, I know. It's awful.

Chris Gammell: So it sounds to me like, so you're trying to create a new toolbox for people. Is that an accurate statement?

Nadia Peek: Sure.

Chris Gammell: So what are some of these wonky machines that you're – so you mentioned that you've made some weird and wonderful little machines. What are some of the weird problems that you've solved or you've seen other people solve with these?

Nadia Peek: Well, so for example, Alon at some point wanted to make a loom, a CNC-controlled loom. And he used electromagnets for controlling which things would pick and which things wouldn't. And so he needed to make – and he decided to make those. So he needed to make a bunch of coils. And then he needed a bunch of custom coils. And where do you get custom coils? You could wind them by hand. Took a long time. Wasn't very precise. You make a machine to wind the coils. Exactly. So he made a machine to wind the coils. And because we've been working on controlling machines easily and prototyping machines easily, I think he made the coil winding machine in like two hours. And it made perfect 50-ohm – or was it 8-ohm? I forget. Coils every time. Yeah.

Chris Gammell: That's intense.

Nadia Peek: And so then he used that to make the loom. And then the loom is like not a – it's a friendship bracelet loom. It's just like a silly project.

Chris Gammell: Yeah. I'm looking at it right now. 24-bit friendship loom? Exactly. Exactly. Yeah.

Nadia Peek: And so lowering the barrier of entry for automation so that you have access to more precision enables you to build things that you otherwise wouldn't. And I think that part is really exciting.

Chris Gammell: Got it. Yeah. Neil's book kind of talks about that too because of like when they pushed the Fab Labs over to like –

Nadia Peek: I totally haven't read Neil's book.

Speaker ?: Okay.

Chris Gammell: I won't tell him. It's fine. But I'm telling you right now. But it's basically that. Like when you open up access, like all of these different applications pop out that most people wouldn't think about. Yep. Actually, that's probably a really good example of this where, you know, like there's stuff around a low-income village that might need a tool, but they can't actually even get to the point where they're making the tool. So you should read his book. He's a smart guy. So telling.

Nadia Peek: I talk to him every day. Don't worry. I know.

Chris Gammell: Well, that's fine too, I guess. I guess you probably get some of that then. Yeah. This is a cool looking thing though. And it's just like a woodcut little magnet machine. Yep. I guess, huh?

Nadia Peek: Yep.

Chris Gammell: Huh.

Nadia Peek: And it has a browser interface. So like your question about like is it Python or is it Erlang or whatever, you know, like people are relatively proficient at using browsers.

Chris Gammell: So you just click and point and it goes. Yep. That kind of thing.

Dave Jones: Yep. My three-year-old can use a browser. Yeah, amazing.

Chris Gammell: That is interesting too. So like, so what would it take for one of us to get started here then? So like, so file design files for the axes are on GitHub or anything like that? Or is this stuff published or?

Nadia Peek: Yeah. If you click on like the reconfigurable stage tab in the MTM site, you can see where like the cut files are linked just right from that page. And then the code run, the interface or the code that runs on the nodes and then the code for the interface is also linked in GitHub. And the bill of materials and the schematics and stuff are also all there. I don't know if you see them.

Chris Gammell: So what about the, so you, you just, did you design the circuit boards as well? I mean like. Yeah. Are those just, are they just drivers on board or are there micros as well? What's, what are those?

Nadia Peek: So that's like a, it's like a, it's an Arduino compatible board. So you can program it through the Arduino bootloader too. Um, and then you can also program them in the network. And then it has the one that, the one that runs the, those reconfigurable stages has a, has a stepper motor driver on it. An Allegro stepper motor driver. It has a really cute key sink on it. I don't know if you can, if you see that. Uh, and, and so.

Chris Gammell: I do like Allegro parts except when you're trying to buy them. They can be a little bit, uh, nasty to buy sometimes. Yeah. They're hard to find in the marketplace.

Nadia Peek: Yeah. Yeah. I, I, it's a, it's a, it's a, it's. I'm trying to keep it relatively portable. So if you wanted to use a different stepper motor driver, um, you could like the, the, the Vladimir and Dimitri and some of the people that we're working on during the, the cool workshop where we started all of this, uh, they, they prefer TI. I don't know, like everybody has their own preferences. Um, and, and so the, the, the motor driver component is, is separate from, from the, from the, um, from the communications component. And so switching them out, um, is theoretically trivial.

Speaker ?: Um.

Chris Gammell: It's just, you're just sending pulses usually, right? Isn't it just like you send pulse trains and then it's just amplifying the pulse trains? Yeah. Is that usually what it is?

Nadia Peek: I mean, but you would have to change the firmware for the microcontroller and maybe that you like that or maybe you don't like that.

Chris Gammell: Oh.

Nadia Peek: Yeah.

Chris Gammell: I mean, I guess if you're, if you're switching parts on a board, you're probably pretty, Yeah. pretty prepared to switch out your firmware. You should be at least, right? Yeah. Right, right. Yeah. So does, does the virtual, where does the virtual machine live then? Does it live on, I'm guessing it's a 328, like Atmel 328 on there?

