Will SE5 AI ever reach that of SeIV? |
Considering that the AI in SEIV is far superior to SEV..
And considering that the modded AI in SEIV is very, very strong...
Why have we still not been able to port that knowledge over to SEV?
I'm just curious as to what the complications are that are causing the considerable difference in AI competence.
In your opinion will the stock/modded AI ever reach the level of SEIV stock and mods?
Thanks,
Javaslinger
Re: Will SE5 AI ever reach that of SeIV?
R U FRACKING KIDDING ME?"?? SEIV AI is CRAP!. I've played a few games of SEIV Devnull 180 while waiting for SEV to be playable and let me tell you something... even after I make 3 or 4 mid level empires surrender to my #1 ranked enemy it's still no contest. Frankly it's about as boring as a nail considering the fun is about gone from the monotonous leveling up.
I gotta say that SEIV AI sucks donkey dicks. That crap shit AI can't provide a challenge unless you are a noob. Pisses me off. I hope the AI is SEV can do something worthy... I hope I dont have to go surrendering lesser empires to make one uber empire just to get a challenge. AI bonuses dont even help...
There needs to be a cheat bonus for AI... Only way they'd be anything other than a doormat to step on. I dont even know why I bother playing this game anymore...
Re: Will SE5 AI ever reach that of SeIV?
I don't know, I think SE:IV in it current form is one of the best games I've ever played. Sure the AI does some wacky thing some times, but I bought this game new in 2000 and it's still installed on my machine.
I don't think I have any game that old still on my machine.
Re: Will SE5 AI ever reach that of SeIV?
While I agree the AI needs work I don't agree that it needs a cheat bonus. It needs to be scripted correctly but that means sitting down, determining how you make your decisions, writing that decision tree into an algorithm and changing the script. Dizzy, are you working on it? I am tinkering with scripting but only to learn how it works so I can "graduate" to the AI but in amongst the other demands of life. Willing to swap ideas. There are others out there working on different aspects as well...
Re: Will SE5 AI ever reach that of SeIV?
I'm currently doing a large amount of work on the construction side of the AI for stock and planning to move onto some other area's of it also once I've finished. I'm doing this as a pure AI mod, though it won't port to other mods very easily. But will be adaptable with a bit of scripting, since I'm trying to write this in a very modular fashion.
I don't expect I'll be up to any releasable version of this anytime under a month, a month and I may have an AI that performs capably in some area's. Though fleet tactics I expect to take me a long long time.
As for an AI 'Cheat' bonus, there is an option in game setup to give the AI a bonus, it's up to the player in that regard, and they can choose to give it no bonus, or up to a *2 bonus I think is the highest.
Re: Will SE5 AI ever reach that of SeIV?
It will but the AI for SEV is a more complex thing than it was for SEV and given that the AI is set to emulate a human player its going about it the hard way. The AI in the balence Mod and IRM is approaching that I remember from Stock SEIV already though as most of my games where PBEM the quality of the AI was rather a secondry factor.
Re: Will SE5 AI ever reach that of SeIV?
I wasn't that impressed with the AI for SE4 as it did not do the basic things like upgrading facilities or build the good production boosters. I certainly hope that with this 5th rendition of the game the AI gets squared away so that it uses what's in the game intelligently. Plus no more silly AI fleets chasing human fleets around from several systems away. I'd like to see more hit-and-run tactics from the AI where it will go in and clear out a system then head home to resupply rather than sending fleets on fuel wasting tours of the galaxy while we drop in on their poorly protected homeworlds.
Omnius

Re: Will SE5 AI ever reach that of SeIV?
Remember this also, the AI script is only as good as the person who programs it. A person with a highly tatical mind will be able to design a very challenging AI. A person who has only limited tatical skills will not program a very good AI. You could be the worlds best programmer, but still not make a challenging AI to everyone if you do not have a tatical mind.
Re: Will SE5 AI ever reach that of SeIV?
