Mike Anderson CTO of PTR Group

(Joel Beasley at 00:00:01) We got Mike Anderson from the PTR Group. Credentials check. Mr. Anderson designs and deploys satellite systems into space. Go, no-go check. Comms.

(Joel Beasley at 00:00:11) Go. Robotics. Go. Embedded systems. Go.

(Joel Beasley at 00:00:16) Engineering, what's the status of the Yeti? Slight adjustment in gain, point 0.1, point 0.2. Yeti online.

(Mike Anderson at 00:00:22) We have a—

(Joel Beasley at 00:00:22) Go for auto sequence start. Jake has all the controls of the podcast critical functions. Counting down from three, two, one. Here we go. This is the Modern CTO podcast.

(Joel Beasley at 00:00:42) Yeah. I'm excited about this, man. I think what you're doing is the coolest thing on Earth.

(Mike Anderson at 00:00:48) Well, it's not on Earth. That's the problem.

(Joel Beasley at 00:00:51) And that's what makes it so cool.

(Mike Anderson at 00:00:53) Yes. Well, I mean, we've been working on this problem for a while. So it's good to finally see something actually happening.

(Joel Beasley at 00:01:02) And what problem is that specifically?

(Mike Anderson at 00:01:05) Well, in this particular case, we have been working on two different programs. One of them is RSGS, which is the Refueling of Geosynchronous Satellites. That's on the DARPA side. It was originally called the DARPA Phoenix program. And the other one, which is what we're working with at NASA, so we have teams at both places. The NASA side of it is called Restore-L, which is a lower Earth orbit satellite. In fact, it's Landsat 7, and we're going to grapple with Landsat 7 and then refuel it in orbit.

(Mike Anderson at 00:01:40) That's never been done before. The robotics and all of the components and everything that go along with it have never been used before. We're inventing all of this stuff as we go. And of course, we're getting some great support from the NASA roboticists, and we have our own roboticists. So most of our guys that work for PTR—I mean, we're a small company.

(Mike Anderson at 00:02:03) We're about 35 folks, but our typical engineer has 18 to 20 years worth of experience. So, you know, we've been doing embedded systems controls, flight software, and all that sort of business for quite some time. The company itself—actually this is our 18th year. We were founded in 2000. So as a corporation, we've been around for a while.

(Mike Anderson at 00:02:28) And then, you know, we've just managed to get into some of these programs like Restore-L and RSGS, the DARPA program, and it's just really cutting-edge stuff. So, as a matter of fact, I had an opportunity to go by and visit my group over at NASA when I was over there teaching a class last week, and it was fantastic to actually see all of the equipment and see what they're trying to do—some real physics problem kind of things. It's like when you're trying to grab hold of a two-ton satellite in orbit as it's moving at orbital velocities, even the slightest mistake ends up in a debris cloud that violates all kinds of treaties. So we're all trying to focus on—

(Joel Beasley at 00:03:13) On the science, and we have to worry about that as well.

(Mike Anderson at 00:03:18) Well, yes. And that's really—I mean, the science part of it is difficult, and it's a really complicated problem. Lots of embedded control theory. Lots of—I mean, the folks at NASA, the folks we've been working with—I mean, we're inventing tools to make this thing work. The little grippers, matter of fact. The arm that comes off of the satellite has to be able to change tools because we need to be able to cut wires. We need to be able to grab hold of a filler cap and screw it.

(Mike Anderson at 00:03:47) We need to be able to pull mylar back off of the heat shields. We need to put in new fuel, and then we have to undo all of that. And that all has to be done while the satellite is orbiting. The bigger problem here is that the satellite is too far away for humans to control it. So most of this stuff has to be done autonomously.

(Mike Anderson at 00:04:08) So the robot—the satellite basically runs itself. Once we get it close to the actual Landsat 7, you know, the bird itself, it takes over and runs itself. So it's really pretty remarkable stuff, even from an artificial intelligence perspective.

(Joel Beasley at 00:04:26) Are you running—how are you testing this? Are you running virtual models that run this over and over and over to simulate it?

(Mike Anderson at 00:04:32) We do have some simulation, but we actually have mockups of all of this equipment. It's in a big bay there at NASA Goddard Space Flight Center, and we're actually testing it on life-sized devices with life-sized robots. So some of the robots we have are, like, six-degree-of-freedom, seven-degree-of-freedom, two-ton robots—industrial robots that have been modified specifically for this particular application. So we're running through all of the software control stuff here on the ground. We've already done some testing on the Vomit Comet.

(Mike Anderson at 00:05:10) That is the aircraft that goes up and down to test microgravity. So we've already done some testing on that with some scale models. So we're actually approaching it from several different perspectives—both a simulation perspective, a scale model perspective, and a life-sized model perspective, with all the weights and everything. We've even been able to instrument the robotic arm so that it behaves like it's in space.

(Mike Anderson at 00:05:38) So if you walked up to it and touched it with your finger, it would actually start to move.

(Joel Beasley at 00:05:42) Whoa. Is it very difficult dealing with—so you have these, you mentioned, two physical things that are in the real world that are models, and then you go into a space that doesn't have gravity. How do you compensate for the difference in running—knowing your physical model's working when you don't have the same—

(Mike Anderson at 00:06:02) Well, fortunately, NASA has quite a bit of history in terms of understanding how large devices operate in space. We can certainly take some of the history that we've gotten from the Hubble Space Telescope. Of course, all the work that they've done on the International Space Station. So we have a fairly good history in terms of being able to understand how the physics are supposed to work. And we actually do have an arm that's on the International Space Station that we've used for testing some of this stuff.

(Mike Anderson at 00:06:34) So we are actually doing some live testing on real robotic arms in zero-G on the space station itself. So it's been rather interesting. I mean, you know, you can actually from the facility there at Goddard, we can actually drive the robot arm around on the space station, which is very kind of cool. As long as the space station's overhead, it's really kind of neat.

(Joel Beasley at 00:07:02) So were you born with a CapCom voice, or did you take vocal lessons for that?

(Mike Anderson at 00:07:10) You know, it's a weird thing. I've been teaching classes and speaking at various conferences around the world for the past 25 years or so. So it just happens. You know? And of course, there's nothing worse than being a loud engineer.

(Mike Anderson at 00:07:28) That's always kind of an oxymoron. You know, the definition of an extroverted engineer is one who looks at your shoes instead of his own when he talks to you. So—

(Joel Beasley at 00:07:43) That's funny. I talk about—I actually wrote, so I wrote this book, and then in the book about being a CTO, I talk about a lot of things that nobody talks about that are just very obvious to me. And one of the things I mentioned—I think I spent a whole chapter on it—it's called "When to Speak Up." I'm talking about the generalization is loud and dumb, smart and quiet. So when you're experiencing life, most of the people who are loud are dumb. It's like the 80-20 rule, right?

