java moo

[Ray K]

Originally posted by Todd Hawks
If you really see this through, then PLEASE make it bigger than MoO concerning map size. I.e. more stars. As many as possible.

That can be done. Right now, the original map sizes are supported. It would be easy to allow this to be customized.

When I start getting warm fuzzies that this will actually be completed one day (like when I start working on the AI), then I'll start taking requests.

[vmxa1]

My suggestion would be to first get it work as it was originally. Then made a version 1.x that incorporates chanes you deem worth doing or are willing to do. This makes it faster to get something done and will allow people to have the original experience. I would not consider graphics or sounds changes to be anything that impacts the gameplay.

[Urban Ranger]

That's true, the only thing is Ray needs to code the program so it can be easily changed.

[Urban Ranger]

Originally posted by Ray K
Actually, it simply requires something that game developers do not have: time to actually play the finished game in order to develop winning strategies and tactics. That is advantage that the human players always have over the computer once the final version of the game is released.

That's true to a very large extent. Another major factor is it is very hard to write the AI so that it "thinks" in similar lines. Human thinking is fuzzy, there are a lot of factors to weight, and we can shift gear without blinking.

The more strategic a game is, the fuzzier is it to make decisions, thus the harder it is to write the AI. Chess is clean and tactical: all the pieces must follow certain movement rules and there are only 64 squares. Compare that to Go, where there are 361 points and the stones can go anywhere. Even though we have almost as large a body of works on Go as we have on chess, Go programs remain lousy players, while chess programs can now be as strong as the world champion.

Granted, MoO will be a bit easier to code. Still, the decision making process will be hard.

[SeanV]

This is really neat!

Ray K you have earned my admiration for embarking on such an ambitious project.

[CenturionV]

yes this is way cool i dont know much about java but is it web based? also i will it have any multiplayer features (i dont know if the original had any as i dident play it for more than a day)

[vmxa1]

No MP in Moo1.

[Skanky Burns]

Originally posted by Ray K
I intend to keep this version of the game close to the original, meaning that I have the ability to encode proven strategies into the AI. That is something that the original programmers could not do.

The problem with this approach is that an AI will always react to a situation exactly the same every time it encounters that situation. For example, the AI in Civ 3 always sends its military to the humans weakest city in preference to their well-defended border cities. A good tactic.

Unfortunately (for the AI) humans understand this, and force the AI to behave how the player wants it to behave by deliberately present the AI with such situation. In the Civ 3 example, this would be making a city just out of the AI's reach empty, letting their units get near it, and then putting defenders back into the city. They then empty another nearby city, the AI moves towards that... rinse and repeat. Rather than the AI using its military effectively, they end up with their units just moving back and forth infinitely, or until the human destroys the invader unit by unit at their leisure.

The point is that hard-coded tactics fail once the human understands what rules the AI uses in its decision-making. Once the human does understand these rules, exploitation is inevitable.

[CenturionV]

i agree. if theres one area of games today i wish they would spend most of there time on its ai, instead they add loads of features thats all games are now, old games with new graphics and features the ai is still really REALLY stuped but yet we see years of development on the gafix and nothing but a passing note (if that) on ai example, NEVER to my knowledge has there been an ai system in a startegy game were the computer actually used only info it gathered on its own. They always play the same and they always make the same mistakes. The ai always knows were you are, they always know how powerful you are, and they always know your weak spots. what we really need is a kind of information based ai that only know from what they've gathered and only reacts on what they've gathered thus providing alot less exploitation because they will always have different info. (btw i'm just stating some facts about games in general i'm not saying that you should spend any time on this sort of thing in your javamoo)

[grinningman]

I would love to play this game against an improved AI.

I played a game recently with Humans, 4 opponents, med map, impossible difficulty. I was lucky in that the first planet I colonised was a fertile terran, with a population max of 140 or something. During the game, all 4 of the planetary shield techs showed up for me to research, so my planets were always well shielded. (I was unlucky in that the first ship drives available to research were impulse drives. Luckily the Silicoids were willing to trade me sub-light drives).