Nadia Peek: No, the virtual machine lives in your computer. Okay.

Chris Gammell: Okay. Okay. So that, so that's just because it's running Python. The Python is. Yeah.

Nadia Peek: So, so, so, so your, your virtual machine could be implemented in Python. So let's say it is. So you have, um, so you have, um, a machine that has different parts that you can call functions on and that, and that, so like that, say it's a class, say that class that you're calling functions on subsets of, that is what I'm calling the virtual machine. So it could be, uh, it could be, it could, it could be any other language too, if that's what you prefer, but it has a persistent connection to the networked nodes.

Chris Gammell: And we say networked, you just mean, I'm looking at the, I'm looking at the picture now and it says just daisy chain just to, they're just getting like unique addresses basically. And driver one is you just say you are X, you're your X axis, number two is Y axis.

Nadia Peek: So, so when you, when you plug them in, if, uh, when you plug them in, if you don't have a virtual machine configuration file loaded already, they'll flash and they'll say they have that little button on them. So all the nodes will flash and, and, and it'll say, okay, please identify the X node. You press the button on the X node. Please identify the Y node. Please identify the U node. Please identify the V node. Um, and then that will get saved into a machine configuration file. Um, so the next time you plug in the same machine, it'll know what it is.

Chris Gammell: Oh, that's fancy. So how far do you want this to go? I mean, like, so there's lots of good pictures that people can, we, we don't need to describe them on here and they can go click through to them, but where do you want this to go? I mean, like, what, like, do, do you want to see this supplant a Haas? I mean, I mean, or do you not care where it goes or what's, what's the plan?

Nadia Peek: I mean, I, I don't, I don't, I don't necessarily think that, I don't necessarily want to supplant Haas, but I do want people to have access to small scale automation that currently don't feel like they have access to it. Um, I feel like, um, I feel, I feel like the lack of access to precision fabrication and automation for researchers like me is in a way, uh, like being stuck in the stone age fabrication wise. It's like, it's like controlling what your output can be. And then I think that really needs to change if we want to be making different kinds of technology.

Dave Jones: What are the main items which we need to, or the main materials or things that you need to process? Have you got a list of like, yeah, okay, we need to process, uh, you know, we, we need like a, you know, ABS plastic. We need to do, you know, we need to cut styrofoam. We need to cut metal. We need to cut all these things precisely. Have you got a list of materials? For manufacturing? A dream list of materials? Yeah. For manufacturing all the stuff you want to manufacture.

Nadia Peek: I'm not, I'm not necessarily just thinking about manufacturing. I think that just automation in general is something I'm interested in. And so, and so, yeah, like if you look at, uh, Amazon bought Kiva systems, right? So there's like a robot that moves around things in a warehouse and that's now owned by Amazon and Amazon is going to start being the only, or not the only, but the most efficient warehouse moving company. Therefore, the people that are going to sell you most of the stuff and that results in centralized control. And I'm sort of opposed to that. And so, and, uh, and, and so how do you, how do you make it easy for someone to say, oh, you know, like the only thing that Kiva systems robot really does is like move in a grid tile, um, move around in a grid and bring different things to different people. So how do you make it easy for people who are interested in doing the software of warehouse automation and the robotics of something? How do you make it easy for people to continue developing those things? Um, cause what I see right now is that, uh, there's this sort of techno elite that is able to make stuff and it's controlled by a certain kind of person.

Dave Jones: Um, and, uh, and evil patents and things like that. Yeah.

Nadia Peek: And, uh, patents are evil. Well, I don't know if patents are evil or not. I think patents are incumbent. I think patents are an incumbent system that, you know, you can't just say it's good or bad. It's something that you have to deal with.

Dave Jones: Yeah.

Nadia Peek: Uh, and, uh, and, and the way that certain companies deal with patents, I do think is bad. Like if you look at 3d printing right now, um, I don't know if you guys are familiar with form labs, the 3d printing company.

Dave Jones: Yes.

Nadia Peek: So form labs came out of the center for bits and atoms to, or at least Max Lebowski. Yes, I do. Uh, he's a, he was from the center for bits and atoms and they're being sued or they're constantly being sued by, um, 3d systems, which who is one of the two big players in the, in the 3d, in the 3d printing market. And, uh, and, and, and, and, and that's just, and that's just a scenario in which patents are obstructing innovation. Uh, and, uh, and, and 3d system chooses to use them in that way. Um, maybe because they don't have any good engineers of their own.

Chris Gammell: Oh, shots fired.

Nadia Peek: I'm sure 3d systems has plenty of competent people because they acquire competent people all the time.

Chris Gammell: That's right. They do. There's only two printing companies left. It's them and Stratasys. That's it. They're, they're huge. Yeah, exactly. I mean, once Bob Jot got bought, it's just like, okay, well now we're down to two.

Nadia Peek: Well, there's, there's EOS.

Chris Gammell: Oh yeah, that's true. Yeah. Um, but anyway, the scale wise, they're not as big, right?