(Mike Anderson at 00:08:14) Well, I wouldn't go so far as to say that I wasn't dumb, but, you know, nonetheless, I do the best I can. Well, the issue is—and this is something that I teach and I try to teach to engineers when I'm teaching the commingling of hardware and software—one of the things that normally happens in the design of a new system is the hardware guys get together, they think of these really cool chips, they've got state-of-the-art stuff, and they're going to build it. And this is really amazing. And then they're going to throw it over the transom into the software engineers, and the software people are just going to make it work, right? It's just magic.

(Mike Anderson at 00:08:50) Unfortunately, when they pick the latest and greatest pieces of silicon, there's no device drivers for those things. The operating systems don't know anything about them. So what I try to emphasize to people, to students who are, you know, or engineers that are trying to take classes for me, is as a software engineer, first of all, you have to know something about the hardware.

(Mike Anderson at 00:09:16) You can't just be a total software person in the embedded world because the embedded world is very unforgiving for just software people. Then the other side of that is you as a software person have got to attend the hardware meetings and say, "Whoa, whoa, wait a minute. Excuse me. You realize that if you do it this way, you're going to cost us six months in the schedule because we're going to have to write all new software for this," right?

(Mike Anderson at 00:09:42) And oftentimes, the hardware people don't understand that dynamic. So they think that, well, if it takes them, you know, a month to make the hardware, then it should only take a couple of weeks to do the software. And they don't understand that the errata in these new parts is enough that software people will spend months trying to debug this stuff with basically stone knives and bearskins, because we don't have the tools to make it work because the stuff is so new that nobody's invented them yet.

(Joel Beasley at 00:10:16) Wow. Yeah. I have very limited experience with hardware. I mean, listening to you talk, I feel like the next time Netflix loads slowly, I'm going to be like, "Mike put stuff into space and moves tiny robot arms and two-ton maneuvers, and you guys can't get some simple streaming service to work," right?

(Mike Anderson at 00:10:38) Well, you know, and fortunately, I guess, fortunately for everybody, I'm not actually putting those things into space. I've got teams of folks that are trained. They know what they're doing, and they understand which end is up when it comes to both robotics and satellite systems. I just kind of provide some guidance every now and then where appropriate. You know, when—

(Joel Beasley at 00:11:02) Do you text Elon Musk and be like, "Bro, got another piece of equipment I need to get up there"?

(Mike Anderson at 00:11:08) You know, I would love to chat with him because there's some of the things that he does that I think that he could certainly do better. But, you know, not that I know everything, obviously. But we have had one of our other programs with OrbComm. And we do the flight software for a new generation of satellites called OrbComm Generation 2. And OG2 satellites, we've had several of them lost on Falcon rockets.

(Mike Anderson at 00:11:40) So I think we've had four of them splashed by Falcons over the last couple of years. So, you know, by the time he finally got the Falcon and the Dragons actually landing correctly, fortunately, we now have been able to get a good number of satellites up in orbit thanks to SpaceX. But for the first few launches, it was dismal. They didn't make insertion orbit.

(Mike Anderson at 00:12:11) They made insertion orbit, but we had a choice. We could either deliver food to the space station or put our satellite in orbit in the proper orbital plane. So, I mean, there are lots of things that went wrong that—

(Joel Beasley at 00:12:24) Did you put them on a diet? Or did you—

(Mike Anderson at 00:12:29) No. No. No. We saw our satellite go around about two and a half times before it burnt up.

(Joel Beasley at 00:12:34) Oh, wow.

(Mike Anderson at 00:12:35) This gets expensive. I mean, satellites—

(Joel Beasley at 00:12:38) Well, who covers that cost?

(Mike Anderson at 00:12:39) Well, see, there used to be Lloyd's of London would cover that, but there have been so many losses on satellites as of late that it's very difficult for them to actually underwrite a lot of this stuff. So in some cases, the launch vehicle people are actually underwriting it. So they say, "Hey, if we fail to put it in orbit, we'll pay for it." Well, okay. But that's not really the issue.

(Mike Anderson at 00:13:05) It's not really the cost of the satellite. I mean, the cost is expensive. But the reality is it takes 18 months or 24 months in some cases to make one of these satellites.

(Joel Beasley at 00:13:16) So what's more expensive? The intellectual property to design and deploy and to do the whole thing, or actually the physical manufacturing?

(Mike Anderson at 00:13:25) Wow. That's a tough one. You know, the intellectual property associated with putting something in orbit and keeping it in orbit—we've seen examples for instance, you know, I'm a ham radio operator, and we've had amateur satellites in orbit now for, you know, better of 20 years. And so there's also a whole new class of small sats that we see universities are able to put in orbit.

(Mike Anderson at 00:13:54) But we had—

(Joel Beasley at 00:13:55) Universities are putting things in orbit on their own?

(Mike Anderson at 00:13:58) Yeah. So well, what happens is they will contract to a launch vehicle, and the launch vehicle may say, "Well, hey. We have this payload that we're supposed to put up, but we have, you know, 200 pounds worth of extra space, of extra lift capability. So anybody who wants to put something up and if it fits in these dimensions, yeah, we'll launch it, and we'll tell you how much it's going to cost."

(Joel Beasley at 00:14:21) If I decide that I want to have—you seen that movie, The Astronaut Farmer?

(Mike Anderson at 00:14:24) Yes. Right. Right. Okay.

(Joel Beasley at 00:14:25) Okay. If I want to Billy Bob this thing and start going out in my backyard, how do you not—how do you do that?

(Mike Anderson at 00:14:33) Well, it turns out that the process for doing small sats, amateur small sats, is actually fairly well known at this point. There have been—they actually sell kits, CubeSat kits. There's a specific form factor. It's like a 12-by-12-inch cube. And these CubeSats, you can basically put anything you want to in them.

(Mike Anderson at 00:14:58) There are some limitations, of course. When you're dealing in space, you're in a very hazardous environment. There's lots of radiation. There are micrometeorites that hit you, and when a micrometeorite hits you at seven or eight kilometers a second, it really ruins your day. So there are some significant problems, not to mention outgassing, because your part, when it's made, it's made here on Earth, and then you put it in a vacuum and it's not made to run in a vacuum. So the little bit of gas that may be inside the chip itself will then outgas through the component and the component blows up.

(Mike Anderson at 00:15:40) So, I mean, there's all kinds of little things like that that come into play. You can fix a lot of that by using conformal coating. So you can take an epoxy and basically coat your satellite electronics in this epoxy, and it will keep most of that from happening. But still, you know, you're trying to get lift.

(Mike Anderson at 00:16:02) And right now, the big problem that the world faces in general is lift. If you take a look, for instance, well, of course, we don't have the space shuttle anymore. The Deltas, the Delta IV rockets, those are relatively old technology. Atlas rockets are still being used. Those are really old technology.

(Mike Anderson at 00:16:24) They're very, you know, they're reasonably stable, but those things still blow up. We have, you know, one of our cases for one of the OG2 satellites, we were on a Russian booster, and the Russian booster had ourselves and 12 other satellites on it, and it blew up on the pad. So it was not just our satellite. A whole bunch of people's satellites got lost.

