 Computers keep changing the world, but their power and safety is limited by their rigid design. The T2 tile project works for bigger and safer computing using living systems principles. Follow our progress here on T Tuesday Updates. This is the 41st T Tuesday Update. Let's get into it. Last time was the big 40th episode. We were supposed to have a ring lotus running on the wall. 100 tiles running on the wall didn't get close. Since then, I've done propaganda stuff, talked about it. Some hardware stuff and some software stuff. Coming up, there's going to be one more T Tuesday Update next week before we head off to the Artificial Life 2019 conference in Newcastle-Pontine in the UK. So, pretty much whatever we're going to take to the ALIVE conference is going to be happening this week and the early part of next week. So, that's what's coming up. Let's see where we got this week. And the Propaganda Division. Yeah, I got invited to this thing. I'm not sure I mentioned it. There's a podcast by a guy named Steve Krause, I guess, called The Future of Coding. It has this slack room in it as well that I joined up that I got invited to. Had a bunch of good conversations there. Other folks from the T2 tile project, Andrew Walpole and AJ Zaff have shown up in The Future of Coding as well and are starting to chat with folks. And it's really great. It's this conversation I had with Andreas S. It was really good. It was the same kind of thing in the sense of the podcast with Tom Barbel at the other week where people are just asking what are sort of the obvious questions coming from a more or less traditional computing background. And so, you know, so Andreas says, you know, if you're going to send a message from China to Africa, you have to go tile to tile, tile the whole way and the answer is, yeah, absolutely you do. And there's a reason for it and so on. It makes totally makes sense. Andreas says quite a bit of work, quite a paradigm shift. Yeah, indeed. So that felt good. I mean, it all takes time out of the day. But this seems like there's been sort of an increasing amount of interest and awareness gradually getting out, cast the bread upon the water. And so some of that's coming back and that seems like a worthy use of time. So that happened. Various other folks. And again, the whole point is these are all public conversations. So other folks, someone was chipping in with giving a little light bulb sort of aha thing, you know, the robust first slogan delay universality, which was from last week's glass sandwich rant. So it was just on my mind and naturally it popped up now. Also, the there's been some renovations in the getter. Instead of everybody sitting in the T2 tile lobby the whole time that we now have a profusion of rooms, one about splat, one about 3D printing, software department and so forth. And conversations are going on in a bunch of these. And it's very exciting. It's like home improvements. I mean, the flip side of it, of course, is now the lobby. It's like, bam, everybody vanished from the lobby and nothing's happened there in most of the week. It's fine. The, you know, growing adjustments is a great problem to have. And this isn't even a problem. So that's in the propaganda division. Oh, and one other thing, some statistics from the T2 tile project. We got our first dislike. So these statistics are from since last October, which is essentially when the channel began. 959 likes, one dislike. See if you can guess which one, which video got it. I always sort of feel good when I finally get a dislike on my other channel and the Dave Ackler channel. It took a long time too. I'm not sure. Certainly hundreds and hundreds of likes before the first dislike. And to me, the fact that I finally get dislikes suggests that I'm actually getting a little traction out there and maybe like YouTube is being willing to like put my videos up next to other things so that somebody might click on it who isn't sort of pre-sold, isn't already in the bag, isn't preaching to the converted. And therefore they're going to look at it and eventually someone's going to say, this isn't what I was interested in. And boom, there we go. So that's what's happened here. When I was looking at this, there was one other weird thing on my likes and dislikes. January, February, March on April 5th. I had negative 26 likes. I don't know exactly what that's about. I assume it's about YouTube scrubbing fake people or whatever it is, what they do. So that's a little bit more in the pushing stuff out in the world. So did you make your bet about which video got the dislike? Dick Viana makes me wonder if perhaps somebody from the sort of startup entrepreneurial business community managed to find this thing. And was, you know, taken it back to say the least that it was no less in the traditional sort of business exuberance model. And maybe that's where the dislike came from. Who knows? Somebody may have fell asleep on the keyboard. So that's the propaganda division. Moving on. Hardware. Right. We've got two topics. The power injector production. Now we talked about this. We've got the regular intertile connectors. These share power and data. We've got the ones that just share data, but don't share power so that we can divide the whole grid into this is powered by one power supply. This is powered by another supply. This is powered by yet another power supply. We still have to get the power in there. And that's what the PIs, the power injectors are for. Ordered some more boards from Osh Park. They came this past week. And so I built, here they are. I got a 21 or something like that because they sell in groups of three. And so this is a bad picture, but this basically looks like all the other connectors as far as the circuit board goes, except it's got these two holes in the middle, which is where the big, pretty heavy cables from the power supply go in and then distribute out to the both sides. So the connector actually goes between two