Nadia Peek: I mean, they're, I guess my point is that for form labs to be able to participate in that space, they have suffered so much and they really are building a product that I think is pretty cool. Um, yeah.

Dave Jones: I thought they settled that though. I thought they were the first one.

Nadia Peek: There's a second. I really don't think they're going to let up on it. Uh, I don't know. Did you guys see that? Did you guys see that documentary print? The legend about 3d printing. It's like, it's a, so it has, it has like three main actors in it. Max from form labs, Brie from maker bot, and then Avi right hand tall from 3d systems. And, uh, and, and, uh, you know, Brie, people have opinions about him. I've met him. He seems like a nice guy. And I understand that going from an open source machine to a closed motion machine alienates your audience. Uh, and, uh, and, and kind of, maybe he's a little bit of a tragedy, but I think that in the movie, Brie Pettis really gets vilified for being, for wanting to be like a Steve jobs, as asshole control, evil empire, whatever, whatever. And, uh, and, and maybe some of that criticism is fair, but at the same time, like maker bot was a small startup working in the 3d printer space and had to contend with big players, including 3d systems and strategists. And they don't receive hardly any criticism for their behavior in the marketplace, um, in the movie. But otherwise it was, it's an amusing movie. You can see, you can see.

Dave Jones: Oh, I'm going to have to watch this because yeah, granted Breen didn't necessarily have a choice because back in the early days of it, uh, a lot of people bought in to make a bot. So he only had a certain amount of say in what went on. They took, you know, venture capital funding and things like that. And once you do that, you don't make the decisions anymore. Right. So you, you just have to either choose to completely leave and quit or go along with the ride, you know, and that, that, and, and that's what he did. And now he's, you know, worth hundreds of millions of dollars because he decided to go along for the ride. And well, I don't blame him for that. Um, so yeah, it's a trickle thing. It's a tricky thing. When you do these startups, you give away your choice really to drive the direction.

Nadia Peek: So maybe, maybe, unfortunately, a lot left the PhD program to do a startup. And, uh, maybe it's just because it's in the early, uh, in the early days of their startup, but they're making, uh, they're making like, uh, uh, uh, a CNC hand tool. So it's like a, you do gross positioning, you do gross positioning by hand and it does smallest positioning.

Chris Gammell: Um, wait, they're not the router. Are they? Yeah. Is that them? Yeah.

Nadia Peek: That's them.

Chris Gammell: Oh my God. That thing is, oh, that was so, oh geez. I think I talked about it on the show a couple. Nerdgasm. Nerdgasm. This one was so great though. So it's a computer vision. They have the stickers that go on the wood.

Nadia Peek: Yep.

Chris Gammell: So fun story about the stickers.

Nadia Peek: Imagine you needed an unrepeated, a non-repeating pattern to be printed on masking tape. How would you do that?

Chris Gammell: Uh, does it have to be different tape to roll to roll?

Nadia Peek: Well, all the, the pattern has to be non-competing, non-repeating for all of the tape that you put down on the material that you're cutting. Oh. And so if you get, if you get tape manufactured, you can print a repeating pattern on tape, but non-repeating patterns are not something that is available, um, for you to buy. So what do you do?

Chris Gammell: Oh, I have no idea.

Nadia Peek: You build a machine. So one of the things, one of the things that they did was build the super cool router that you saw. But another thing that you haven't seen is that they, that Elan also prototyped a tape printing machine to print non-repeating patterns on tape. And he, he's able to do that because we build machines.

Chris Gammell: That's right. That's, you should lead with that. You should definitely lead with that. That is really cool. So, wow. So wait, so they're going to just end up manufacturing their own tape then too? I mean like that's, or.

Nadia Peek: I think so. Yep.

Chris Gammell: Huh.

Nadia Peek: Or maybe, uh, or maybe they'll, they'll manufacture, uh, maybe, maybe that's going to be like the actual thing where the investors say, we don't care about this router anymore. You guys just have to become a tape printing company. But I don't know. So far, it seems like their investors are like relatively cool. Um, and, uh, and, and I guess now.

Dave Jones: I, yeah, there's no such thing as a cool investor. There's not. Oh, you know, like they might be relatively cool at first.

Chris Gammell: Does that mean that they have a load tape button on their machine too? Like the one that you were talking about? No, because that actually would make sense.

Nadia Peek: Because you put the tape on by hand.

Chris Gammell: Right. But you need to like load it when it's like looking at the tape, right? Oh, once you put the tape on?

Nadia Peek: I don't think that there's a button that says load tape, but I'll suggest that I bet along with the gift. It's hilarious.

Chris Gammell: There you go. That's good. Um, so what's it called again? So, cause I mentioned it, but that, that thing is going to do so well. Like it is the coolest machine.

Nadia Peek: Well, actually they haven't actually named the router tool itself yet, but the company, they call it Taktia, T-A-K-T-I-A.

Chris Gammell: Okay. T-A-K. Yeah.