(Mike Anderson at 00:16:47) But, you know, when people think about space today, we still tend to think in terms of, well, you know, we did all that stuff in the Apollo era and Gemini and Apollo, and that was back in the sixties. Surely we know how to build a rocket today that won't blow up. But the reality is when you're running fuel through turbo pumps and all of the mechanisms that are required to launch one of these things and the temperatures are thousands of degrees Celsius at the nozzle when the fuel is being burnt, things go wrong. Stuff happens. And when it happens, it usually happens in a very spectacular way and the whole thing blows up.

(Mike Anderson at 00:17:31) So the moral of the story is don't be too close to any launch because you never know what it's gonna do.

(Joel Beasley at 00:17:39) No. I got my wedding ring made out of tungsten because they had some really high melting point, and they said that it was used in nozzles for rockets or something like that.

(Mike Anderson at 00:17:48) They're very heavy metal. You know, so you probably noticed your wedding ring, if it's made out of tungsten, it's really heavy.

(Joel Beasley at 00:17:55) Yeah. I figured it symbolized the marriage. Right? They say ball and chain, so why not get the... I'm not gonna go down that path.

(Mike Anderson at 00:18:02) That's between you and your respective other.

(Joel Beasley at 00:18:05) Oh my god. I'm not gonna go that way. No. She's—we just had a baby.

(Mike Anderson at 00:18:11) Oh, well.

(Joel Beasley at 00:18:11) So I'm really proud.

(Mike Anderson at 00:18:12) Congratulations. Now you definitely have the chain part of it.

(Joel Beasley at 00:18:16) Yeah. No. It's awesome, though. Do you have any kids?

(Mike Anderson at 00:18:19) Yes, I do actually. My daughter is finishing up this semester at Carnegie Mellon. Nice. And she has—definitely, she's very happy to be finishing.

(Mike Anderson at 00:18:31) My wallet is incredibly happy for her to be finishing. Carnegie is a wonderful school. Pittsburgh's a great town, but, man, the cost is through the roof.

(Joel Beasley at 00:18:42) What did she go for?

(Mike Anderson at 00:18:44) She went, remarkably enough, she is in a double major: art and Japanese.

(Joel Beasley at 00:18:55) Uh-oh. Your microphone is—Houston, we have a problem. Oh, wait. Are you back?

(Mike Anderson at 00:19:04) Hello. Testing one, two.

(Joel Beasley at 00:19:06) Mike is back. Yay.

(Mike Anderson at 00:19:08) Yay.

(Joel Beasley at 00:19:09) Oh my goodness. You had us there for a second, buddy.

(Mike Anderson at 00:19:12) Yeah. Well, you know, it's somebody—Carnegie must be monitoring. They pulled the plug on me. You can't possibly tell them how expensive we are. No.

(Joel Beasley at 00:19:24) Oh my goodness. Alright. So she's gone Japanese art. That's interesting.

(Mike Anderson at 00:19:28) Right. Well, it's Japanese and art. So two different majors. It's a blended major. And it turns out she has a real interest in astrophysics as well.

(Mike Anderson at 00:19:38) Oh. So, you know, I said, okay, you can actually get any degree you want, but here's my stipulation. You have to be employable. You cannot be living in my basement when you're 35.

(Mike Anderson at 00:19:53) You have to be able to be self-sustaining. And I did find a job at the Large Hadron Collider—

(Joel Beasley at 00:20:01) Oh, yeah.

(Mike Anderson at 00:20:01) —that requires artists who understand physics. So I said, okay, there's at least one job for this mix.

(Joel Beasley at 00:20:10) Yep.

(Mike Anderson at 00:20:10) Okay. At least you might be able to get a job here. So this is good. This is good.

(Joel Beasley at 00:20:15) She gets some great travel in too.

(Mike Anderson at 00:20:18) Well, yeah. She actually—she was in the UK last semester for a semester abroad at the Chelsea School of Art. So that has also been a lot of fun for her. So I am cautiously optimistic that this will mean good things for her future. If not, I'm not gonna quit my day job.

(Joel Beasley at 00:20:38) Have you ever run into Neil deGrasse Tyson or Bill Nye?

(Mike Anderson at 00:20:45) No, I have not. I have communicated back and forth with them on occasion. I've also in touch with several other, you know, kind of physicist types that, you know, they will see me at a conference or they'll see me on YouTube. I do a—I've got a lot of videos from Embedded Linux Conference and some of the Linux Foundation organizations, some of those conferences out on YouTube.

(Mike Anderson at 00:21:09) So I'll hear from people occasionally, but I would dearly love to meet both of them personally. That would, of course, be, you know, one of those, oh, wow—especially Neil deGrasse Tyson. That would be just absolutely awesome. I've met Michio Kaku. Definitely an interesting fellow, and I bet that the other ones are just as fascinating as he was.

(Joel Beasley at 00:21:34) Well, I'll pull some strings. I'll see what I can do.

(Mike Anderson at 00:21:36) I appreciate that.

(Joel Beasley at 00:21:37) Jake, power up the Twitter.

(Mike Anderson at 00:21:39) Yeah. There we go.

(Joel Beasley at 00:21:41) I'm actually talking with the—I think Wednesday, I'm talking with the Linux Academy.

(Mike Anderson at 00:21:45) Oh, yes. Okay.

(Joel Beasley at 00:21:47) Yeah. So there's some cool people over there.

(Mike Anderson at 00:21:49) Yeah. We have—we're actually working—there's a group of us that have been in the embedded Linux community for, you know, a couple of decades. And we're starting to realize that when we go to these conferences, we all know each other. We know what we're doing.

(Mike Anderson at 00:22:05) It's basically, you know, buddies and old friends, and it's kind of like big user group meetings. But there are a number of people who come to these conferences who this is their first intro to embedded Linux. They've never seen the real-time side of Linux. They don't really have any history. So we're now setting up an apprentice Linux engineer program that will allow people to—basically we'll bring—they will actually get a target board, and we have a whole series of a couple of days of, you know, folks that are well-versed in the technology that'll be presenting information.

(Mike Anderson at 00:22:45) I'll be going into things like GDB, using the GNU debugger in user space and in kernel space, how you debug device drivers and things of that sort. So these are all kind of mysterious things that if you've never had to do it before, it seems insurmountable. But once you have somebody sit down with you and say, here, here's how this works, then it's like, oh, of course it works that way. So we're hoping that we can help foster some younger engineers, get them into the embedded Linux realm, and get some contributions going.

(Mike Anderson at 00:23:20) You know, it's—you know, even though it turns out that, you know, 90% of all the code that's generated for Linux is actually being generated by actual corporations. So Intel, HP, Qualcomm, all those folks are actually contributing. There is still a place in the Linux community for individual contributors. And we absolutely need people to even just look over code to make sure it's formatted correctly, that their warnings are gone. I mean, there's a lot of work that has to be done, and I think the Linux kernel right now is up to about 18 million lines of source code.

