Ape Reality #52

On Noble Ape Philosophic
September 28th, 2006

Transcribed by Jason Howard.

This is a transcript of a podcast by Tom Barbalet.

This evening I'm going to talk about Noble Ape Philosophic.

I've spent a bit of time through these podcasts talking about various bits of documentation associated with the development, but Noble Ape Philosophic is the concise and relatively helpful view, with regards to the background principles in the Noble Ape simulation development.

What motivated the Noble Ape simulation development? You go to the original manuals for that. But in terms of the kind of meat and veg of the simulation development -- the ideas and the methodology -- Noble Ape Philosophic provides that information.

An interesting sideline story: I was chatting with Pedro a few days ago, and he had mentioned a discussion he'd had with a fellow online about what this Noble Ape development was all about. In fact, this conversation was the inspiration for this podcast more than anything. Pedro had noted that the fellow had said that he read Noble Ape Philosophic, but having read that, he wasn't sure how the particular agents were controlled. My response back to Pedro was, ''Well, he didn't look at the current manuals for the simulation?'' Because that obviously explains through ApeScript how the agents can be controlled, although the hard-coded agent control is still within the simulation itself.

So, having developed the simulation for more than 10 years now, my view has always been that documentation is just a complete black hole. I wrote the documentation detailed initially and only a certain group of users read it. I wrote a concise version of the documentation, and again, a certain group of users wanted more. I put more information online, particularly programming information, per request. A lot of people read it but I got no feedback. And now the situation is that the documentation associated with Noble Ape is spread out over a few different documents and now a suite of podcasts.

The feedback I gave to Pedro was, ''I can not be changing the development in some regard based on people that aren't interested in using the simulation.'' I've come to this realization in many years of development that people say, ''I read something and I'm not going to look at it.'' You know, ''I'm not going to look any further.'' There's no way you can motivate people to look any further. The development that you can do only relates to people who are already interested or already motivated or already using the software. Moving people from that initial reading to actually downloading the software can be quite ethereal. There are a number of reasons people use software, and without getting too deeply into that idea, it's very difficult to say, ''Here's someone who hasn't done anything. They've looked at one document and walked away. If that document were better or had more information what would they do?'' Well, the conclusion is that they'd walk away, so what more would motivate them? Well, they didn't look at any other documentation, and it's almost impossible to lay out some kind of causal argument about how things are done. So, as an open source developer, I put back to Pedro that I'm investing my time currently in podcasts, funnily enough -- another means of evangelizing the Noble Ape development and getting people interested in artificial life in a kind of broader scheme of things.

To this end, Gerald de Jong has produced a splendid podcast. Another splendid podcast from Gerald. If you're not on the DarwinAtHome podcast I thoroughly recommend that you subscribe. Darwinathome.org will give you the method of subscription. I think they're available through iTunes as well. I've included a number of links through to his site. I will do so again in these show notes.

I am talking about Noble Ape Philosophic. The history of this document was that I wrote the original manuals, and the original manuals were so completely overwhelming, and so completely scattershot, in some regard, that it was very difficult to say to people, ''This is why the Noble Ape simulation is the way it is.''

You also need to appreciate, because I didn't take a single idea -- because I wasn't talking Daniel Dennett or Richard Dawkins or Sartre or Foucault or any philosophic work or philosophical theorist and say, ''This simulation is based on this philosopher's work and this philosopher alone'' -- I had to create a fundamental document that was easily accessible that people would read relatively quickly that would give them the essence of the philosophical motivation and the practical motivation behind the Noble Ape development.

Fundamentally the development was based on this idea that you take a view of the external world and you form a view internally of this. Then, through a variety of mechanisms, you create a society based on that. I think these components came together in the Noble Ape development relatively well.

I'll go through the various components of Noble Ape Philosophic and comment on them sporadically, but the last link between language and the creation of society -- as you know, from listening to these podcasts -- is a relatively contemporary development in Noble Ape. So when I come to that I'll talk a little bit more on that.

Really the other purpose of Noble Ape Philosophic was for people not familiar with philosophy or what the mechanisms were -- particularly computer science folk -- I had to give a kind of philosophy 101 in Noble Ape Philosophic as well.

I started by talking about the difference between internal and external. Noble Ape explicitly simulates the external stuff but allows the internal stuff to evolve. This is an important distinction, particularly when you talk about ideas of space and time, because these both have external realities and internal perceptions of space and time.