Nadia Peek: I think they're, they're going to, they're going to kickstart that at some point. No. Yeah.

Dave Jones: I'm shocked.

Chris Gammell: So I had a question about that too. Do you, do you feel the urge ever? Are you like, you know what? I mean, like, so Elon did this, right? He left the program and he's just kind of like going off and making this thing. Do you ever, do you ever look at one of the machines you make and you kind of like scratch your head and you're like, this is the one. This could be it. Do you ever, do you ever feel that urge or no?

Nadia Peek: Um, no. It's really fun. All right. It's really fun to make more machines. And, uh, I don't know. So there's lots of people at MIT that kickstart things. Um, uh, yeah. And, uh, and even a small kickstarters, like, uh, like, um, do you guys know public lab that make, they make open source, um, science and instrumentation tools. And so. Uh, I don't know my name, maybe their tool.

Chris Gammell: What do they make?

Nadia Peek: Uh, um, like $10 spectroscopy tools.

Chris Gammell: Oh, no.

Nadia Peek: And so, and they did the kite mapping of the, of the Gulf oil spill. Um, and so they sell kite mapping kits on Kickstarter. Anyway. So I watched another, some of my students at some point did a, did a Tesla coil for Valentine's Day Kickstarter. I don't know.

Chris Gammell: I've watched. I think I saw that one. Yeah.

Nadia Peek: I've watched a lot of people fulfill Kickstarter campaigns and it just doesn't look like that much fun.

Chris Gammell: You don't like doing t-shirts and stickers as well? Dave's done one. He's, yeah. He can tell you all about the, uh, Dave, are you done yet? Are you done fulfilling yet?

Dave Jones: Yes.

Chris Gammell: Yeah. Finally. Yeah. Yeah. It took about six months. Yeah.

Nadia Peek: I guess, I guess there's certain things that I do think are really exciting. Like, have you seen, uh, have you seen the open source, the open source hardware laptop?

Chris Gammell: Oh yeah. Bunny. Yes. Yeah. Bunny. Yeah. So, so that was, that's, that's, that's cool.

Nadia Peek: Like, and I think, well, I think one of the things that's super cool about the Nofino laptop is it's not just a, it's not just a lap, like as a laptop, it's not even as a laptop for most people, it's really not that useful.

Chris Gammell: Right. Yeah. Or as a platform, right?

Nadia Peek: But as a, as a platform, it's cool. And, but, but I think the thing that I think is most cool about it is that I think it's a, in a way it's a kind of investigative journalism to figure out what it's figure out. Can you manufacture things at a affordable, if not cheap price point, um, using industrial manufacturing processes? So not like the Tesla coil where you make everything yourself and then ship it and then hope everyone is satisfied. But can you, can you, can you design a board, um, keep everything open the way that you want it? Use crowdfunding to go into production and make something that any investor would say, this is not a business model. You can't do this. Um, and, and, and, and like Bunny, Bunny and Zob's, you know, he's, Bunny's like a, he's an electronics guy and, and Zob's a software guy. Um, and, uh, and like neither, neither of them are manufacturing guys. Right.

Chris Gammell: So are we getting, well, I'd say Bunny, Bunny writes about manufacturing all the time though. Yeah. Yeah. Yeah. Anyway, sorry. I, sorry, I shouldn't catch up.

Nadia Peek: I guess manufacturing, manufacturing hardware, but not manufacturing, not manufacturing, not manufacturing mechanical things. And granted like a laptop is not a very complicated mechanical thing, but it's still, I mean, you know, they're, they're CNC milling aluminum there. They're going to injection mold ABS that there's.

Dave Jones: Well, to do a modern laptops, very like a modern Mac or something. A laptop is, it's all about manufacturing. It's all physical, mechanical engineering. And all of that. Anyone can lay out the board and, and do the PCB side of things. Yeah. Anyone.

Nadia Peek: Bunny, you're so trivial.

Dave Jones: Well, you know, no, but any decent electronics designer, right. Who can lay out boards.

Chris Gammell: Anybody who's worked at Altium as a PCB designer could definitely lay out a circuit board. Like, like it's no problem.

Dave Jones: A motherboard. Yeah, exactly. Yeah.

Chris Gammell: DDR3. No big deal.

Nadia Peek: And, and, uh, yeah, no, it's not.

Speaker ?: All right.

Nadia Peek: Well, you know, I, uh, I think they have some more boards that they're working on if you want to, uh, if you want to lend a hand.

Dave Jones: It's an open source project. Yay. Yay. I got plenty of time to do that.

Nadia Peek: No, but I think that the cool thing about it is that this is kind of a new point where you can specify something that you need without being an expert and still get it manufactured and have it be something that is useful at the level of not being a toy or like a, uh, uh, uh, uh, kind of a gimmick, but, but it's like a real tool.

Chris Gammell: That's true. That's an interesting look at it because yeah, I mean, some of the, some of the Kickstarter campaigns are pretty, pretty in depth, but usually they have some clout behind them. Whereas it's not just a couple individuals that are doing it.