(Mike Anderson at 00:23:58) So there's a lot of stuff here, and we're always finding little nits and things in the system that was like, oh, I thought we fixed that. So getting enterprising young engineers up to speed in how Linux works. Understand that Linux as an operating system is running on everything from the very smallest Internet toaster all the way up to the Fortune—to the top 500 supercomputers. Linux runs across all of those platforms. There's no other operating system in the world that can do that or can claim that.

(Mike Anderson at 00:24:35) And it just—for something that was started as kind of a lark back in 1991 when Linus Torvalds posted the first version of it, it has just absolutely grown. We support 26 different CPU architectures. And it turns out that in Restore L, we are actually using Linux, the real-time Linux to control a lot of the robotic arms.

(Joel Beasley at 00:25:02) That's so interesting. So it first came out in 1991?

(Mike Anderson at 00:25:05) That's correct. So Linux has been around for over 25 years.

(Joel Beasley at 00:25:09) That's when Ruby came out. It's so amazing to see how far—the difference. They both came out in '91, and look what happened to Linux. And look what—obviously, they're not the same thing. They're totally different, but it's just amazing to see the spread of them.

(Mike Anderson at 00:25:20) Oh, yeah. I mean, when you consider that most television sets these days run a Linux derivative. Android, of course, is Linux. It's the largest Linux distribution on the planet. You know, there are just so many things that you look at it, and it's a truly embedded system. It's an appliance. You don't think of it as a computer, but you know there's gotta be a computer in there someplace, and that's the magic.

(Joel Beasley at 00:25:46) So I have limited hardware experience in—I have limited low-level operating experience. Most of my stuff is in application software, things like Ruby. Do you test when you're writing? Is the Linux code base—do you monitor test coverage? Do you write tests for this?

(Mike Anderson at 00:26:04) Oh, absolutely.

(Joel Beasley at 00:26:05) Okay. Cool.

(Mike Anderson at 00:26:05) Yeah. Yeah. No. There is—we have access to static code analysis systems. So we do static code analysis against the Linux kernel. We also do a lot of active testing. So when we put together a release candidate, the RC candidate then goes out to several thousand developers, and those that have the time and the hardware will run their tests to make sure that it does what it's supposed to do. If we find a bug, we'll report that, and that goes into the individual subsystem. So whether it's the USB subsystem or the PCI Express subsystem, the individual teams are organized in such a way that they will handle their own subsystem, and then they'll push that code up to the next level.

(Mike Anderson at 00:26:47) And ultimately, it gets all the way up to Linus, Linus Torvalds, and Linus is the final say. He will—

(Joel Beasley at 00:26:53) Really?

(Mike Anderson at 00:26:54) —either bless it or not bless it. But Linus works at the Linux Foundation, and he has access to the servers that the Linux Foundation has. So he's got access to lots of different kinds of hardware, and, you know, it gives him something to do and definitely keeps him up at nights, I think, at least in some cases.

(Joel Beasley at 00:27:14) Does he have early access to Elon Musk flamethrower?

(Mike Anderson at 00:27:20) That, I don't know. I'll maybe—I'm sure that I'll run into Linus at the Embedded Linux Conference in March out in Portland, Oregon. He lives in Portland. So maybe we get a chance to ask him.

(Joel Beasley at 00:27:33) I wonder if that runs on embedded Linux.

(Mike Anderson at 00:27:37) Who knows? We can certainly—we can certainly hope, or maybe be afraid—be afraid. I'm not sure which one we should do. But—

(Joel Beasley at 00:27:46) Well, a little bit of both.

(Mike Anderson at 00:27:47) Perhaps a little bit of both. Yeah. It's—we're starting to see Linux in a lot of automotive applications now. We already see it in some aircraft applications. So even though it's not ARINC 653 compatible or IEC 61508 or any of those nasty certifications from the government, it's still—there are key pieces like the entertainment system where we'll see Linux being used in the entertainment system on aircraft.

(Joel Beasley at 00:28:16) Look at that. Tesla vehicles use a modified version of Ubuntu.

(Mike Anderson at 00:28:21) Yes. Yes, they do. We're seeing the—it's called the GENIVI Foundation, and the GENIVI Foundation is all about Linux and automotive. So we're seeing Linux, we're seeing Android, we're seeing various derivatives, Tizen and other derivatives that are showing up in automotive.

(Mike Anderson at 00:28:39) So as we start getting to more and more artificial intelligence, we get to automatic driver assist systems, the ADAS systems. We start getting to level three and level four ADAS systems. You know, this is where you have to have a lot more understanding of what's going on. You know, it's kind of a—at level two, you know, you've got the lane keeping and the radars and all that sort of business. And actually, I heard a very good kind of quote on it.

(Mike Anderson at 00:29:08) It was—at level three, it's eyes off. Level four is hands off. Level five is brain off. So, you know, the car takes more and more active control of the system as you get to higher and higher levels. Level five ADAS is completely driverless.

(Mike Anderson at 00:29:28) Basically, it's—you don't need to have a human in it, so you can, I guess, use it to make pizza deliveries without humans, I guess?

(Joel Beasley at 00:29:36) There you go.

(Mike Anderson at 00:29:37) That could be a threat to the pizza delivery industry. Who knows?

(Joel Beasley at 00:29:41) But you could deliver pizza to humans or humans to pizza.

(Mike Anderson at 00:29:44) That is the other possibility. And, you know, one of the things that we find when you start talking about automotive systems is, you know, what happens if you get to the point where Uber, let's say, has a self-driving fleet that—you know, why do you have a minivan? Well, you have a minivan in your family because your kid has to go to soccer once a week or he has to go to the game and you gotta pile everything in and it just takes a lot of space. What if you could simply order a minivan to show up at your house at the right time and take everybody out to the game and then come back, drop you off, and then it goes back into the fleet? That changes the dynamic of how automotive works worldwide.

(Mike Anderson at 00:30:29) You're gonna talk about impacts on taxes, road use taxes, personal property taxes, the amount of vehicles on the road themselves, parking locations. I mean, you're gonna change everything with this kind of technology.

(Joel Beasley at 00:30:46) That's what I'm so pumped about. The market's moving so fast. I'm really interested to watch how the government transitions will happen because right now, it's a nice-to-have, but things are gonna be moving so fast. You're gonna have to adjust the systems to be able to move faster. Right now, it takes so much time to respond to things like that, and we're gonna have to change our government systems to be able to respond to these things faster.

(Mike Anderson at 00:31:10) Well, if we could get some artificial intelligence in government, that would probably be a good thing. But I'm not gonna go down that path either. That's—

(Joel Beasley at 00:31:17) Well, you know what? We'll just—we'll leave that part of the conversation, and I'll settle for any intelligence.

(Mike Anderson at 00:31:25) That would also be a good thing. But, again, I probably shouldn't go down that path. So, as much as I might agree with you. Yeah.

