 Computers keep changing the world, but their power and safety is limited by their rigid design. The T2tile project works for bigger and safer computing using living systems principles. Follow our progress here on T Tuesday Updates. This is the 10th episode of T Tuesday Updates, and how about that? The goal from last week we had just built a new T2tile based on the revised circuit board that we got from Seed Studios a few weeks ago. We saw that it was able to boot, but didn't really test it very much, tested it a bit more. In addition, we knew there were problems. For example, the J10 header with its six pins were completely in the wrong place. At least for two of the pins, couldn't put the Beaglebone down on top of it to do the revs. So today I want to try to do a little demo, show you what we've learned about the Tile. A lot of it's working, some of it not so much, but I'm not entirely sure in some regards whether it's actually a hardware problem or whether we're actually climbing up to the level that it's the development of the software that needs more work. I want to save ten minutes, half the time, something like that, to say, you know, wow, we've made it to ten. Let's take a look back, take a look forward, sort of just take the moment to see, you know, for where we're at, okay, that's the plan for today. All right, so in the new rev of the board from last time, we added a light sensor that we needed to test, and then we needed to test the intertile connectors, the ITCs. So this is the board from last time. We've actually got it plugged in and powered up here, and we've got it connecting with Ethernet so that we can actually SSH into it and take a look at it, and that's what we've got here. So the upper window here is monitoring the analog to digital channels on the board of which they're seven, and we're using essentially five and three and four are reserved for future work, future we'll see. But the grid voltage, channel A, channel zero, it's measuring 12.5 volts, 12.25 volts, that seems about right. One and two are measuring temperatures in the center of the board and at the edge of the board, 97 and 91 degrees for us, you know, Fahrenheit wings. It's fairly hot and that is not the CPU temperature. That is not the chip temperature. That's just measuring the board near the center one is measuring the board near the chip, near the CPU. So in fact, the CPU is quite a bit hotter and, you know, these guys do produce a non-trivial amount of heat. We'll see how much of an issue that is coming forward. Now, in addition, number five, the user act is what used to be on a digital pin, but I actually wanted the digital pin back more than I wanted. I had extra analog channels. So I said, well, what the heck, we can read the state of a pin with an analog channel. And that seems to work fine. Six ADC light, that's the main event. That was the light sensor. The fact that it's currently reading 3,500, 3,400, whatever it is. These are 12-bit analog to digital channels. That means their maximum range is 4K. So 3,400 like that is near the max. And the only reason we're seeing that is because there's a ton of light on in the room here for shooting the video. But it works fine. If we put the thing in shadow, look at that, thousands to hundreds like that. And we can get it all the way down further. Like I said, when there's this much light, there's absolutely no problem. In typical room light, the ADC light is typically at, you know, low hundreds. But still, if it goes into shadow, it goes down to tens or ones. So in fact, it is quite sufficient to detect what's light changes. And I think it'll be fine. It'll be sort of fun when people walk by, cast shadow on the things or shine their phone flashlight at it or whatever. It'll get detected pretty well. So that's good. But that's just the analog to digital channels. We really need to be testing the intertile connectors to see whether they can negotiate locks with each other, which is one of the main things that these tiles do so that when an event happens near the edge of the tile, they can ask the neighboring tile, could you please not mess with it while I do this? And I'll tell you what the results were. And it'll send a packet with a cache update through the, say, East or West thing. And then it'll release the lock. And maybe the other guy, it'll be able to grab the lock and say, okay, now I want to do an event and tell you what happened. This is meant to be our best effort attempt to make a single event reliable. And it's a lot easier when the single event is in the middle of the tile, but when it gets to an edge or worse, when it gets to a corner and have to coordinate with the neighbors. So, but we can't really test communications if we only have one tile, at least not well, but we can do the standard thing, which is to have the tile talk to itself. And that's what this bottom thing here, let's see. Okay, now this is not very easy to read, but the main thing is just, you know, Northwest, Northeast, ET, East, Southeast, Southwest and West. So we have a little display for each of the six directions. And we have what state it thinks it's in. So for example, Northeast is currently in state failed. East is in state reset. Southeast is idle, failed, flocks is another state and idle for Northwest. Now, this tells us several things. The first thing it tells us is that since there