August 2016
8/11/16 Finished the core code for serialization, importing and modding, but I’ll have to hang back on scripting a bit longer. I got the August build out early (or maybe it was the July build out late), and there haven’t been any major problems with that. So at long last I can go back to the brain and move on past the fixed throg-hunting behavior to more varied and deliberate learned actions. I need to consider a rewrite first, though. I had some ideas about associative memory and it may be that a bunch of previously disparate things could now come together into a single concept. I can’t afford to do this for purely academic reasons but if it works out and I do it soon then it might pay off in the next stage of development. It’s a scary thought, though. It would involve some big changes to the genome. Hopefully changes that can be automated. Worth a couple of days’ thought.
8/31/16 Dammit! I’m working on the design for the more executive functions of the brain and the way that they need to interact with short-term memory is much more involved than I’d been assuming. I REALLY thought I’d got it all worked out today, but then a single exception to the rule screwed it all up again. It’s just the usual highs and lows of inventing, I guess. Usually there’s euphoria when inspiration strikes, followed by “uh-oh”, followed by deep despondency as it all unravels, followed eventually by a kind of synthesis where something useful comes out of it. Hopefully that phase will come tomorrow. On the bright side, I’m getting a lot of exercise, pacing around Arizona in deep thought.
September 2016
9/14/16 What this week has been like so far:
void Main()
{
Main();
}
9/27/16 Finished a major rewrite of the mouth, muscle sequencing, action signaling and suchlike. Not what I INTENDED to be doing right now, but it needed an overhaul.
9/28/16 A slow and tiring day, but one of those where I feel like biology is being revealed to me. The various drives and emotions need to be grouped and handled in ways that turn out to be reminiscent of the amygdala versus the hypothalamus. Stuff like that. It’s surprising how little clarity and consistency there is about the categorisation of emotions in the psychological literature, but actually trying to implement them is proving quite revealing.
October 2016
10/13/16 Boy, is this hard going at the moment. I can tell it’s autumn - I feel the need to hibernate! I figured out how I wanted emotions to work and thought I’d start with some really trivial drives, like feeling uncomfortable because we’re hot and sticky or cold and wet, from which we can learn to seek shade and shelter. I thought this would be a quick and easy test for the basic drive system before I set out on the harder stuff like the emotional and autonomic responses to threat. But then I realized that to test any of this the creatures need to be able to remember where there is shelter and navigate over potentially long distances to reach it. Except it’s no good just going to the general area where we know there’s shelter; we have to be able to find it and shelter under it. So I have to finish the high-level nav maps and add the ability to recognize the utility of different sub-locations. All of which I knew I needed, but I didn’t realize I needed it right now. But then this got me thinking about a new way to handle high-level nav itself, which, despite being one of the first ideas I’d worked on, still wasn’t working very well. I like what I’ve come up with - it solves several problems at once. But of course I have to build it before I can test that “really simple” drive-related behavior! So much for five-minute tasks…
10/28/16 At last some movement in the right direction! The learning method I’d used for recognizing things (objects, creatures, weather, landmarks, etc.) was never very good - too slow to learn; too unstable in its representation; not very self-organizing - but finally I’ve come up with a simple mechanism that seems to work well enough. So that means I can soon test all the other changes I’ve made, which means I can get on and test high-level nav, which in turn means I can test the simpler kinds of drive-based learning. I just need to change the way short-term memory works to fit all the new features, and add some executive maps and it’ll start to come together. He says confidently…
November 2016
11/20/16 I’ve had one of those “this is how you should have done it in the first place” periods. I guess I’ll feel good about it one day, but right now it feels like I’ve written a huge epic poem in English and then suddenly realized it would have been better written in Anglo-Saxon. Except now none of the puns will work.
December 2016
12/1/16 This is so complex and massively parallel that I’ve had to write a simulation of my simulator. How meta is that???
And that's literally it.
Author
Topic: Grandroids: Secret Supporter Stuff! (Read 65841 times)