(Joel Beasley at 00:31:33) I think logic in general, we should encourage logic in society in general, not just government. I mean, just in general, we should encourage logic and critical thinking a little bit more.

(Mike Anderson at 00:31:44) That would be great. I mean, I am a robotics mentor for a high school robotics team. And it's really interesting to kind of help the kids see how the real world works. You know, I did have one student that I caught was actually Googling "what should I do with my life."

(Mike Anderson at 00:32:05) And—nice—I said to him, you really don't expect an answer, do you? And he goes, well, Google can answer everything else. Why not this? Uh-oh.

(Mike Anderson at 00:32:13) And I'm like, okay. Let me explain to you how a search engine works. And once I had finished explaining it to him, he then realized that it was a very stupid thing to do to ask Google what he should be doing with his life. But, you know, those are—that's a level of misunderstanding that exists today in reasonably intelligent people. They just don't understand how this technology works.

(Mike Anderson at 00:32:40) And as a CTO, it's part of my job to help the customers and help people that I've met at speaking engagements and things of that sort to help them understand that the technology is not magic. It is well understood by a few people, and we need to make it more understood by others.

(Joel Beasley at 00:33:02) I agree.

(Mike Anderson at 00:33:03) And this is no less the case than when you start talking about the Internet of Things.

(Joel Beasley at 00:33:07) Well, I want to talk about that right now. So I've been talking and looking at a couple different companies, have the conversation up. Right now, a really hot topic—you see all the enterprises jumping on it—is this IoT enterprise management system to manage all your devices, control systems, things like that. They're all saying Amazon's better than this, so-and-so is better than that. What do you think of this battle going on right now for IoT enterprise market share from a management standpoint?

(Mike Anderson at 00:33:35) Well, you know, the issue is that a lot of it is hype. I mean, unfortunately, there is a lot of smoke and mirrors going on here. Understand that the Internet of Things has actually been around for thirty years. It's gotten the new name and it's gotten the hot moniker associated with it, but the reality is the embedded systems that we're working with today in the IoT are roughly equivalent to the same things we had back in the nineteen eighties. I had a 32 bit 33 megahertz 68030 running with, I think it had 16 megabytes worth of RAM, and that was a fast machine back in the mid nineteen eighties.

(Mike Anderson at 00:34:21) Today, we're dealing with a Cortex M0, M0 Plus, M3. It's got 256 K bytes worth of RAM. It's got a 48 megahertz part, 32 bit processor. So they're really not that much different than what we saw back in the nineteen eighties. However, today, the mindset is very different.

(Mike Anderson at 00:34:42) The mindset today by the developers is, well, everything they've been taught in school—that everything's a virtual machine. Memory is infinite, CPU is infinite, storage is infinite. And on the x86 platforms, that's essentially true. But when you get on to some of these resource constrained platforms like the ARM Cortex series, the Cortex M series parts, you're dealing with 128 K, 256 K bytes worth of RAM, not megabytes, not gigabytes, but K bytes worth of RAM. So the knowledge of how to work in constrained resource systems—unfortunately, most of the universities are dropping the ball on that. So it means that old fogies like me have to try and explain to people, no. No, no. You can't write code like that because that won't fit. You can't just simply say, well, we'll run everything in Java because Java needs two megabytes of memory just to run the VM. And I only have 128 K, so it doesn't fit.

(Joel Beasley at 00:35:46) Right.

(Mike Anderson at 00:35:47) So it says you have to design the code differently. Now, when you look at kind of a total package, whether it's Amazon's IoT or ThingWorx or ThingSquare or Verizon's version, whichever one you want to talk about, you have to think in terms of how does it scale down to the small sensor. If we're dealing with the Internet of Things, we're dealing with two different big markets. One is the consumer market, the other is the industrial market.

(Mike Anderson at 00:36:17) The industrial market is typically estimated to be three to four times the size of the consumer market. Understand that when you put something in an industrial environment, it stays there until it dies. This is not something that you can—oh, you know, the new Fitbit 3 is out. Let me throw away my old Fitbit and buy a new one. No. No, no. That doesn't happen in the industrial space. In the industrial space, we're still running devices that are based on RS-232, RS-422, 20 milliamp current loop. All these old technologies that have been around for the past forty years, we're still running things on that stuff.

(Mike Anderson at 00:36:57) There is a water company, basically. I can't say what city they're involved with, but there are about 2,000,000 people in the greater metro area for this city. They are running the entire water system on Windows 95. And they don't dare change it because the people who designed it have retired, and nobody knows how it works.

(Joel Beasley at 00:37:26) I love it.

(Mike Anderson at 00:37:28) So when you talk about some of these new systems like ThingWorx and Amazon IoT, you have to understand what application is it going to be used in. What is it trying to address? We need to think in terms of the communications, the radios. Are we over the 2.4 gigahertz band? Are we sub gigahertz, 900 megahertz? Are we 432 megahertz? That all changes the radio propagation. Are we using a mesh topology? Are we using a cellular type topology? That has to be taken into account.

(Mike Anderson at 00:38:02) How much power do the radios use? You know, we need to understand if our goal is to run ten years on a coin cell, how do you do that? Can you do that? And we get unfortunately a lot of hype from the individual manufacturers in IoT frameworks that say, oh, we're fantastic. We speak MQTT. Well, congratulations, you and everybody else. What makes your system different? How can I achieve the goals that I have to achieve given that your code base stops at the Linux border gateway and does not get down into the distributed sensor nodes? You know, if you've got a piece of code that requires four megabytes to run, then I'm sorry, it won't run on the platforms that we have to run it in. We have to do something different.

(Joel Beasley at 00:38:55) Are there companies bridging that gap?

(Mike Anderson at 00:38:57) There are. There are some that are working towards it. Certainly, Intel and the Linux Foundation has been working on an operating system called Zephyr. Zephyr is targeting these very small platforms. There is IoTivity, which goes all the way down to the Arduino platform, I think. Of course, the AllSeen Alliance has been doing some stuff, and the Thread Group also has been doing work. So there are a lot of folks that are looking at this problem, but understand that many of these announcements that were made by Thread Group, for instance, they were made two years ago, three years ago in some cases, and they're just now getting the product out there so that we can actually get our hands on it and see how it works. The Thread Group—in order to get access to the Thread Group's SDKs, you had to be a member of the Thread Group. The smallest value you could be a member of the Thread Group was $2,000 a year. So the do-it-yourself home inventor, I'm sorry. That's out of his range, right.

(Mike Anderson at 00:40:00) So, out of his or her range. So it really is a case where you can have these big IoT frameworks, but understand they are targeting big dollar investments. And they oftentimes don't go far enough into the ecosystem to be able to handle the actual sensor as it exists out on the edge. So there is—

(Joel Beasley at 00:40:26) Are they able to monitor it or they just can't change it? Or—

(Mike Anderson at 00:40:29) Well, if you could come up with some other piece of software and radio connectivity and everything else to get the signal to the border gateway, then they can certainly do that. Now, the problem is that their framework usually stops at the border gateway. So if you are an industrial facility and you're trying to then automate with, you know, replace your old stuff with new stuff, then basically, you're on your own. You might be able to get from the border gateway, which is usually a Linux machine, all the way up to the cloud. And the data—most of these frameworks, if you look at ThingWorx and some of these others, they're focused on the data analysis piece.

