 Hey folks, it's t-minus-8 and counting of the artificial life creation challenge a lots of stuff to look at here Let's start with this So long or so membrane started to work one thing I want to call your attention to is Occasionally so you see these orange flashes happening in various places. That's where Contact information is being signaled from the edge of the grid in to the center and sometimes I think there was one right here especially from the very yeah, there was one you see things that aren't actually real That it's just a phantom of the signaling system. I think of the thing being itchy We'll watch this just for a little bit more I Let this simulation run For two million apps two million average events per site across the whole grid because I wanted to get a lot of Interactions and make sure that these things weren't still smashing each it's fusing themselves together getting stuck to each other But you know when it starts going faster like this, I mean it really starts looking very I mean to me You know pretty organic except for the fact that it's a diamond. You could sort of imagine Like looking at some pond scum and a microscope and seeing some of these things or something Anyway, so this is the main the main progress for this week but Hello, hi It's t Tuesday 31 27. Here's where the goals for this time did pretty well on most of them Did not do loop operations that will continue over to next time that doesn't put us behind on the overall I'm not even going to talk about the outline. I'm just going to do it So here's our overall outline spreadsheet that I made in July. That's actually still holding together pretty well And here we are at t-minus a hc3 sequencer that's hard cell three the diamond grids Sequencer and borders what we have today is borders. We'll have next time is The hc3 sequencer and and hopefully the rest of the countdown. We will continue to count Moving on so why do we care so much about cell borders? Well once again here is the overall global view of what we're trying to do going from a mother to grow Reorganized copy the genetic information Split and have a mother and a daughter that can do the same thing and right there at the very beginning We're talking about a discrete organism that has an edge that I drove drove drew a wall around This is the cell wall. This is what we're trying to get done with the border around the diamond matrix So let's take a quick minute To just look at the simulator All right, I've got it set up here so that maybe we can sort of see what's going on So where the heck is all right here? We are so hc3 that is the hard cell three these so those are the matrix You can see 3 3 3 they're each of them three apart 1 2 3 and so forth, but we got these MBs next to it those are the Membrane bases and what they try to do at the edge of the hard cell 3 matrix We don't have normally we have a complete two little two by two Units all over the place, but at the edge we have three Hc3s and no fourth one because that would be off the end So I call those broken pockets and the MB the membrane base they colonize the broken pockets that are supposed to be at the edge of the thing and They're looking out to see stuff that's going on and they're also looking out to see ms's which are the membrane Sensors those are the distant early warning Systems that look out further and in fact what we're seeing right here In this this is a replay. I've backed up the events This is about to have in its event just at the edge of its event window Can you see that we've got the this green outline showing how much an individual event can see and it's fundamental to everything About the movable feast is to understand the event window that when one atom that's having an event That's what they can see and alter so in this case This guy is a pat can just barely see a membrane sensor from a different cell And it can tell it's a different cell because this the sensor has a Open dir that says which direction it's looking in versus which direction it thinks is its own cell and these differ So when we go ahead and have an event here Boom now this event is starting to signal It's signaling that there's a contact directly to the west and if we let this thing keep going that'll keep spreading on through But that's the idea. I mean so if we look at this Here we've got a this hc3 is one two three four five six seven eight nine ten eleven It's twelve sites away from the nearest hc3 of the other cell But yet they have detection they have contact and in fact the other cell have already seen what's going on so The diamond cell membrane that goes around the outside. It does all of these things for us I mean number one it you know It helps spatially isolate the inside of the set diamond cell from the outside to let each guy know I mean the the hc3 atoms itself I mean they don't actually they know which direction they're going based on the neighbors that look downhill We've talked about that before but if another Hc3 diamond grid comes too close they can get confused they can think do I belong here or do I belong there and Having the membranes around it having the layered defense in depth the final point there This is defense in depth goes back. It's a military strategy of having instead of having one Absolutely firm line that you're gonna hold forever. You have fallbacks and layered defenses got taken over by Computer security talks about defense in depth Which is an easier said than done when you're talking about computer security within a single computer defense in depth Doesn't really work because as soon as you have a breach of the program RAM the random access memory You can take over the whole RAM like that. That's kind of the point of random access memory But in the movable feast in cellular automata, we have actual spatial locations And we can actually make certain amount of progress by saying I control this much space and in furthermore Don't even come within, you know, and like 12 Sights of me or I'm gonna notice and I'm gonna react now They don't have a lot of ability to react