Just giving you guys an update and some biography crap...
No work has been done on KAI for almost a month now, and in the way things are, it will most likely remain so for another month. The motivation has been pretty low and I├óÔé¼Ôäóve been doing other things like going to music festivals, camping and international trips, as well as pre-uni pissups. I can hardly bring myself to stay at home and code.
I├óÔé¼Ôäóve tried, but it just made me feel miserable to code in the summer when i could be partying with some friends for the last time before i move! Those other activities are not the real issue however. They are just what i chose to do with my free time otherwise.
People that start big projects and lose motivation follow an unfortunate and predictable pattern. Firstly they come up with one excuse or another (I├óÔé¼Ôäóm moving, i got this trip, i got lots of work to do), then they post/come online less and less around. Updates come more and more slowly and smaller in size and then they usually disappear without a trace or explanation. Id feel bad about doing that though, so ill explain the past, current and future of KAI as concisely as possible.
I started with KAI around November last year, from scratch, without knowing c++ or having any programming experience. This was around the time for the university interviews (i them at six places). Upon visiting the universities and seeing some projects i became more interested in learning programming. At the interview for Imperial College i showed my previous AI project, which was much simpler and easier but the interviewer seemed very impressed by it, and by looking at the third year projects at some universities i could see why, they were piss poor one-day jobs!
I also wanted to get into Cambridge, but at the time that was just an unrealistic idea. I was doing badly at school, missing lessons and failing tests. Lots of letters were sent and my parents received phone calls from teachers, i had less than 50% attendance for physics by January! Since i would probably not make the grades i thought i might as well try to make an even bigger project and try to use it to persuade my way in. I wanted to do that for a while anyways so that was the spark i needed.
Soon after I decided to create an AI for spring i had lots of ideas and planned it all out. My motivation was really high since i really wanted to see those ideas in action. For the first week i was reading c++ books like mad, averaging 400 pages a day! After a few months i had trouble with the main pointer which caused it to crash all the time and i had no debugger either, so I worked less on it.
Once the solution was found by firenu i was again extremely motivated, development went very very fast. Never mind that i was actually taking my final exams and that my future depended on them. If you really want to do something you just do it and well, fuck the world! KAI 0.1 was actually released two-thirds of the way thru my exams, i had spent a lot of time coding and not so much time revising. I expected to do badly on them anyway so i had to work on plan B to get into uni!
In the two days before the results day on august 17th i actually wrote a 6000 word document on KAI, with plans to send it off to Cambridge and hopefully get a place. Turns out i actually got the grades i needed (by just two percent in maths! phew). I was really happy on one side for getting there but on the other it felt like a bit of pointless effort. I did not have the pressure to work on KAI anymore. Also, I had proof-of-concepts for most of the ideas i had for KAI, and there├óÔé¼Ôäós little more to try out, just coding and tweaking, which can be boring.
Tournesol and firenu have not been very motivated (or been busy) lately either so we have a standstill.
Now what will happen? I plan to get into uni first, see if i get any fresh ideas and motivation to work on it (I├óÔé¼Ôäóm doing computer science after all). It always seems more easy to do AI work out of holiday time as well. If that fails, well I won├óÔé¼Ôäót force myself into pushing a 1.0 out, but instead of disappearing I should probably release the source if I decide to stop working on it for good.
*****
If you didn├óÔé¼Ôäót read all that, basically:
No work is being done until October and then I will decide whether to stop working on it and release the source or carry on and release a final version.
While some might not be happy about this, I though it would be better to let you guys know what├óÔé¼Ôäós in my mind instead of simply vanishing like everyone else...
KAI status and future
Moderators: hoijui, Moderators
At the very least, I think that it would be better to release the source now. Having been to college (a long time ago), I know that by October, you will be very un-motivated to pay attention to this project, because you will either be drilling down on your studies and being studious, or getting heavily involved in the much more adult social world of college life... or, like most people, both at once.
Basically, your mea culpa is good, and all of us who've worked on big software projects know exactly where you're coming from.
But KAI has a lot've good ideas in it that should be seen and used by others. While it's not a very complete project in some respects, it was fairly innovative in others, and I think that understanding some of its more innovative features would be very helpful for other projects here.
The longer you wait on something like this, the less ultimately useful your work becomes... and, to be frank, the higher the chances are that you will just not care enough to release the work. AI development for Spring is still in its early stages, and projects will continue to come and go. As you've actually advanced certain areas of interest to all would-be AI designers for this game engine, I think it would be in everybody's interests, including yours, to have your source code out where it can be used and improved.
Basically, your mea culpa is good, and all of us who've worked on big software projects know exactly where you're coming from.
