AI enhancement (test games)

[kyrub]

Inspired by sargon0's view , I decided to put up a small thread, dedicated to the thoughts and to the practical trials of new ways to enhance the in-game AI behaviour.

I know that many people here love the game simply as it is (me included), and I do not plan to offend anybody who plays it the old right way. Still, there are a couple of AI performance issues, annoying weaknesses, and a couple of exploit points, which could be possibly sidestepped or played down.

Process and rules:
1) I intend (and I invite others) to post altered test game files.
2) If anybody wants to try the altered game and share the view over the enhancement, I will be delighted. Hopefully we can gather some info and get some idea about an enhancement.
3) The initial poster always gives the reason and the explanation of the intended effect and supllements the exact hex values changed.
4) The changes in one "version" will touch just one area and be as minor as possible.
5) Contra arguments are welcome.

[kyrub]

Test game I - Ship maintenance

FILE REMOVED.
NOT WORKING. :thumbsdow

[kyrub]

Test game II - Have you ever killed him?

Myself, I never did - it's so boring when you have won it all already. Yes, I mean the Guardian.


Current state:
- the Guardian is overpowered (especially on impossible level), so that Mastering of Orion system is never a real strategy. This concept should have added some spice and a sense of urgency in the game (have THEY conquered it already or not?), but it has failed horribly.
- The Guardian is too unbalanced, great against big ships, but sucks against the fighters (only ScatterPacks). It should be so, I know, but maybe we could let the numerous scatter packs stay in our new version, for a slightly more balance.
- Death ray is useless (it is simple to change, but will be adressed later)

Enhancement:
Cut down the Guardian stats and make him a viable alternative for a mid game.

--------- original values - simple....impo ----
Scatter Pack X's.......5 25 45 65 85
Stellar Converters.....5 15 25 35 45
Plasma Torpedos.......6 9 12 15 18
--------------------------------------------
Hit Points (x1000)......2 4 6 8 10
--------------------------------------------

--------- test values - simple....impo -----
Scatter Pack X's.......20 40 60 80 100
Stellar Converters.....04 08 12 16 20
Plasma Torpedos.......06 07 08 09 10
--------------------------------------------
Hit Points (x1000)......2 2,5 3 3,5 4
--------------------------------------------

(it may be still too much... let us try)

Effects
1) Enrichment of player strategies.
2) What if AI race kills him first?

Hex locations
Uff, several.
158ba and 158bf are the HPs and the increase of HP per difficulty level (both 2000).
(the others are nearby, just compare it with original via Hex editor)

Feel free to test. Or to suggest the right blend of Guardian stats!

------ Starmap test II a: added RefSteels' suggestion, Death ray is now a streaming weapon // somewhat less Scatterpacks 5-20-35-50-65 for the balance of Guardian's power -------

[RefSteel]

I'm almost sure the AI is programmed never to attack the Guardian, no matter how powerful its fleet, so creating a "race to Orion" would require modifying a different part of the code entirely. I must also admit that I like the Guardian's near-invincible status, though making larger ships more effective vs. smaller ones might be a good thing. (When I defeated the Guardian in Imperium 11, I used what turns out to be a very sub-optimal ship design - a medium hull with gauss autocannons - because I thought the Stellar Converters were streaming weapons for some reason. Of course this is wrong, and fighters are therefore the better option absolutely.) Increasing its shield levels (and perhaps decreasing total hit points) would probably be the best way to do this, as the Scatters just get dodged regardless of their numbers in an optimal attack on the Guardian.

[kyrub]

I'm almost sure the AI is programmed never to attack the Guardian, no matter how powerful its fleet,

I read somewhere a report claiming that AI tried repeatedly take Orion in his game. But the author did not seem smart enough for me to take it for granted.
I will run a preliminary test to see if you're right, RefSteel. Will make a feeble Guardian inside AI territory, hold a good position, max missile bases, sit down and push Next for a long long time. Then we'll see.

Your other suggestions
- the new Guardian is no pushover, it is basically an Average old one on impossible level.
- on impo level, Guardian has shields 9 and beam defense 8. It is quite good already, do you think it should be even more?


Thanks for your thoughts.

[vmxa]

I once figured I had played 2000 games of Moo and that was years ago and in all of those games I have never seen Orion taken on my the AI. I ran into 32k bug a couple of days ago and they did not go.

The fleet had two sets of large ship of 32,000 and did not attack the guardian, what does that tell you. My late best friend once sent me a game where one of the races had a fleet of 64,000 Huge ships and 2 stacks of large at 32,000 and Orion was intact.

I feel quite safe in saying the AI will never attack Orion. Making a change to that would require a code change.

I do agree it would be good, if they did as they will in Moo2. If you wait too long they will have Orion and the tech they get.

[RefSteel]

Sorry I missed this for a while....

- the new Guardian is no pushover, it is basically an Average old one on impossible level.
- on impo level, Guardian has shields 9 and beam defense 8. It is quite good already, do you think it should be even more?

The reason I suggested more shields actually is probably moot: Prevent small ships from being effective by obviating weapons they can actually mount, via shields. Trouble is, that doesn't make huge ships any better-suited for the task; it just makes small ones ineffective until very high tech levels (so lots of mediums like I used in Imperium 11 might become the best bet, for instance). A better solution to the "Always use Fighters" situation might be to modify the Guardian's weapons. E.g. ... Death Ray --> Streaming weapon? That would make the Death Ray incredibly powerful, of course ... which, considering what people now think of it, might not be a bad thing.

[kyrub]

Now this is a cool idea (seems to me at least)! Two birds with one stone, death ray useful and Guardian hard to take with fighters.

Added another testgame, see above. Thanks, Refsteel.

[kyrub]

Test game I - Ship maintenance

FILE REMOVED.
NOT WORKING. :thumbsdow

Here we go again:
A testgame with ship maintenance halved and missile base maintenance raised to 10BC/base (used to be 6BC).

And it's working

reasons:
- to help CP who creates clouds of ships in the midgame
- to make missile base deffensive tactics (a la diplomatic win without any ship) ecomonically detrimental
- MAYBE to eliminate the negative fleet bug (needs testing)

Hex:
0x88a9 32>64
0x898d 32>22

Feel free to test. I will be grateful for any reports or suggestion... etc. You know.

- credits to sargon for his analytical work on savegame files -

[sargon0]

Interesting ideas!

Firstly in the file I downloaded x898D was x16 (decimal 22) which affects results.

Ship maint was halved, as expected. Feelings are, as mentioned, this could help CPs with large fleets. This could make CPs more challenging or could unbalance some areas eg. CPs could become too powerful compared to Human player, Militarists could become too powerful compared with other objectives, ships could become too powerful compared to bases. Also do CPs decide number of ships based on cost (unchanged), maint (halved), objective (unchanged) or need (abstract)? Even if maint is a deciding factor this probably will just mean more CP ships which is not better strategy but could enhance challenge or unbalance things depending upon difficulty.

Missile Base cost & maint is impacted by tech level and, for CPs, by difficulty level. On early game on Hard I found Human Player and CPs MB maint more than doubled. I did not try to see if CPs built fewer bases from a new game. My expectation is the CP strategy is fixed by difficulty level (they start very early on Imposible) and they could cripple themselves on higher levels with too many expensively maintained bases but this is just my guess.

Good work but would need some serious trials to determine impact on CP and Human strategy. Nice to have the choice though!