Part of the problem is how poorly balanced the stock SEV is, all AI concerns aside - fighters, anyone? Before you can start making a challenging AI that uses different weapons/units effectively.. you have to balance the weapons/units so there's actually a reason to do more than simply mass produce whatever happens to be overpowered. The best AIs come along with mods that remake the game itself from the ground up, and then create an AI to play this new reblanaced game, there's just not much you can do with the stock SEV. Unfortunately all that seems to be slowing down the AI development and dividing up the modders and their incompatible mods and AIs. I used to play balance mod and now I'm enjoying IRM, but I'd much rather see a playable stock game so we (and our AIs) are all actually playing the same game..
Re: Will SE5 AI ever reach that of SeIV?
It will not only reach the levels of SEIV it will surpass it. Allowing modders to actually tinker with the AI scripts means no limits if you have the time to work with it. Keep in mind for every adjustment there is a reaction. EXAMPLE: The AI built way to few attack ships, I was able to increase the number of ships only to find with more ships.....now the maintenance costs have risen.....ok let's get the AI colonizing quicker as more colonies will bring more resources to pay the extra maintenance costs....OK that means now we have to improve the AI's exploring so they will find more planets to colonize on. Hmmmmm more ships & more exploring?...now we need more resupply bases as lots more fuel is being used & ships with no fuel don't explore very quickly.....What?...more resupply bases will mean less planets for resource production. etc. etc. etc. Since I 1st played the stock AI back in Nov. 06 I've been determined to make a better AI...so much so that I actually haven't played the game as a human player since, only hours & hours & days of AI testing trying to encounter as many different situations as I can to see how the AI reacts to it. What I've discovered is that MM has presented us with a blueprint with endless possibilities in the scripting. Keep in mind I'd be willing to bet that he has never had the time to actually play his game. The advantage I have is I've played on PBW a bit & have a good idea of what will actually make a competative AI. MM has to depend on his beta testers who are doing their part as every patch has improved the scripting of the AI in different respects. I'm very very excited about what I've accomplished so far but until I work out a few more kinks I will not release it as of yet.
Some examples:
Ever see an AI fleet in SEIV that has Mine Sweepers in it?
I've not only accomplished this with SEV, I also have created not only a repair ship but a resupply ship that travel with the fleet as well.
Ever see the AI mine warp points in SEIV
My AI mine layers & satellite layers launch units on SEVERAL different defensive warp points.
Just a couple small examples. I could type for an hour telling you stuff I've improved on & other surprises that I'm resisting talking about.
Let me also say I am not a programmer or even a modder for any SEIV games(discovered seiv when sev was already in development & seiv had countless mods already) But I am very determined & as I said before Aaron left us a basic blueprint to improve upon to no end. I have not rewritten any scripts from to ground up, only improved upon them using the function choices installed in the game. As beta testers & modders suggest more functions to Aaron, he continues to add them in giving the modders even more to work with on improving the AI. So hang in there for a bit, my up & coming release will be enough to show what is possible, especially for a modder that knows a bit more then me.




Re: Will SE5 AI ever reach that of SeIV?
Good AI takes time. Maybe you weren't around when SEIV was released, but the AI was pretty bad then just as it is in SEV now. It took years of hard work by Arron to get that to a decent level, and it also took as long for the modders to come up with something that is worth playing against. Now you might say that SEV is the same as SEIV, and because of that the AI should be great due to experience, but that's just... well, wrong. There are many little things that we humans considered minor changes that causes the AI absolute hell... things like a changed tech tree and the resulting change in components and facilities, things like a different ship design mechanic, or hell, just the fact that Meson Cannon is now the uber weapon versus pretty much everything else. The balance of the game has changed, and it will take time for humans to learn this balance and then use that knowledge to design (script) the AI to do the same.
Unfortunately, the ones who has knowledge of AI design also likes to tinker with the game in other forms (they like to mod the game to their own taste), thus further changing that balance. These modders usually needs lots of time to mod, test, remod, test, etc... until they are finally just too tired to do anymore. All of this takes time, lots and lots of it...
CAN the AI in SEV get as good as the ones in SEIV? Yes. WILL it get as good as SEV? That depends on the patience of the people working on it.