Nadia Peek: So they're, they're like hardware. I hate this term, but they're hardware accelerators. You know, the first, the first time I met someone who said that they worked at a hardware accelerator, I was like, Oh, no, someone is like, I run an accelerator. And I said, Lydia. And then they were like, you are retarded. Who are you? What are you doing at this point? And, uh, and so anyway, hardware accelerators, they, some of them like, yeah, the one, there's like one in Boston called Bolt and they are kind of partnered with Dragon Innovation, I think they're called.

Dave Jones: And they, and they do all, uh, they've, they've been on the show. Everyone's been on the show.

Nadia Peek: See, it's a small world and they do. And, and, and I guess Dragon Innovation is the, are the people that have the expertise of how to manufacture things in China. So they did the, they did the, the fulfillment for iRobot and, uh, and, and then as the, as I guess maybe the cohort or the person that's doing the, that's doing the, uh, that's doing the product, you don't necessarily learn that much about how to get things manufactured in China because they already have that expertise. However, what if you want to build something that they don't care about? How do you have access to?

Chris Gammell: Well, you probably wouldn't go to their program in the first place. Yeah. First of all, first of all. Could you, could you give me 30 grand for 6% of my company you don't care about at all?

Nadia Peek: Have you ever, I noticed that, that, that, that, that, that their accelerator is called Bolt and yet their, their logo is a nut.

Chris Gammell: I did not notice that. That's, that's a good, that's a good catch.

Nadia Peek: You should bring that up with them. I've, I've, I've mentioned it. I, I, I think I'm kind of annoying.

Dave Jones: That's funny. Oh no, here she comes again. She's going to mention it.

Chris Gammell: Who is that crazy person? So, so you're, but you, so you're saying that the idea is like that you think it'd be interesting to move outside of a system where you need infrastructure to build a lot of something. That's what you're saying. Is that kind of the point behind this?

Nadia Peek: Well, I guess, I guess it comes back to this notion that I think autonomous infrastructure would be a good thing in case you want to manufacture something that isn't necessarily something that's supported by the status quo of infrastructure.

Chris Gammell: Okay. And I mean, and I guess you could, you could argue as well. I mean, so a lot of these things are, they push their way into the mainstream as well, right? You say, okay, well, why would you need a router with a stepper motor on it in the first place? But then when you go and try and sell it to furniture makers, they're like, hell yeah. Right. That, that, uh, the Taktia thing that Elan's working on, right? That, I guess it makes its way into the mainstream as well. Eventually someone's going to be like, yeah, we'll make it into a real thing or a mass produced thing.

Nadia Peek: Yeah. Yeah.

Chris Gammell: Well, that's pretty crazy.

Nadia Peek: I don't know. Is it? I think this is all just kind of steps in the obvious directions.

Chris Gammell: I guess so. But I, if I look, if I look at what you're doing, I never would have been like, yeah, yeah. I would go make tools that make tools. Like that's not a, I mean, it seems like you got that. It's not there over time as well, but, um, it's not like a natural progression to go backwards, like not backwards, but like to go back to the primitive levels where you're like, oh, we need to break this down to modular machines. So I guess that's what makes it interesting is that you have to kind of break it down to build it back up, right? You, you're not saying that I'm going to start with a CNC machine. You're saying I'm going to break it down to a single axis, abstract that, and then build it back up to something that looks like a CNC machine.

Nadia Peek: Yeah. I don't know if you're familiar with other research from the Center for Bits and Atoms, but generally everything that we do is like small things that connect together, that have local behavior, that have some kind of emergent global behavior.

Chris Gammell: I did see the one, I saw a presentation from a couple of years ago, I think it was about, it was a similar, like a kind of networked small thing, machines. It might've been, it might've been, uh, Neil that was presenting or something like that. It was like in a boardroom. It was the same kind of thing.

Nadia Peek: It's also like, for example, um, Kenny Chung, who used to be a PhD student at the Center for Bits and Atoms, now he works at NASA and he does small, um, yeah, he does these like small shapes made out of carbon fiber that you can snap together to make aggregate carbon fiber structures. Uh, so you can have space filling carbon fiber stuff or, and like the, the, the, the final, the final material that you're engineering there doesn't necessarily have the same material properties as bulk carbon fiber, but you can approximate a stiff light material. Right. But make it much lighter because you're using this engineered material. So, so that's the example of like snap together something. So he's not snapping together electronics, he's snapping together materials. Um, or, uh, there, there's another student who at some point worked on, uh, worked on snapping together small, small, um, small propellers so that you could make instead of like a single, instead of a single flying object, you have something that's more like a flying carpet. Um, and yeah, also the, uh, the, the, the processor design that was kind of before my time at CBA, it was the same thing where you do, you need more computation, you add more processor notes instead of.

Chris Gammell: Sweet. So does a lot of this stuff come from like a computing surplus? Like the fact that micros are so cheap and everything's so cheap, that's the idea to kind of distribute the, the processing idea?

Nadia Peek: Uh, and, and for the, for the processor, the, so that was the, called.

