Re: [Fear-devel] SOD Questionnaire

[Alex J. C. <al...@ai...>]

>At first, I see the development of components that should be familiar to
>budding AI programmers that have read the Game Programming Gems and/or AI
>Wisdom books. This would include FSMs, FuSMs, Fuzzy Logic etc.

Yeah, this seems relatively straightforward for NN/GA/DT, as they can be 
wrapped inside modules with a clean interface.  For FSM, the task is harder 
to do "cleanly" as this component is extremely intermixed with the AI code 
itself (not a single module with a nice interface).

I do plan to do all these personally 
(http://sourceforge.net/pm/task.php?group_project_id=24163&group_id=51070&func=browse) 
eventually.  Simple implementations of course, but enough to support my 
future efforts with Fear.  Of course, assistance is more than welcome.


>Ideally visual tools and/or scripting interfaces would be developed for all
>these components to make it possible for non-programmers (i.e. games
>designers) to use them but this is definitely something for Fear 2.0 and
>beyond.

Scripting is controversial, as it really needs to be (i.e. I'd like to make 
it) a fundamental part of the FEAR core (helping in inter-module 
communication for example).  This causes problems as proprietary scripting 
languages will be potentially required by many teams using FEAR.

The design of 0.1.x is already taking into account a GUI / IDE, and all the 
implementation is also geared towards less programming for developers -- 
and more AI experimentation.  So I guess it's good to think about 2.0 now ;)


>OK. Let me rephrase. I think this is a good time to recruit new developers
>so that Fear will acquire interesting new features quickly. We will each
>have different ideas for what we want to do with Fear and that should
>increase its scope.

I think I understood that initially.  My point is, initially there was 1 
dreamer.  He sold a vision to another programmer, and the project 
started.  Having convinced a few other AI fanatics that this is a good 
idea, the project has grown to 5/6 people (thanks to some simple docs and a 
bare bones implementation).  However, everyone seem to be holding their 
breath for something (time, missing feature, etc. -- depending what they 
want from the project).   It's the same for game programmers (RTS, 
RPG);  they're waiting for maturity and a good demo (i.e. it's no use to 
try to reach out to them until we're ready to take them on).  So, what i 
meant is that we're fundamentally limited by the potential of the people 
who bought the dream, until the dream turns into a more solid reality.

I've done my best to nurture any signs of interest in the project, getting 
developers and researchers involved as much as possible.  However, getting 
more people involved is a matter of putting in x amount of work, producing 
an impressive demo with a reliable and mature API to get mod teams, hobby 
game developers interested.  And that's something that this team alone will 
have to do if the project is to succeed.

The snowball effect I guess.   Now is time to push.  I push a bit so you 
want to push more, and we all push together.


>Easing the integration with multiple platforms is probably the hardest
>problem facing the development of Fear but also one it's most important
>things to get right, otherwise, Fear will not gain acceptance as a viable
>platform.

I think that is something that Fear has done well.  To get Fear running in 
another environment, the existing engine interfaces need to be ported, and 
that's it.  You recompile, and voila!   Gaining wide acceptance is a matter 
of having mature and elegant specifications that define these interfaces.


>Hopefully, the benefits of using Fear to gain access to interesting and
>convenient AI technologies will encourage developers that specialise in
>games platforms to to provide new integration layers for Fear.

Indeed, having a core NN module for example, finally implementing RPROP 
(which people know is great but don't know how to tackle), would do 
wonders...  Then GA / DT/ FSM and we're in business!

Alex