Late into the game, the Silicoids, my largest neighbour, declared war on me. Unfortunately (for them), their fleets didn't have any weapons that could penetrate my shielded missile bases. Over the course of the war (at least 20-30 turns) the silicoids attacked my planets several times, but NEVER with a fleet that had any weapons that could even damage my missile bases!

It wasn't that the Silicoids didn't have the appropriate tech available - they had at least anti-matter bombs at the start of the war, and a high enough tech level for them to be miniaturized enough to put on medium or small ships. They were just designing the wrong sort of ships! Loaded with pulse phasors that could shred any of my defending ships if they came close enough, but not scratch my planets.

After I finished with the Silicoids, the Meklar got a bit worried and declared war on me. The same thing happened. They attacked 3 or 4 times, different planets, in each case with no weapons that can hurt my bases. Unlike the Silicoids, they did start putting the odd bomb on their ships, but of course by this time, it was too late.

Anyway, what annoyed me about this is that through spying, all the other races should know what tech I currently have (or had a couple of turns ago). So they should know the shielding level of my missile bases (assuming I've spent the production to build shields) , and know that they will need bombs to attack my planets.
This doesn't seem to me (not that I've done a huge amount of programming) to be a super-hard thing to code in for the AI.

*sigh*

Sorry for this rant It's just the first game of MOO1 I've had when the computer AI did something so frustratingly stupid.

grinning man

[vmxa1]

The AI is not likely to be code with a dicesion tree that would be able to handle all the choices it would need to deal with to use proper tactics for all the permutaions that are possible.
It is more likely to have primitive tree that figures out if it should attack you or not. If so then send either ships or troops or both. If the ships it has on hand do not fit the bill, it will eventually change the design, but you can prevent or delay this by not destorying all or nearly all of the ships in a given style. As long as it has a bunch of a given design, it is slow to scrap them and make new ones. If you kill all of them, then it has a chance to change.
I am surprised that they did not have any torps that late in the game. In fact I rarely see them with pulse phasors.
You see the change in games were they show up with bio weapons and you have the antidote. They will drop that design after a while and go for bombs. Later some races will go with torps, especially if they get to plasma torps.
BTW I do not think you can start with any planets of 140 or better legally. Standard HW is 100. Largest starting default planet is 110, unless already fertile. Fertile has never occurred in any starting HW I have ever seen in over 2000 games at all levels.
Even Saks start at 100. A 100 planet can reach 270 with complete terraform and advanced soil. A few planets, such as Orion can reach 300. If you started at 140 (standard) the planet would be able to grow to 330, never seen that.

[Spaced Cowboy]

Hmmm...It has been almost two months since Ray posted here???

Does that mean he is busy coding like a madman...or he was overwhelmed

[vmxa1]

I sure hope not the later one.

[Dwedit]

Originally posted by vmxa1
Even Saks start at 100. A 100 planet can reach 270 with complete terraform and advanced soil. A few planets, such as Orion can reach 300. If you started at 140 (standard) the planet would be able to grow to 330, never seen that.

Planets get even larger if you have Advanced Soil Enrichment + Atmospheric Terraforming, and there's an industrial accident on the planet. The planet's new base pop becomes HALF the terraformed pop, then you raise it from Radiated to Minimal raising the max pop, then you make it Gaia raising the max pop, then you Terraform it +x raising the max pop... It's a GOOD thing to get the industrial accident.

[vmxa1]

I don't that is correct. First we were talking about a start of the planet (HW). Anyway I have captured many radiated planet from accidents and never had one go over 300. It would be unusual to see the accident so late in the game that you have all of the boosters.
I say again, I have never seen a planet over 300, ever. I would be very suspious of any game posted with that condition and would wonder if they used an editor.