(Mike Anderson at 00:41:13) That's where the real money lives for them. So you're talking about storage of lots of data in the cloud and the data scientists to do the analysis of that data. So you're looking to try and find that nugget that's buried in the data someplace. But that again, it usually stops at the border gateway. So the actual end device, the actual edge device, you're kind of left on your own.

(Mike Anderson at 00:41:39) So hopefully, that will change. I know that there are some people, some organizations that are working to try and change it. But, you know, how long is it going to take? That's an excellent question.

(Joel Beasley at 00:41:51) So I want to talk more about the robotics program your company is sponsoring.

(Mike Anderson at 00:41:55) Yes.

(Joel Beasley at 00:41:56) When I read it, I was like, alright, that's really, really cool. And then I kind of thought, are you guys taking an incubator approach? Like, you're sponsoring the program so you could pay attention to the smart talent that's coming out of it?

(Mike Anderson at 00:42:08) Well, what we do is we sponsor FIRST Robotics teams. And the teams that we sponsor, they're all high school teams. I have had the real luxury of working with some incredibly bright young people that have now gone off to Virginia Tech and Georgia Tech and other places, and they're continuing to work on robotics in those areas. So, you know, I look at it from the perspective of, you know, at some point in the future, I might be on a respirator that these people actually wrote the code for. So I've got an incentive there to make sure they understand how to do it right.

(Mike Anderson at 00:42:48) And sometimes it means that I have to undo the damage that has been done to them by their computer science departments. You know, oftentimes, they will not be trained in how to do things in real time mechanisms. They won't understand timing. They won't understand caches. They don't understand the hardware, and you have to fix that.

(Mike Anderson at 00:43:10) So when you're dealing with deeply embedded systems and control systems—automotive, aircraft, satellites—you have to know how the hardware works. That is not something that you can escape. So it's important for software developers to understand how the hardware actually functions. I've got examples that show, you know, if you just make the small little tweak in the memory alignment, everything runs 600% faster. I mean, those are demonstrable examples that people—software developers—are totally blown away by, and that's because they don't understand how the underlying hardware works.

(Joel Beasley at 00:43:50) So we have a lot of software developers that listen to this. And if you were to give them a place where they go just poke around on a Saturday afternoon with a beer and kind of see a few cool things about where the hardware meets the software just to kind of peak their interest a little bit. What sort of resource would you direct them to?

(Mike Anderson at 00:44:12) I would, you know, one of the things that's really exciting is what's happening in the maker space these days. We're seeing a lot of work on Raspberry Pis, BeagleBone Blacks, BeagleBone Blues. It is really exciting to see your code make something move.

(Joel Beasley at 00:44:30) Oh, yeah.

(Mike Anderson at 00:44:31) There are tons of small robot devices that you can pick up—four wheel drive or tank drive type robots—that you can actually get the basic chassis for about $49 to $50. And it could range all the way up to $100 depending on how much money you want to sink into it. But then you can take an Arduino or you can take a Raspberry Pi or a BeagleBone Blue and go through the process of, you know, watch YouTube videos, see what other people have done. There is nothing that's going to bring you up to speed in the hardware faster than actually working with it.

(Joel Beasley at 00:45:08) I fully agree with that. Now let's say that that 13 year old that you were helping with the whole Google problem, let's say, we'll call her a she. Let's say that she turns around and asks you—where should she be putting her time? If right now, I started writing code before 13 because my father did it and he taught me. So I had the luxury of being able to sit there and write code all day with a mentor, essentially. Right? So where would that individual put their time so that when they hit 18 or 19 and they're going to enter the workforce, that they have five years of experience in something that's hot today?

(Mike Anderson at 00:45:47) Absolutely. I think that probably the best outlet that we have available for young people today is probably FIRST Robotics. There's another program called CyberPatriot. If you happen to have an interest in cyber defense, cyber offense, or cyber defense. The CyberPatriot program, I think that one's sponsored by DARPA. I don't remember right off hand, but CyberPatriot is definitely one you can look up. There's FIRST Robotics, there's BEST Robotics, there's VEX Robotics. There are a lot of these programs that are available in the elementary schools, the middle schools, the high schools, that if you simply take the time to investigate them, you will have an opportunity to learn as much software in the amount of time that we have access to. I mean, we're professional engineers. We're working with these kids. We're mentoring these kids.

(Mike Anderson at 00:46:42) And, frankly, I've had some fantastic young ladies in the robotics program. Right now, my team captain for the software development squad is a—she's a senior this year. I'll actually be sorry to see her graduate. I've offered her a possibility of keeping her back a year just so that I can keep her for another year. But, you know, this is really—you've got to take kind of the bull by the horns here. You have got to, as a young person, find access to the programs that are available to you. Sometimes they're sponsored by churches, sometimes they're sponsored by NASA. You know, you just have to look for them. And oftentimes, if you can find yourself in one of those circumstances, you will have a tremendous leg up by the time you get to college.

(Mike Anderson at 00:47:39) I've had some of my students that are 14, 15 years old working for SAIC writing code for a defense contractor because they understand how the code needs to be written.

(Joel Beasley at 00:47:53) So if a software developer—if I had a 13 year old ask me today, he was very—he or she was very interested in sort of the next generation of just writing code based on my experience. Right? Because that's what I have to go off of. I would say start paying attention right now and start at home writing code with the voice systems and understanding the Alexas and the Google Homes because they're going to get a lot bigger in the next three or four years, and then the jobs are going to be really hot when you're graduating high school.

(Mike Anderson at 00:48:23) Correct. Voice enablement, I've done a couple of programs on voice enablement. I actually have a companion robot that I'm working on right now for the elderly that is Alexa enabled. So you can actually talk to it. And, you know, part of the thing that you need to watch for when you're trying to build companion robots is you don't want them to look like humans.

(Joel Beasley at 00:48:44) Right.

(Mike Anderson at 00:48:44) Because you don't want them to personify them simply because they're going to break down. And when they break down, you don't want somebody to get depressed over the idea that this friend of theirs doesn't work anymore. But that also means that you need to have sensors on it to be able to determine, you know, maybe you work in multiple modalities. You're looking in both the infrared and the visible spectrum to see if they're still breathing. You know, part of the problem is I have friends of mine whose parents, they don't want to go into kind of a nursing situation. They want to stay at home, and they want to be responsible for themselves. And you absolutely want to try and support that. But, oftentimes, they need to be reminded to take their medicine. They need to be reminded to have dinner or eat lunch.