But KAI has a lot've good ideas in it that should be seen and used by others. While it's not a very complete project in some respects, it was fairly innovative in others, and I think that understanding some of its more innovative features would be very helpful for other projects here.
The longer you wait on something like this, the less ultimately useful your work becomes... and, to be frank, the higher the chances are that you will just not care enough to release the work. AI development for Spring is still in its early stages, and projects will continue to come and go. As you've actually advanced certain areas of interest to all would-be AI designers for this game engine, I think it would be in everybody's interests, including yours, to have your source code out where it can be used and improved.
Aha, there're similarities in us both then, I too had a similair experience with physics lessons. My uni offers both changed to unconditonal the week before my results though, and whereas you've had a big spurt that slowly unravelled into a trickle and now almost nothing, I've done the opposite and if anything its an unhealthy obsession by now.
I've offered to help with KAI before and I'll still do it, and I may even pick up the project once its opensource and continue it. Super AI ftw!
I've offered to help with KAI before and I'll still do it, and I may even pick up the project once its opensource and continue it. Super AI ftw!
I wont be releasing anything until october, that is 100% sure, i can live with the risk of forgetting to do so. I might in fact take a third option and open source it , create a detailed roadmap and get people to help with it, nothings certain for now :D. And if AF is really determined, then hopefully a new uber AI will come out with the next spring version, rekindling the competition fire, that could sure motivate me as well...
honestly i think that KAI was the most promising AI out there, given its still very early status of development.
On a sidenote this is something every programmer should have seen once:
http://swissnet.ai.mit.edu/classes/6.00 ... -lectures/
its rather old (early 80ies) but trust me its well worth the effort and the ideas are far from obsolete.
On a sidenote this is something every programmer should have seen once:
http://swissnet.ai.mit.edu/classes/6.00 ... -lectures/
its rather old (early 80ies) but trust me its well worth the effort and the ideas are far from obsolete.
This is an opensource community, everything should be opensource, anything else simply causes negative effects such as paranoia, suspiscion, unhealthy competition, or general lack of progress and fustration.
But why you made that comment I dont know, afterall I thought I'd expressed that to you specifically in private many times before. Afterall cosndierign the reasons you gave above for starting it why not keep it opensource? I got praise for keepign NTai opensource, and submarine got a big thumbs up for putting AAI in the svn, and AAI benefited as a result. Tobi wouldnt have been able to fix a few NTai bugs, fix AAI on linux, and add AI multiple directory support for both AI's if they where closed source. Then there're the bugfixes people have sent me in the past.
Daraknor I think it was told me to read an article called the cathedral and the bazaar before he left, and it helped, I reccomend you look at it too.
But why you made that comment I dont know, afterall I thought I'd expressed that to you specifically in private many times before. Afterall cosndierign the reasons you gave above for starting it why not keep it opensource? I got praise for keepign NTai opensource, and submarine got a big thumbs up for putting AAI in the svn, and AAI benefited as a result. Tobi wouldnt have been able to fix a few NTai bugs, fix AAI on linux, and add AI multiple directory support for both AI's if they where closed source. Then there're the bugfixes people have sent me in the past.
Daraknor I think it was told me to read an article called the cathedral and the bazaar before he left, and it helped, I reccomend you look at it too.
I hope you didnt think I was trying to snipe there, just trying to express an opinion and explain it.
http://www.catb.org/~esr/writings/cathedral-bazaar/
http://www.catb.org/~esr/writings/cathedral-bazaar/
the ability to keep units out of enemy range while still bombarding the target with long range units was the deadly feat of this AI, it is only how a human would play and that killed me off most of the times (combined with the maphack). in KAI vs AAI games hordes of AAI units fell victim to a few long range KAI units that were properly managed.
I too experienced this, and i'm a little disappointed that Krog won't be going too much farther with it. It would be cool if AF would be able to improve it a bit more. (make it spam more units, but keep the map hack.)
I honestly don't mind the fact that the ai cheats, it provides a challenge and is quite fun.
I honestly don't mind the fact that the ai cheats, it provides a challenge and is quite fun.
I took that class . . . . good timesImperator wrote:honestly i think that KAI was the most promising AI out there, given its still very early status of development.
On a sidenote this is something every programmer should have seen once:
swissnet.ai.mit.edu/classes/6.001/abelson-sussman-lectures/
its rather old (early 80ies) but trust me its well worth the effort and the ideas are far from obsolete.
Defintely a pretty wierd language to use, but great for AI. (scheme/LISP that is).
Back ontopic:
Good luck krogothe with whatever you do!
-Redcoat