 Hey folks welcome back. This is official live creation count up. We are t-plus-four and Sinking if we're supposed to be swimming kind of sinking Multicellularity is hard That I have struggled a lot this month Enough so that I'm wondering a little bit about whether I can even really Do this or do this in certainly any kind of finite time, you know I look at that natural evolution the Cambrian explosion Was this you know half a billion years ago when over a relatively small period of time 10 or 20 million years? there was obviously a architectural breakthrough in Multicellularity that allowed Multicellularity to take hold and then go crazy And you know, I love the Cambrian explosion. It's super fascinating But you know as far as what I'm concerned with here about half a billion years ago is not the number that's important What's important for me now? It is that you know, here's the earliest multicellular life fungi possibility about a billion and a half years ago But single-celled life is more like four billion years ago So there's something like two and a half billion years that I'm supposed to be saying I'm gonna do in six months or some kind of analog to it Maybe So can I do this? I don't know I Show you what I got. I made progress, but it was it was tough. It hurt the development process is painful. I mean I had other things that were kind of fun, but Man, uh, so, okay, here we are. All right, let's try looking at This is a this is the state of the art if folks that have been familiar with this This is a whole new kind of cell membrane Looks kind of like a chip now with little lead sticking out instead of all that sort of brown goo intercellular goo The intercellular goo is gone. I was trying to get more precision For how they register with each other and be able to send more information Not more information, but more accurate information through and you know this at the moment I've lost a whole bunch of other behaviors So this thing doesn't know how to change direction or lose any fins or so on and so forth. That was the big success and So and again, I don't know why I'm having so much trouble with my buttons. I did not look into them And okay, so where were we supposed to be? Today we were supposed to have clean living fish, no Well, I mean, they're kind of clean only because they don't do much stuff second species started no way not even close and I am increasingly resigned not to say happy to think about it being And the multi-cellular menagerie will be All about the fish if there's any little Happy mushroom or something that would be great Second half plan. I have some ideas more big fun, you know when I put aside the development stuff, you know April was good in a lot of ways stuff with friends and family was good and Stuff with other things were good. So let's just pushing to it Good and that Okay robustness demo. Oh, yeah, and so this happened I mean, so the previous fish with the intercellular goo that have all kinds of terrible problem They've been running on the grid for all of April basically and here's a little sample of it Oops, and if I could just try to figure out how to get the correct button here. All right. Here we go And folks that were watching the live stream, they were seeing the April run on the thing look at that Lotus 2 Power supply failed. It didn't fail hard to zero It failed to like six volts when it was supposed to be 12 volts and all of these tiles up in Lotus 2 just went nuts I bought a new I sprung for a mean well as I think oh, this is nice I soldered the connections on and I put the wrong gender on So, okay, but but now it is now we're getting 11 and a half volts and so forth and you know To put to find a point on it, but this thing survived. Ah, it Lotus 2 crashed it didn't lose itself. It kept on going. It was down for a day or more Actually several days because I ordered new power supply and so on and eventually I stuck a new power supply in and It actually came back up I don't want to claim anything for it because I think it was luck in most purposes, but hey, that's what was supposed to happen The tile, you know grid so big Matrix so big that parts of it are always failing. You know like that. So anyway All right. So now yikes So the schedule Supposed to be this is almost identical to last month's schedule, which is as far as the progress Which is a little scary. I've called increased code density. I've greened that out because I think That is not a problem for what we're currently doing Having that we can get enough, you know In fact, I've stuck a few extra instructions in and move them around and stuff like that And that has not been a problem it the existing machinery the loader and all that stuff takes care of it And it's good Coordinated movement while you saw it's hard there and junctions and intercellular signal were supposed to be the topic for this month And in fact, that is where I ended up going and we'll talk about it Right now, I guess. So yeah, so membranes 2.0 You know part of this whole project of trying to build a whole new computational stack Is one of the hardest parts of it is when you actually succeed at some given level And now you're supposed to go and build on top of that Just like when I was in grad school and you know, the older grad students when I was in graduate school They were working on parallel computers that their faculty and hardware Students were building new parallel computers or exploring it. It was the new stuff And there were other graduate students that were doing parallel software And what all of us little munchkins learned the new grad students learned from this Is that you don't let your thesis depend on somebody else's thesis Because the software parallel soft guys were getting really good Folks were really getting screwed by this whole thing because the hardware stuff wasn't working They were changing it out from underneath them and it was terrible. And that's what I've done to myself here, right? I built our own hardware I got our own programming language our own compiler and then this giant mountain of luam code that just to get the Diamond matrix going and so on. So, you know, yeah, okay Maybe I was a little glib thing a little bit easier. And as I said last time The the design, you know, and this is again the same thing, you know, my thesis depending on my previous thesis It currently takes eight minutes to build the software from standing start because it builds from a standing start each time It doesn't have any incremental compilation facility and anybody who's a programmer knows that the amount of time it takes To make a tiny change and then try