(Mike Anderson at 00:49:34) You know, those are the kinds of things that small companion robots that are voice enabled can easily do. And you can have it then, if it doesn't see any movement in the house when there's supposed to be somebody there, it can then kind of send a text message back to say, hey, I haven't seen anything. And then you can come in with your smartphone. You can attach to the camera that's on the device. You can drive the device around looking for people, seeing if there's a big lump in the middle of the floor that's not supposed to be there, that kind of stuff. And Alexa and Google Home and those voice enablement systems, they are coming along. There are lots of examples. Again, Raspberry Pis and BeagleBone and those types of things. They're all Linux based for the most part.

(Mike Anderson at 00:50:19) So, you know, get a device. A Raspberry Pi is $35. I mean, there is no excuse for not having something like that. And, you know, you can get a Pocket Beagle, one of the Pocket BeagleBones for $20. It's $19.

(Mike Anderson at 00:50:34) There's absolutely no excuse for not having an opportunity to sit down, get up to speed in this technology, extend it to an Alexa. There are lots of examples out on YouTube, on Google sites that show you how to build a Raspberry Pi that is enabled with Alexa. Those are the kinds of things that somebody can do for a reasonably small investment. And if I'm a parent and I have a choice between buying the kid a $60 copy of Grand Theft Auto or $60 worth of computers and robotics equipment, there's no question in my mind what I'm gonna do.

(Joel Beasley at 00:51:17) Well, it's gonna be a lot harder to explain what hookers are to them, though.

(Mike Anderson at 00:51:20) Well, that's, you know, different kind of hookers, I guess.

(Joel Beasley at 00:51:25) But, you know.

(Mike Anderson at 00:51:26) There are hooks. They're operating system hooks. So there's a, there you go. Different kind of hook. But, you know, callbacks.

(Mike Anderson at 00:51:33) That's right. Callbacks. Yeah. Callbacks have a different meaning, I guess, in that game. Nope.

(Mike Anderson at 00:51:38) But when I talk to parents who, you know, we come in and we're kinda doing the parent introduction to what kind of STEM programs are available, we talk to parents and we say, look. You know, they say, well, we hear the robotics team is a lot of work. It's a lot of time. And we say, absolutely. That is correct. But understand that you would spend the same amount of time in basketball practice, lacrosse practice, football practice, the football games.

(Joel Beasley at 00:52:11) Oh, no. We're losing Mike again. Let's see if we can resolve it faster. Dude, so I was thinking about when he was talking about, there were these people that did this wristband that you'd put on the older people. They were trying to do a startup. I don't know if it ever got off the ground. But it basically monitors their movement, like, if they're moving around and stuff. Then I was watching these image recognition, video image recognition, like Ahroo and stuff, all those guys we had on. And they were able to, and like the Amazon, how it can tell when you're picking up objects and stuff. You could put one of these smart robots in there with a camera, and it could actually tell if they're taking, like, their prescription. Really? It could tell if they're eating food. And so, like, if you forget, you could be like, did I eat lunch today? And the robot would be like, yes. Right?

(Joel Beasley at 00:52:43) Because it would be watching you eat lunch, or know you made lunch, or it could tell you, remind you to eat the lunch if it noticed you hadn't.

(Mike Anderson at 00:52:59) Yeah.

(Joel Beasley at 00:53:00) I think, man, oh, it would be unbelievable, man. Yeah. For people that have these memory problems and stuff like that.

(Mike Anderson at 00:53:06) No, it's, I don't know, the possibilities or—

(Joel Beasley at 00:53:10) That would take, like, a minute.

(Mike Anderson at 00:53:11) Can you hear me?

(Joel Beasley at 00:53:11) Oh, yeah. Hey. We got you back.

(Mike Anderson at 00:53:13) Okay. Good.

(Joel Beasley at 00:53:14) Did you hear us talking about the nursing home thing?

(Mike Anderson at 00:53:17) Yes. I did.

(Joel Beasley at 00:53:18) Are we doing, are you doing that or are we doing that? What's happening?

(Mike Anderson at 00:53:22) Well, I don't know of anybody who's doing that right offhand, but I know I have talked to several people about the little companion bot idea. And it has a LIDAR on it, so it actually has a rotational, a spinning LIDAR to map the house. So the goal is to make it cheap enough that you can have one for every floor so it becomes more like a Roomba than it is like a little robot that runs around. It's not like a BB-8 or anything like that. It's, you know, it's more like an appliance.

(Mike Anderson at 00:53:54) And if you can make it cheap enough to be able to do that, then there's a lot of people who would pay to have somebody be able to check in on their parents and be able to keep their parents at home in a place they're comfortable, and then be able to just check in on them periodically. Of course, there's all kinds of cybersecurity issues. There's, you know, firewall issues and all that sort of business. We're gonna have to deal with all of that. But at this point, I'm just working on trying to get the system working.

(Mike Anderson at 00:54:22) Once I get it functioning and you can then talk to it, they can say, hey, Alexa, order me some pizza or, you know, change the—

(Joel Beasley at 00:54:29) The most requested skill this month. Did you know that?

(Mike Anderson at 00:54:32) No. I did not.

(Joel Beasley at 00:54:33) I got the email from Amazon two days ago. It said the most requested skills, there was, like, three of them of last month, and one of them was to order Domino's.

(Mike Anderson at 00:54:41) Well, that wouldn't surprise me. At least it's something useful. Right? Right. You know, the, and but that's the kind of thing.

(Mike Anderson at 00:54:50) You wanna be able, first of all, you change the wake word, so it's not Alexa. It's robot or computer or whatever. You know, make it more like Star Trek. What can you say? But, you know, you wanna be able to set up a SLAM system, simultaneous location and measurement system so it can measure the house.

(Mike Anderson at 00:55:09) It can then determine that there's something that's in the middle of the floor that's not supposed to be there, and then make alerts. I mean, there are, and you were not talking about something that requires incredible CPU time or the use of GPGPU. You can't buy GPGPU anymore. It turns out all the Bitcoin miners have bought all that stuff up.

(Joel Beasley at 00:55:31) Have you seen what they're doing with the video tagging that's gone open source, and it's just unbelievable what Amazon's using to detect when people are picking up objects and stuff like that?

(Mike Anderson at 00:55:43) Oh, yeah. Yeah. No. There's a lot of stuff that's happening there, but unfortunately, a lot of that is gonna be dependent on some fairly significant image processing. And image processing, when I was speaking at the Embedded Systems Conference back in December, there were a lot of presentations on image processing, and you then come to realize quite quickly the state of image processing today.

(Mike Anderson at 00:56:09) And it is making progress, but nowhere near as good as it needs to be. As a matter of fact, I've got some NVIDIA Jetson boards that I purchased for the robotics team, and those are gonna be used for vision processing, target tracking, and things of that sort. But trying to explain to high school students exactly how you do vision processing, it's a tricky problem. You know, but you have to basically distill it down to something that's easily attainable. We are very fortunate on my home team in that we have three Microsoft HoloLenses.