except to run but that's really important So provide early warnings of outside threats and didn't talk about it But the membrane base and the membrane sensor are both sacrificial We can just roll over if the thing does get too close those things will get put out of the way And the membrane base not only reacts by copying what the membrane sensor says They also react if they don't see any membrane sensor And in fact, that's what we saw at the very beginning of the opening clip that when the cells were first coming out There was all kinds of orange signaling going on and that's because the membrane sensor layer hadn't grown yet So the whole thing was feeling very itchy as it was scabbing up until it got running. All right. I did it I Took the you know, Cold War the the due line the distant early warning line that was supposed to be like Watching Russia so that we'd see the missiles coming in I understand from Wikipedia that it's not the due line anymore They've replaced it, but you know, I'm a boomer And the idea is we have four levels of contact in a deaf con. I'm mixing my metaphors Open meaning I don't see anything sight meaning the sensors have seen something, you know, just barely on the horizon Contact meaning that we've seen something much closer that the membrane base I've seen not just with the membrane sensor cells But they see some problems directly or the hc3 tail the hc3 edge is seeing something Which is even worse and then the worst case is damage when then we have actually an hc3 in the interior of the cell Is seeing something that's supposed to be there. That's not completely implemented yet But that's the worst case so the root when it's making a decision about what to do will have a choice about well Okay, I can see some sightings, but I'd rather get away from the damage or whatever and that's all it seems working So this is the title card Cell defense in depth in depth meaning layers. We here we have our two giant Political powers cell one and cell two they've come in sighting distance of each other and the reports are now flowing up Back to the root for decisions to be made All right moving on debasalification we you know, this was the Clip I showed last time the screenshot that had you know, you know, you know, all right There's a diamond grid, but because the t2 tiles have these These bezels these frames around them. There's connectors for the intertile connectors There's a grid power light over here is the tile power light all of this stuff That means the LCD in the middle is quite distant from the adjacent LCD of the tiles. And so I worked on that Let's take a look at this Looks like a brick wall in the dark with stuff being projected on it, but the stuff that's coming out of the wall I didn't I didn't trim it any closer because there's enough variation from position to position This is the best I dared to go to the color rendition is also not great A lot of that is due to the cheapness of the LCD panels. I'm using May be able to fix that. That's already trying to do some color correction You might have seen some some itching going on there on the lower left One This is still running. Hopefully we'll have a longer one with stuff bouncing around a lot later All right, and so that's all using this FFM peg program Which I think stands for fucking fast and peg It doesn't but I wish it could I mean and this is this is stressing it pretty hard You know each one of those images is a separate image crop which are then overlaid one at a time on the output thing And you know the bottom line for me is it takes over an hour to process a day's worth of data Which is kind of a long time, but on the other hand is less than a day so we can get this working So that's that Finally You know thanks so much for taking a look at these videos or this video being here now Wherever you have managed to find this, you know the t2 tile project is still super duper tiny But I think the artificial life creation countdown is helping it's drawing in a few new people YouTube says that the previous update to t-minus nine is reaching a wider audience And so I thought maybe in the very end here I would spend a minute trying to just you know Reset a little bit of the backstory and you know how I got to be doing this and you know just you know What's going on here now? You know I could say you know I am a emeritus professor of computer science and I have done blah blah blah and so forth But I thought instead I would take one minute and show you this Hi I'm Dave Ackley like to present the results of some work How can an organism learn during its lifetime given only its own death as feedback so organisms with good evaluation Genes will tend to learn good things and will proliferate a source of reinforcement signals Here's the data we obtained. This is an overview of the world This diagram depicts the array of inputs that each agent receives at about 600 time steps He was quite lucky to survive an extended encounter with a carnivore The result of the learning algorithm in this interaction was such that it corrected the genetic defect in the action network He was born shortly thereafter. He produced the first of his four children Of course if you took a stable population of natural guys like this and stuck in a couple of those hand-designed guys They'd take over in a couple of generations wipe these guys out Guess that's life Okay 30 years ago. All right, and that is it So the goals for next time we will be back on August 30th is to get loop operations demo I mean the demo here means, you know, we want to be real It's not just talking about it not just thinking about it but getting something working Also want a hard cell three sequencer demo. We don't even really know what that is yet We shall find out and also as always have some fun Um again, uh, I will pause here at the end and get the render starting I will then stay will still be going live in the live stream until the top of the hour in the meantime Thanks so much for checking this all out. I hope you're doing all right. I hope to see you next time