again has a tremendous impact on how much productivity you're going to end up getting And an eight minute loop is tough. It's really tough Um, so Nonetheless Where we got to last time was the whole idea of sending Signals from a controller near the center of the diamond out to the edge to say, you know, what is out there? Do you see other diamonds? That you know, if so, can you get me some information like about how big they are and exactly where they are and Send that back to the center. That's the fish fin and the fish pod diamond controller And we sort of got it working But it was working so sloppy and terrible that the fins kept flying off And I became convinced that what we really needed Was more accurate. It would be nice to have more timely But the speed of light the speed of light on the grid that you can only go so fast Uh, so did a whole new implementation Um, and I don't know that should be Chords not cores So I did something I didn't want to do The the diamonds the the hc3 Atoms that make up a diamond now have an actual absolute coordinate within relative to the diamond There's still no absolute absolute anything, but I had been avoiding putting an absolute chords in because it felt like Going with the the old school thinking where I'm just lay everything out flat And then I'll have all the flexibility I need as opposed to thinking about what do you really need now? Well, I broke down and I made hc point the hard sell point and it's an absolute coordinate and I used it Some I re-implemented the membrane and in particular Instead of saying we're going to take all our information and ship it to the center And then it's going to decide oh, we've got an edge over there and an edge over there. We should do this Instead we're going to make these junctions. We're going to distribute the Information back to the edges or somewhere in the middle of the diamond Saying I want to have a fish fin junction on this side and over on the other one It says I want to have a fish bod junction on this side And the fish fin and the fish fish bod are supposed to have enough Information enough knowledge to know is this coordinate where I want it to be how how well are we lined up? So instead of sending the entire Position and orientation of neighboring edges back to the center the junction reduces that information itself It boils it down to a diamond stance that says for each of north south east west How much it likes or dislikes movement in that direction? So it it reduces it all and just passes up I would like north or west please and I would really like dislike south or east Or whatever it is and the cool thing that came out about that was that Once I had it spread out once I had the information spread out four things representing liking the directions four numbers representing disliking the directions Now it was possible to merge diamond stance signals coming from multiple edges and multiple atoms along an edge In the in the mesh in the grid without having they don't have to all go and reach the Route and instead as soon as they see another one of their own kind they do a merge, you know You want south? Oh, I want south too. You you want no east. Oh, well, I want no west And so you put it all together and it gets to the top and the and the thing at the center Does very little beyond saying what's what possibilities do we still have now that there's an opportunity to move? And that's what we just saw in the antenna. Oh, yeah, there's I didn't even mention antennas here So instead of the intercellular goo, we have these probes that stick out that get exact information It's always outdated by the time it gets to the edge And from the edge to the junction and the junction all the way to the center It's always outdated, but now it starts out actually site perfect that from this site That guy's center is minus 26 plus 13 whatever it happens to be uh, uh, all right, so I Running a little bit long. I'm gonna skip the demo now. I will come back around maybe during the uh Livestream afterwards and play with it some more uh Okay, big writing fun. So, uh, we had the submission to the artificial life 2023 conference waiting for results through most of April and and the results are the paper got accepted for a normal presentation And and that's as good as it gets pretty much through this channel of uh, submitting a paper. So that was cool and I have to say I had quite a bit of fun actually writing it You know, uh A programmable replicator for an indefinitely scalable machine We demonstrate a programmable mobile self-replicator for a non-determinist. So there's our there's our jargon and so forth Write down to it. Um, and I felt confident About, you know, just say it. I don't I don't need to puff it up. Just say exactly what happened 20 day case study Here it is so on and so forth did a Big essentially a whole poster version for figure one this is in the negative and so forth and there's one and they're reproduced and here's like a whole bunch of them or it all reproduced out, um and so on and I kind of Let myself go a little bit and you know Uh, one's supposed to be careful. So in the conclusion section, for example, I said Talking about uh, in particular, we may admire but we'll not Emulate heroic attempts to do the least possible thing use the least number of bits the least number of things Because what we really want to do is be useful not to be minimal It is possible to do all computing with just NAND gates, but real computers don't Indefinite scalability is hard enough. We don't need to also tie one hand behind our back So that was being able to talk about a little hubris and everything and indeed Uh reviewer number one really did not like The paper they suggested a rejection. They didn't suggest a full-throated rejection But they had lots of problems and kept telling me I needed to do better science and so forth I asked for that. I'm okay with it. I will not be actually traveling for this This is going to be a hybrid thing. So we'll have to deal with it but step by step And all right, and and we're at a time, but you know I want to step back from even monthly Updates at the moment. So I'm saying two more updates. Maybe three months apart something like that and instead I want to work on Coming out with a video on the Dave Ackley channel a more approachable by more folks kind of video And I'd be interested to hear your thoughts about what kind of topics might be fun might be into it And publishing something we'll see And that is it Thank you so much for coming and I hope to see you in a month I hope to see a little better multi-cellularity in a month as well