(Joel Beasley at 00:56:48) So I hate you. I'm flying out to hang out with you.

(Mike Anderson at 00:56:52) And we were the, we were the only team last year to actually use a HoloLens on the field. So, we are, you know, we're sponsored by NASA, and the NASA house teams have all got HoloLenses that they've been kind of given to them by NASA through Microsoft, through donations from Microsoft. And so, you know—

(Joel Beasley at 00:57:16) Smart move on their end.

(Mike Anderson at 00:57:17) Well, absolutely. And then it's not so much the virtual reality that is interesting to me. It's the augmented reality.

(Joel Beasley at 00:57:26) Oh, yeah. Speaking my language.

(Mike Anderson at 00:57:28) Because virtual reality by itself is, yeah, it's kinda cool. It's great to play a game. But if I need to know how to start a generator and I've never started a generator before, I wanna be able to put on a pair of glasses that say, here, push this button, turn this gauge, pull the crank.

(Joel Beasley at 00:57:46) So pumped for the future, Jake. I lost my voice last week, so I've been trying to talk quietly, but I wanna, like, man. Oh, dude. I wanna unleash.

(Mike Anderson at 00:57:57) Yeah. You know? And it's one of those cases that, like I say, my first programming job was back in 1977. So I'm looking at where the future is going on this, and I'm like, you know, the kids today, those young whippersnappers, they can stay off my lawn, but by God, they have an opportunity to do some really incredible stuff.

(Joel Beasley at 00:58:18) We're gonna, I'm not gonna let that thing go only because I have a friend who's building a 600 unit retirement home about two hours south of us. And, you know, they're all, like, the same floor plans and stuff.

(Mike Anderson at 00:58:30) I gotta talk to him and be like—

(Joel Beasley at 00:58:31) Dude, you gotta, we should all talk about that.

(Mike Anderson at 00:58:35) Absolutely. I think that there is a place for automation where people then start to feel, well, people are more and more comfortable with it. Even the generation above myself, I mean, my dad certainly understands computers these days. He understands what a smartphone is, and you just simply make it so that they are smart appliances. And taken from that perspective, they're not scary.

(Mike Anderson at 00:59:02) You don't have to worry about them administering you drugs or anything crazy like that. It's not like the scary movies that, you know, computers run amok. So in all those cases, I mean, you can make the technology attainable. You can make it less intimidating and get people to accept it. I mean, that's really the thing. You know, who would have thought twenty years ago that we were gonna have octo-core computers in our pocket and rely on them every day? I mean—

(Joel Beasley at 00:59:31) I'm gonna start calling on that now. Yeah. I found now I got my octo-core computer.

(Mike Anderson at 00:59:36) Absolutely. People will look at you and wonder what the hell an octo-core is, but, you know—

(Joel Beasley at 00:59:42) I'll throw an English dictionary at them.

(Mike Anderson at 00:59:44) There you go. Exactly.

(Joel Beasley at 00:59:46) I have to go get one from the antique store.

(Mike Anderson at 00:59:48) Well, that's the other thing.

(Joel Beasley at 00:59:50) I mean, you know, we used—

(Mike Anderson at 00:59:51) To have these things called books.

(Joel Beasley at 00:59:54) Oh, yeah.

(Mike Anderson at 00:59:55) And, I mean, I'm just as guilty. Certainly, I carry my, my Samsung tablet around with me, and it probably has a 150 different reference books on it.

(Joel Beasley at 01:00:06) Yeah.

(Mike Anderson at 01:00:06) Just simply because there's no way I could carry that kind of stuff. I only need it once in a while, but when I need it, I need it, and I don't wanna have to go back home to try and get it.

(Joel Beasley at 01:00:14) It was always my job. My dad had this, like, roller thing that would collapse. It was like a metal frame, and he'd put his crates and his books and everything on it. And it's always my job to, like, put the bungee cords on it and wheel it into the office. You know what I'm talking about too.

(Mike Anderson at 01:00:26) Absolutely. Well, you know, I used to work on old TV sets back in the day. And, you know, we actually had a flat panel TV that went out, and my wife says, well, let's take it to a TV shop and have it repaired. And I go, well, first of all, those things don't exist anymore. But turned out it was a bad capacitor inside of the TV set.

(Mike Anderson at 01:00:50) And by going on YouTube and just scouting around, it was called the click of death. And it was a characteristic of certain Samsung TVs. So for $3, I was able to buy the parts to replace the electrolytic capacitors in it and bring it back online. And we got another couple of years out of it before it finally died. But when it did die, it was like, okay. It's dead. Let's just throw this thing out. You know?

(Joel Beasley at 01:01:14) Put it, take it—

(Mike Anderson at 01:01:14) To the recycler, and let's get something else. I was tired of working on it.

(Joel Beasley at 01:01:19) Oh, dude, Mike. This is, I, you're, we're best friends now. Like, I don't care if you want to be or not. So I wanna wrap up the call with the last question. Elon Musk texts you back. He said, no problem, bro. We'll get those satellites into space on our sixth attempt. And then he asked you, come over, hang out with him. He's got a time machine, and you're gonna get to go back into the past and talk to yourself ten years ago, and you get to hang out with yourself for a few minutes to give yourself a piece of advice. What would it be?

(Mike Anderson at 01:01:52) Wow. If I were to go back and talk to myself ten years ago, first of all, I would have gotten into robotics a lot sooner than I did. That would have been one thing. The other aspect of this is I probably would have spent more time in artificial intelligence.

(Mike Anderson at 01:02:12) Back in the, now understand that back in those days, a decade ago, I've done some things in parallel and distributed computation theory. And in that particular model, a lot of the things that were being hyped as artificial intelligence were actually not. They were queuing theory or they were, you know, some sort of optimization theory. And, you know, we were just simply relabeling them, calling them artificial intelligence so that we could get funding from the government. You know, that kind of thing was not terribly interesting.

(Mike Anderson at 01:02:44) What we see today, however, is where we're seeing, you know, deep learning. We're seeing reverse chaining neural networks. Those are the kinds of things that if I could go back and say, there really is something here, pay attention to it, that would probably be what I would do.

(Joel Beasley at 01:03:01) All right. Pay attention to that space?

(Mike Anderson at 01:03:03) Well, I mean, if I could go back and do anything, I would go back and become a special effects expert for Industrial Light and Magic. But—

(Joel Beasley at 01:03:13) What? We're not gonna go there. That is for our next call. All right. Awesome.

(Joel Beasley at 01:03:20) Dude, thank you so much for coming on the show, Mike.

(Mike Anderson at 01:03:23) Well, Joel, it's been my pleasure. I'm glad that we were able to suffer through the couple of outages here.

(Joel Beasley at 01:03:29) Totally worth it. Thank you so much for listening to the Modern CTO podcast. Share this. Get the word out. Thank you guys so much.

(Joel Beasley at 01:03:41) I couldn't do it without you. I appreciate it. You guys are the absolute best.