Chris Gammell: Oh, I just meant in general, like, like even the, I guess not carbon fiber stuff, but like your stuff, right? I mean, you're pushing control down locally to the, to the nodes, right? Yeah. I mean, the node, the node design is the idea of everything does its own processing and then you just give it these, these abstract commands. Yep. Yeah. Uh, yeah.

Nadia Peek: Sure. Yeah. I have never really thought of the possibility of microcontrollers being expensive.

Chris Gammell: Well, go back 20, 30 years when the, you know, like, yeah. I mean, like when, when, like, like the, the Haas machines we see today, like the big ones, right? Those were all based on mainframe or like old stuff is all mainframe centralized servers, right?

Nadia Peek: Yeah. Yeah. Yeah. Yeah. Yeah. I guess you're, I guess you're right. Um, sure. Cost is, so cost is not something that factors very heavily into academic research.

Chris Gammell: I never would have guessed that.

Nadia Peek: We're stunned. No, but no, it's actually, it's definitely something that we thought about. It's a, it's a, and, uh, the microcontrollers that we use, I think we always try to use ones that, uh, um, in quantities of a hundred or less than $4. Uh. Yeah. So, so yes, I'm going to, I'm going to go with the answer is yes.

Speaker ?: Okay.

Dave Jones: You don't have to, it always comes down to the best tool for the job. You know, if you've got, yeah, if your micro costs a hundred dollars and it's the best tool for the job, well, use it.

Chris Gammell: Right. But you're not going to see that pushed out in the world in a practical implementation unless it is.

Dave Jones: Well, it depends on what's the market you're in as well. Yeah. No, exactly.

Nadia Peek: If you're using a hundred dollar ADC to replace a $3,000 oscilloscope, then it's pretty cool. No? Exactly.

Chris Gammell: Yeah. But I mean, so like, so here's my thing, right? So, so people like, so you're basically you work on modular, you like modularity and stuff like that. Right. And oftentimes when I've looked at modularity stuff in the past, people are like, well, you know, you just put FPJs on everything and you have these standardized interfaces that are a hundred pins wide. And, you know, they use these fancy Samtech connectors and, oh, everything works great. And then it's like, and then you look at the bomb, it's like a thousand dollars per node. Right. And it's less good.

Nadia Peek: Well, I mean, I think that if you have, so did you guys see Project Aura, the phone that like Google announced? Yeah. So Aura used to be a CBA student too. Oh, it's named after this guy, Aura, who used to be a CBA student. And he's a, and he's making, and he's designing, he's designing like the modules that are going to make up Project Aura, the phone. And so they have, and so they're using something, they're using, you know, like a high speed interconnect where they have to design things, communicating very quickly in a way that not kind of. Yeah, so does something like that. They're going to have to have fancy, fancy ass connectors. Yeah. Right. And, and, and, and I think that, you know, they're not necessarily thinking about the possibility that each of the individual nodes that they're developing right now is very expensive because they have the expectation that they're going to be manufacturing that at scale. And so then the cost that the prototype costs right now for them is irrelevant.

Chris Gammell: But I mean, I guess, yeah, I guess there's that assumption, but then, you know, I guess the only thing that that thinking really helps is like that, you know, like you've mentioned at the beginning of the show that sometimes you have to push the limit. Oh, it was the Gary building thing you said. Sometimes you have to push the limits in order to, you know, force them to change manufacturing processes to make things cheaper. Yep. I could see that making sense. Yeah.

Nadia Peek: Well, so, so I think that, I think that's what they're hoping is going to happen with the Project Aura thing. I mean, they're doing a lot of things with the Project Aura thing that, you know, I'm like, I'm like, I'm going to see it and then I'm going to believe it. Like they, they want to do a machine that prints the entire phone. So it prints the circuit board and the enclosure and the display and everything in one machine. Good luck with that one. No, but I mean, I think it's, it's one of those things that is theoretically possible that it's going to require a metric shipment of engineering.

Chris Gammell: Yeah. Right. And that could be a good thing, right? Yeah.

Nadia Peek: And so, and so I guess it depends like where, who, who, who is angling effort in what direction.

Chris Gammell: Right. My thing is always with those, like, so like the Aura is a good example too. It's like, it always becomes a, the problem with, with modular systems as well as like, it always becomes a lowest common denominator problem. Like if you have a module that you want to plug in, that's an LED, you know, you want to light something up, but the connectors and everything else on, on the, you know, basically you're flipping a bit, but you needed to have all the other infrastructure there in order to flip that bit and to push it down through. That's always the problem with, with modular systems as well as the, like the overhead on a, on a node cost basis. I'm such a downer, aren't I? I mean.

Dave Jones: No, but that, that's been, that argument's been around forever. I mean.

Chris Gammell: We were talking about before the show about a, a modular system that someone in this call had, had designed. It was a little pricey. Yeah. Yeah. A little pricey.

Dave Jones: Altium. Yeah. And yeah, it's, it's, it's always the same when you try and do things modular like that. It, you know, there's a very niche window where it's viable. It's, it's always going to be the case. Always.