are no connections here, in particular, since there's no connection at all on, so Northwest, you see, if we look at the back, Northwest is this one right here. And there's no, there's nothing connected there. And idle sounds like it's not doing anything, but in intertile connection space, idle is a very active state. It means we know we're connected. We've reached synchronization for exchanging packets. And we currently just have no packet actually to send. So the fact that Northwest and Southeast both think their idle is a problem. But let's, so let's do East and West here. So we'll take a cable like this, I guess. And we'll, which way does it go? To obey the key, it goes like this. All right, so now if we're lucky when we plug this cable in, the East and West, which is in reset and flux, will actually change to something else. Look at that, okay. So now East and West are both idle because they spoke to each other through the cable and now they got coordinated back and now they know. Okay, so that's good. Why? And in fact, in a proper setup, we'd have more loopbacks. We'd have a loopback for, what is this? This is Southwest would go to Northeast and so on. Oh, and Southwest and Northeast, they actually got to idle as well. So that's good. And then there would just be one more here, but we know Northeast and Southwest. Northwest and Southeast, we know they're messed up. But so you can tie these things together. You kind of get Taurus connectivity except that the two, we have half of the tile is mapped to this side and half the tile is mapped to the other side. So it's some kind of, well, you tell me, what do you get if you have the X-direction wrapped around like a Taurus and you have the Y-direction split in half and the side swapped? Anyway, so we need to do more work there, but we've made a lot of progress. And in addition, we have revised the board in several ways. We've taken the light sensor footprint made a little bigger so that we could hand solder it, even though our goal is to get the stencil and try the reflow soldering when the thing comes back. A bunch of other little changes I won't go through, although they really are a little bit of a sign of increasing confidence on my part in that I have never actually changed a footprint that I got from Kaikad itself to say, well, I don't actually want the pad to be quite this big. I think it has too much solder because of my previous efforts doing reflow soldering. I tended to get a lot of bridges, solder connecting adjacent pins on certain of the chips that I was using. So I said, you know, if there was less solder, maybe that wouldn't be such a problem. And this past Sunday, two days ago now, I ordered a set of three prototype boards from Osh Park and they told me it could take, you know, up to two weeks or something like that. And I was waiting to have my guy assigned to a panel. But then yesterday, they told me that it had already been assigned to the panel for December 12th, which was already great since I was not expecting it necessarily to happen that quickly. And that wasn't even it. I then also got a follow-up today saying that the December 12th panel has already been shipped off to the fabrication place. It may take up to a week, a week and a half for it to come back, we don't know. But after my previous experience, my most recent experience with Seed Studios where it took kind of longer than I expected, this is all going much quicker than I expected. We'll see how it actually works out. I will let you know. All right. So that's the current state on the hardware. Let's talk about T Tuesday updates, you know, good news, bad news, always the case. Overwhelmingly, my feeling is, you know, it's really been incredible. I made a little bit of a little montage all the way back to the first one. You know, the basic idea. Let's just build out this new idea. Looked at how things used to work, earlier stuff. People showed up and they asked questions. They thought about it. We tried to make, you know, question and answer format. We got a little video bumper to put on the front. I think that needs work. Got a JavaScript simulator just came out of the world. Got my brain re-going on the thing. Tested out this new light sensor using Bunny Huang's ideas. Looked at fork bombs, all of this stuff. Got t-shirts. I got my t-shirt. I mean, the amount of stuff, you know, for 10 weeks and I built a board like that is phenomenal. We have a chat room where there's lots of folks in there. We have a subreddit that, you know, it's warming up. And all of these things are fantastic. And my overwhelming feeling is I started this for two purposes. The one purpose, number one was to keep myself in line and not be able to just sort of slide around and slack off and not actually make progress week by week. And that purpose of keeping me focused week to week has worked really well. And the fact that as many people as there have, all the, you know, people that we know, the questions, all these great things, have wanted to help out and have helped out is wonderful. Like that. On the other hand, there's always possibilities for improvement. There's always issues of some sort. So let's talk about them briefly as well. I have two major issues. I mean, one issue is, well, let's just say this. This is some stats from my YouTube thing. This is the first month from October 10th to November 