and does its job that way. So here it is with the wire. This is under the microscope where I've managed to get the red and the black through the holes and got them soldered down. I then clipped off the excess on the bottom because we've got a plastic strap that needs to go right across that stuff between these two U-shaped holes to become part of the handle for removing the thing. So I wanted to get the solders down as far as I could. I stuck it in the third hand so that it could just be sort of all pushed down on the bench so that I could solder these things a little easier. Got them all soldered up. It's got flux rosin all over the place, but it cleaned up pretty well. And so now the goal is to get one of the handles out of the tabs. And here are the pieces. It's a four-piece unit to make a power injector handle. And step one is to get this strap up through the thing, coming up through the bottom. And that's when no slight problem. There's no way that this thing was going to go through that gap. It's some combination of printing. Maybe I was printing at a slightly thicker layer height so it rounded off differently. But I think mostly it's the Osh Park routing is a little bit narrower than what I saw with the other ones. Not sure. But I went at it with a nail file and filed the board back. I also filed the plastic strap back a little bit. And eventually it was a squeeze. I managed to get it through. And so the idea is that once you get it through you push it up there. And it provides two hooks for the next piece, the shell of the piece to fit onto. And the shell has got a nice little snake in it that provides stress relief for the cable. So you fit the cable through the S-curve. Like, there we go. And then we've got a cover that goes on. Well, I took the opportunity since I was assembling this to put a piece of shrink tubing at the point where the cable comes out of the handle. So give it a little bit more strength there. So there. At the top it's got little buttons and indentations to help push the latches from the strap coming up as well as the S-curve down. And there it is. And then finally the fourth piece comes, which is this cover that slides over the whole thing and slides on down. And then right here that little bump right there. That's a new addition that I put on just in the most recent revision of this thing to help latch this cover thing once it's slid down so it doesn't come loose and come off. And there it is. So a completed guy. And that all seemed to work fine. It's actually running on powering the grid that I've got in front of me right now. So I went back to the printer to print up a bunch more shells. There's actually a little bit of foreboding here. If you're a 3D printing person, you see these rims here and here are kind of pulling up a little bit from the bed. It turns out not to affect the performance of these things. I went ahead and used them but it was foreshadowing. And here they were. So this is after I finally got a little bit of a rhythm going. The first one took most of an hour but the last couple I did half a dozen took maybe 15 minutes each. And I hooked them all up with tabs and here they are. So we've got a half a dozen of these things now ready to go. Two of them deployed but it's in pretty good shape. Got to do maybe two more sets like this and that will be enough for the not just the ring lotus but all of the tiles that we managed to put together which we still don't know how much it's going to be but we hope it's going to be like over 150. Right, so there were the pieces that I mounted on here that you've already seen and same thing as before. But another problem. You saw these before. These were the bases for the these things for the feet on the T2 tiles to go in. It was a prototype for seeing how we could maybe mount these things onto acrylic base or something like that to make it easier to handle. But as I was warned by Andrew Walpole my resident helped me out with a 3D printing guy these female feet that were supposed to be joined up with the male feet that are now on the tiles are printed vertically which means the layers are going this way which means they break quite easily across the layers and you can see this happening already. So that one's busted off. That one's busted off. I think there's a piece busted off for this one and so forth. Now we can handle a fair number of little pieces of it coming off and still hold the thing together but it's really a problem. Andrew and Joby sort of got together with the idea that I could keep using these round feet which are nice because you screw them into the holes and they match up no matter how you do it because it's round but instead of printing these things vertically on the printer like this print them horizontally so the open thing goes this way but then it's got to lay on the bed in order to work so it needs at least one flat side in order to lay on the bed so I started trying to design stuff based on suggestions from Andrew and Joby about how to make it sort of a square outside with a circular inside I started to realize that actually programming this stuff directly open-scad what I used for this 3D design is a programming language and you say things like circle, cube, ellipse scale, stretch, stuff like that which is very clean and you know where you are and I'm a programmer so it feels alright but it feels quite limiting and finally this past week I had a tool breakthrough I learned how to actually import files from Inkscape the open-source vector drawing program which is sort of like an illustrator or something if you're coming from closed source land and get it to be millimeter accurate so it could be exactly what it is so I designed this thing in Inkscape and brought it into open-scad and extruded it into 3D and modified it and so forth and it's a new workflow for me, I like it so and here is an example it's square but the end down here is where the male foot would go in and it sort of clips