Nadia Peek: And I think it's one of those things where, where you can always say it's a bad idea and you can always try it anyway. And probably if you work hard enough at it, you'll figure out some way to make it

Dave Jones: work, but you have to. It's never a bad idea. It's just niche. Yeah.

Nadia Peek: You don't want to be the person who's like, oh man, that was a terrible idea. And then like, they go off into it.

Chris Gammell: Right. Yeah. Yeah. You never need more than five computers in the world.

Nadia Peek: Computers in the home. What in hell? Yeah. Seriously.

Chris Gammell: So, uh, you've, you've been traveling a lot for, uh, conferences. Are people going to be able to hear you speaking at any future conferences?

Nadia Peek: Uh, I'm going to be at the open source hardware summit. I'm going to be talking about sourcing and open source hardware.

Chris Gammell: Oh, cool. That's a good topic.

Nadia Peek: Awesome.

Chris Gammell: So that'll be in Rome.

Nadia Peek: Sometimes I have one of my core competencies.

Dave Jones: Oh yeah, that's in Rome. That's right. It's moving, isn't it?

Nadia Peek: Yeah. It's going to be in Rome.

Dave Jones: Jeez. What a, that's rough. Gosh darn it. You have to go to Rome.

Nadia Peek: I've never been to Rome.

Dave Jones: It's pretty awesome.

Nadia Peek: Yeah. All right. Good.

Dave Jones: Yeah. You'll love it.

Chris Gammell: So what is the, you said that's one of your core competencies is sourcing, like sourcing parts and stuff.

Nadia Peek: Oh, I was going to say that it's one of my core competencies is knowing how to buy shit. Sometimes people just don't know what the thing is called that they need. Oh yeah. Oh yeah. Totally. Those are obviously people that don't spend lots of time reading part catalogs.

Chris Gammell: Yeah.

Dave Jones: Well, no, that, that even happens to me. I've been in the industry forever. Right. And I know a connector, but I won't know the name for it. Right. Right. You know, and I go, oh, look, I know that bloody connector. That's the one I want, but I can visualize it. I can picture it. I'd pick it out in a catalog, in a visual catalog, but you think I can get the name of it? No.

Nadia Peek: Well, here's a, that's a kind of a thing that's been a pro, no, I don't know about a problem, but at least a challenge for the Fab Lab network where we don't only, for Fab Labs, specify kind of a shared set of tools that people should have access to, but also a base set of materials and parts that they can, that they can prototype with. And so maintaining, maintaining that base set of parts. So, so I was talking about it with Bunny at some point, not so long ago about, could you specify a base set of machine parts that people could buy and build machines from? And that's kind of hard because machines that you want to be, that should be custom designed should also have parts that are specified for the application at hand. Um, but if you go down that road too far and you do everything custom, then if it's an open source project, how is anyone supposed to replicate what you did?

Chris Gammell: Right.

Nadia Peek: Um, and I guess, I guess there's different, there's different levels of open sourceness. Maybe, uh, maybe you just have it open source so that people know what it is, how it's running code and what that code is. Um, but I, uh, you know, part of being of the school of modularity is, it's also, it's also like, uh, wanting users to extend things. And, and if you can't buy things that you're able to incorporate into your projects, then you also can't really extend them. And I think that's a big difference between open source hardware and open source software, or should we call

Chris Gammell: it free hardware? Uh oh, you're going to go all stalling on us?

Nadia Peek: Oh yeah, it's not going to be. Uh, let's just, let's defer. Let's defer that discussion.

Chris Gammell: Next time, next time, next time. And you can see Nadia speaking on the free hardware ad.

Nadia Peek: I just, I don't, I don't necessarily want to, uh, I don't want to, I don't want, I don't know if like open source hardware is, uh, I don't know if open source hardware is a fair description because it seems to imply that with just the design files, you're enabling people to interface with the hardware. I think that's kind of a misnomer. So, so maybe free hardware is better. Is it better? Yeah. But free hardware also just sounds like you can get free.

Dave Jones: Right. Sounds like something. Give it to me for free. I'm not behind that crap.

Nadia Peek: Hey, what you calling crap?

Chris Gammell: Yeah. No, that's actually, that's the talk I gave at SolidCon. It was about, uh, you know, open source hardware, but like if the tool, if the, the, I don't know if there's that talk, but it was basically about if you aren't using open design file formats, you know, how deep does it go? Right. It's like the inception.

Dave Jones: Well, it goes right down to the silicon level. The AVR processor in the Arduino is not open.

Chris Gammell: Right. Exactly.

Dave Jones: Right. So, well, yeah.

Chris Gammell: Right. And you couldn't go and you couldn't replicate the 328, uh, on your chip printer.

Dave Jones: No, exactly. Yeah. You're tied in.

Chris Gammell: Nadia, can you make me a chip printer?

Nadia Peek: A what printer?

Chris Gammell: A chip printer. Could you make that for me?

Nadia Peek: Uh. Dude. How much do you want to pay me?