From the external reality that was a defined space and obviously a defined time, vision -- and really here this is all the senses, not just vision, but vision is used as a descriptive sense -- takes in this information and forms what I call an identity which is really, in some regard, the brain simulation. Possibly additional components of the brain simulation. Things like the neurons and the interaction within the brain simulation form the identity.

The identity is fluid. It's not a fixed thing. It's something that evolves over time as the vision information is passed in. This develops a very abstract idea of what space and time is internally. The other core components to the identity are fear and desire. These are actually defined mathematically. The fear is an instantaneous reactive component which is represented, basically, in the changes over time of individual parts of the brain simulation. Desire is a long-term evolution. Kind of putting points in time. You know, ''I want to buy this car in the future,'' or ''I want to move to such and such a location,'' or ''I like sushi so I want to go out and have sushi in the future.''

These ideas are part of the identity, and they're long-term evolving in some regard, but typically reinforcing. Whereas fear is very reactive and a survival instinct fundamentally, although it's arguable that desire is also a survival instinct. The desire to eat food that you like is inherently a survival instinct if not taken to an extreme. So these were the components of the identity, and from the identity, some kind of language gets created. This is what I've talked about in the purest form of language in the Noble Ape simulation, where the simulation entities -- the Noble Apes -- actually construct a view of language over time that evolves and is referential.

Maybe groups of apes on one island have a particular language. Another group of apes on another island have another language. This kind of language produces, in some regard, a society. The idea of the combination of referential aspects of space, which have obviously been internalized in the way they're described -- things like their diet and the food that's accessible. If there's not a lot of accessible food how will the language and the society evolve because of this?

These are the kinds of ideas that I really wanted to simulate with the Noble Ape simulation. This was the core of it. It wasn't a genetic simulation. It wasn't more traditional artificial life, cellular automata. It was creating an artificial society and seeing how the low-level effects -- the very basic components like space and time and the absorption of this information, and the evolution of movement through these environments that the Noble Apes have -- would actually create societies that were put together with language and maybe even longer time emergence.

This is what has interested me through the whole Noble Ape development -- the ability to get to a point where these ideas become emergent and produce far more than was originally posited by the original manuals or Noble Ape Philosophic. The Noble Ape Philosophic provides these guidelines of ''This is what I think will happen. These are the stepping stones that move between what is a relatively abstract simulation that does something that is to some degree intelligible.'' Whilst things like visualization and the ability to actually see the apes eating, and things like that, are not even really evident in a contemporary simulation, I wanted to show people very quickly that there were underlying ideas that I was working towards.

While the original manuals, that are kind of overwhelming and have produced a lot of these ideas in scattershot, my original aim was to produce a series of documents -- Noble Ape Philosophic, Noble Ape Mathematic, Noble Ape perhaps Psychologic, Noble Ape Biologic -- all of these kinds of components in similarly printed out to five pages at most, basically. Just so people could have a look at them and know immediately what's going on. I have drafted a Noble Ape Mathematic. I wrote it around the same timeframe, '97, mid '97, that I wrote Noble Ape Philosophic. I've started a new one for Noble Ape Mathematic because I think this is something that a lot of people currently ask questions about.

''How do you actually simulate the brain? What are the underlying mathematical principles you are using in terms of desire and fear and how is the weather simulated?'' All the mathematical components that go in the simulation. Even visualization components, things like Ocelot. People have questions about that.

Although I've written about some of that, for example, through the IEEE article which is still available online, I think. I'm not sure if they're still charging -- they were charging $15 for a period of time for the article. I thought it was a bit over-priced, but it's theirs to sell.

So all of these ideas come together in various bits of documentation. It might be useful to create a superdocument in the future. My only concern with a superdocument is the kinds of users that use Noble Ape and are interested in particular areas and not others, and actually cruising through a superdocument, actually getting readers to get to Chapter 16 that deals with their particular interest is a little bit problematic. I thought maybe individual documents would be a better way.

The Noble Ape Mathematic document isn't that long. It's shorter than Noble Ape Philosophic. The mathematical principles of Noble Ape are pretty simple. They were designed to be pretty simple because ultimately I had to maintain it. I had to maintain it in a relatively leveling degree of complexity. I couldn't see the complexity going up even at a linear rate. It had to keep relatively stable in that regard. Anyway, all this stuff can be described in a future podcast. If you have any questions, tom at nobleape.com.