on and it gets it the idea was that we had another clip sort of like a clothesline clothespin that would go through a sheet of acrylic and my goal was to be able to drill round holes in the acrylic and put this thing in and pop it in so that it would then be rigid and round holes are convenient because I can drill round holes with a drill press so these things you can sort of see it that these things which are meant to be the clips that go through the board are trying to be rounded off this caused all kinds of trouble for me in part because I'm having problems with this fulfillment pulling away from the bed something I don't completely understand because I'm not really that great at 3D printing here's another view of it with one of the male feet in it and it's pretty satisfying it snaps in and so forth and most of them at least when they print well or they've been broken in a little bit they'll spin fine I guess I don't have an example here at the moment but they're actually click and then it's loose so it's not actually stressing the female side when it goes pretty nice but I was having more and more problems coming off of the bed and you know so as I started to think about redesigning this thing and I said well you know maybe maybe I don't need to do a round hole thing maybe I could do a square hole and I could hire someone to do acrylic for me or whatever but I realized that I could make the thing sort of wilder shapes I mean I was sort of doing things that were basically straight lines and roundish shapes even though I was in an inkscape where I have full on splines and so forth so I started cutting loose and it was really quite a bit of fun so here's the next version that now the part that was going to go through the acrylic is no longer trying to be rounded it's expecting to be going into a square hole and it's got a nice swirl in it there's another angle on it here's a pair of them coming off the printer don't they look nice so I enjoyed that having that additional capability to do you know essentially not really hand drawn but much more flexible shapes and I know people coming from Fusion 360 are going oh my god this is the first time you've done this step by step open source doing what I can I finally realized that just like these guys where I was always ending up with pairs of holes because the two tiles that want to be joined are always have pairs of holes next to them but I could do it as a single piece instead of having two separate ones I could have a unit of two and then I could have that just attached and that gave me a lot more room to use the legs in the final version I've been getting rid of the legs entirely and thinking about gluing things down again I'll talk about that a little bit when I come around to dry it I've got an idea in a way that's sort of not too much work and doable and maybe will not require getting the measurements to exactly up front and then spending a bunch of money for someone to laser cut acrylic we'll see so there's an example of one coming one of the pairs coming off here's a cutaway version that's not actually cut away it's just in the middle of being printed where you can see the this part right here is the main part of the snap and there's a secondary one but really this is the one that counts and so forth again having issues with pulling away from the print bed so I've done a bunch more and I've sort of added extra stuff to help stick it down like that it's still pulling away I'm not entirely sure what's going on but at this point most of the pulling away doesn't actually affect the functionality it seems okay so I'm going with it so Andrew will also convince me to buy a new 3D printer which I did not realize I was going to have to assemble this far that's an adventure for the future that's it for hardware and so we're almost out of time so I'm really not going to be able to tell the software story so I'm just going to cut to the chase which is the goal is to get intertile events happening where an atom running on the outside does something make some changes and those changes appear on an adjacent tile and it can go from there and in order to have an intertile event we have to do two things number one we have to be able to negotiate whose turn it is who gets to have an event here because we try hard as possible to have the effects of a single event be consistent with what the person that made the event expected so that requires locks and then they're sending cash updates to notify what happened the stuff that we were talking about a little bit before was about the cash updates this time I was working on the locking the locking code was very old I refreshed my brain and I was like oh my god it's like taking the lock is the innermost loop this is meant to be as fast as we can possibly make it even though it's going between one tile and one or two or three other tiles and the code I had was incredibly slow I got it down to 44 milliseconds each which is still incredibly slow but it was faster than the 300 milliseconds that it was to begin with and I started a saga of trying to figure out what it was so when I got it down to 6 megahertz 6 milliseconds per lock grab attempt it wasn't actually working the locks weren't necessarily completely grabbed by the time I checked and so on I worked through a bunch of stuff and eventually I got it down to something like where is it here I got it complete 148 microseconds to grab the locks that's not bad it'll get worse from there when you feel like there's a chance that's not pretty bad it's going out of one tile into the kernel in another tile it's responding, coming back to the first tile doing a bunch of stuff like that all in 150 microseconds people that do serious optimization this is terrible they could be getting down to double digits in microseconds if not better but we'll take it alright, at a time the next update will be out in a week it always takes longer but we're making progress have a good week