Chris Gammell: That's, that's a longstanding.

Dave Jones: It's not a matter of pain. It's just not possible. It's a trick question.

Chris Gammell: It's a, it's a longstanding, uh, uh, point of.

Dave Jones: Argument slash joke here on the AVR. Yeah, more of a joke these days. It's a joke because it's not manufacturable.

Chris Gammell: But basically, uh, Dave and I had a bet a while back about will a chip printer ever be possible on the desktop in people's homes?

Dave Jones: Ah. Well. The answer is no. Once again, it's a trick question.

Chris Gammell: No, I'm just saying if she would make it, that would be a lot easier for me.

Nadia Peek: I think that, like, there's, there's, like, technical impossibility and then there's complexity impossibility. Or I don't know, what should we call it? Like, there's, there's things that, like, this is, this is not a thing where you're like, okay, if we worked, it's not, it's not in some limit where you can say this is physically an impossibility. Right? Like, but, um, but it's kind of practically an impossibility because.

Chris Gammell: It's just dumb.

Nadia Peek: Nobody's going to work on it, right? There's no, there's, there's not enough people like Chris who want this. And, uh, I like it.

Chris Gammell: I have to make it because I want it mostly just to prove Dave wrong.

Nadia Peek: Yeah, sure. Or, like, if you were Richard Branson and felt like spending a billion dollars on it, I bet you would get it.

Dave Jones: But, um, you can't even build a traditional fab for a billion dollars.

Chris Gammell: Give up, Chris. I'm not going to give up, Dave, but I will, uh, accept that Nadia will not build it for me. That's fine. She doesn't, it's not her job. That's fine.

Nadia Peek: Oh, goodness. I can make a, I can make a potato chip machine for you.

Chris Gammell: Well, that's a good start, actually. Yeah. Chip printer. Oh, my God, we found a loophole. Chip printer on the desktop. It's potato chip printer.

Dave Jones: There's a semiconductor company called Potato Semiconductor.

Nadia Peek: Really? What do they make?

Dave Jones: Yeah. They make, uh, um, TTL, like, really fast TTL chips. But it's, like, almost one guy in a garage. I'm not kidding. It's like, you know. But you can buy one, you can buy them on eBay. You can go to eBay and buy Potato Semiconductor chips. I kid you not. That's their name. Potato Semiconductor.

Nadia Peek: No, I see it. The USA office is in San Jose. Why call Potato Semiconductor chips?

Nadia Peek: Yeah. Potato Semiconductor chips are the most popular chips in the world. They are high volume, low price, and taste good. All of the people like to eat them. All of the people are happy with them. This is exactly our goals. I'm just reading from Potato Semiconductor.com's About Us page.

Chris Gammell: Oh, my God. Yeah, it's great.

Nadia Peek: Oh, I love it.

Dave Jones: And if you hit their online store, it takes you to eBay. It's great.

Chris Gammell: There's a sourcing problem for you.

Nadia Peek: All the computers and electronics devices like our chips taste.

Dave Jones: Oh, God. Don't get me started.

Chris Gammell: We should probably wrap it up before we get into other ludicrousness here. Where can people find your work online to go check out more of your stuff?

Nadia Peek: You can see all the machines and their source files, their free files, whatever you want to call it, at mtm.cba.mit.edu. And you can read my, I call it my invariably irrelevant weblog at infosyncratic.nl.

Dave Jones: Why dot nl?

Nadia Peek: I'm Dutch.

Dave Jones: There you go.

Nadia Peek: Simple answer, huh? That's it. Sorry.

Dave Jones: Duh. Of course.

Nadia Peek: That's one of those nice things about, you're like, oh, we might live in a cosmopolitan globalized society where everything is on the internet, but the internet is still geolocated.

Chris Gammell: That's right. Yes. Yeah, they don't let you get Estonian dot ee TLDs. I looked into that at one point. Oh, really? I really wanted a dot ee TLD.

Nadia Peek: Even if you know someone in Estonia?

Chris Gammell: Well, yeah. Maybe you could do that, but I thought that'd be fun.

Nadia Peek: Yeah, yeah. That would be fun. Dot ee. Yeah, dot it is popular.

Chris Gammell: Oh, yeah. Italy. Yeah. All right. Cool. Well, I look forward to seeing more of these machines and hopefully building with some of them at some point, too.

Nadia Peek: Yeah, yeah. Oh, yeah. I forgot to tell you about the rotary ones, but whatever. You'll see it online someday. On the webs, yeah. Like next week.

Dave Jones: Yeah.

Chris Gammell: Cool. Well, thanks again. Thanks for being on the show. We appreciate it.

Nadia Peek: Yeah. Thank you very much, Nadia. Nice talking to you guys. I'm going to go boot back into Linux now.

Chris Gammell: Okay. Sorry about Windows.

Nadia Peek: Okay. Okay, bye, everyone. Bye.

Dave Jones: See ya.

Speaker ?: Bye. Bye.

Nadia Peek: administered administered administered administered administered