5th, that the first five episodes, the Y axis is minutes of watch time per day. The different colors or different videos like that. And it's building more minutes each week than the previous week. And that's good. That made me happy as well. I wanted to see. But if we look at it over the full 10 weeks, the picture is somewhat different. The first five weeks, in fact, is where it peaked. As far as the release day minutes of time watching. And, you know, to some degree, that's a misleading statistic because the watching over the rest of the week is so here, you know, in the early days, if people didn't watch it on the first day, they hardly watched it at all. Whereas now there's actually people who are checking in later in the week. Thank you. Like that. Although not every day. So the fact that the peak is going down here is not as bad as it might seem. But in fact, there is essentially a leveling out. The overall watch time of each episode is not going up. It's perhaps going down. Some subscribers have lost a way and so forth, you know. So I offer that in the spirit of transparency. But then on the other hand, I want to say, well, so what? As far as, you know, mission number one of keeping me honest and saying, you know, okay, I need to be able to explain to somebody each week how I'm actually getting closer to the goal. The T Tuesday updates are doing great at that. But there is the second goal, which is to spread the word about indefinite scalability about robust first and about this alternative approach to architecture and an approach to sort of computation and programming almost generally. I mean, they all touch again at the end, but you have to go way, way, way, way and work your way back before they do touch. And so part of the goal of having this weekly update is to help find people that could understand this stuff. I mean, yes, this is very esoteric. This is not just for people who are familiar with computers, not just with people who are programmers, but kind of with people that sort of know all of the above or at the very least are seriously interested in trying to think about alternative ways that we can understand what computation is. And we've found some of those people and that makes me happy and they're contributing and just, you know, asking questions and commenting. That makes me very happy. But I also believe there's an awful lot more people out in the world, even though they're, you know, if they're one in a million, right, there should be an awful lot of people out in the world that just don't know that this is here. So how to actually do better about that, I'm not sure. The second issue that I know is a problem from my point of view is, okay, this feels like a once a week class, you know. And I have to be there. I have to show up. I have to have a lecture. I have to have something to say and I have to say it, you know. The professor can't say the dog ate the homework. You've got to be there and have something to say. But once you achieve that, and certainly 10 weeks ago in the pilot, just the idea of doing that was absolutely terrifying. But now, okay, I get it a little bit more. I can show up. I can tell what progress it is. But there's also this risk of sort of having the least progress per week possible. Okay, I debug the board. I try it out the things. I got the light sensor tested and so forth. That's good enough. And so teaching a class is not just lectures each week. It's the whole class. How long is this class supposed to be? How long is T Tuesday updates supposed to be? Well, on the one hand, who knows? I don't know. It depends how long it's all going to take and we'll keep doing it as long as it seems useful, as long as we can get some progress. But if there isn't a unit of granularity bigger than a week, not just weekly and then forever, then there's the risk of just sort of having this, you know, sort of non-progressive seething. So, my conclusion from all of this is we are now at the 10th episode of T Tuesday Updates by the 20th episode of T Tuesday Updates. We shall have 133 tiles manufactured, interconnected, and running in a grid. And why 133? Because we have 12 tiles surrounding 6, surrounding 1. 12 plus 6 plus 1 is 19. That's a power zone. 19 tiles that will be connected up to a single brick to power them. And we're going to have 6 zones surrounding 1. That's 7 zones. 9 times 7 is 133. Assuming I did the arithmetic right. Okay. 133 tiles in 7 zones. Actually, you know, this is actually a question that I have for y'all. With the 19 tiles in the little cluster counting as a zone, these 7 zone group like this, it needs a name of its own. What should we call a group of 7 zones arranged in 6, around 1? Because that's what we're building. We're building one, what? This is completely terrifying. I said the first time the pilot of doing these updates is terrifying. And it was. This is super terrifying because it means I have to commit to parts. I have to commit to the board layout. I have to make them 100. I have to start laying down serious money. And if I mess it up and the boards don't work or they don't fit in each other or the thing doesn't connect or whatever it is, I don't get my $15,000 back. So the goal for the next week is come up with a schedule for the next 10 weeks that leaves us with 133 tiles running. The next update will be out in a week. Thank you so much for watching.