What are your early game god-mode mod/unit combinations?

[Ferrus Animus]

Because measuring 2 objects to an inferior benchmark does not make them equal, as I pointed out. The context of pve or pvp is irrelevant to how the data is measured. The difference is pvp notices the imbalances because the other player(s) are actually capable of using them.

Consider the ai to be the control group in context of balance. Everything the player has is pitted against this test.

The expiremental test is determining if a player with x is greater than player with y. Any huge imbalances can more or less be found on paper without applying pvp situations at all. If a little bit of math shows that x=50 and y=25, there's probably a game breaking discrepancy between the two.

I know why you think this is the best approach, but I disagree.

First, calling the AI an inferior benchmark ignores that it is the one that matters for SP.
And second it leaves completely out that the AI is a differently weighted challenge. The AI has advantages and disadvantages compared to a human player, that another human player doesn't have. Using it as a control group doesn't apply either, because it is not baseline, compared to players.

The issue is more that when X returns 50 and Y returns 40, and the AI returns 45, then simply comparing X and Y and deciding to bring them in line with each other, is a hail mary with regard to the AI. And that's before we compare summands.

 

[Sinsling]

I know why you think this is the best approach, but I disagree.

First, calling the AI an inferior benchmark ignores that it is the one that matters for SP.
And second it leaves completely out that the AI is a differently weighted challenge. The AI has advantages and disadvantages compared to a human player, that another human player doesn't have. Using it as a control group doesn't apply either, because it is not baseline, compared to players.

The issue is more that when X returns 50 and Y returns 40, and the AI returns 45, then simply comparing X and Y and deciding to bring them in line with each other, is a hail mary with regard to the AI. And that's before we compare summands.

The ai isnt a differently weighted challenge. Game settings might create a differently weighted challenge(see:conquered worlds, 1v11ais), but the ai player is a set of codes designed to play with the system with variable cheat factors included to attempt to make it a competitive oppenent.

You've also showcased variables well within margin of error for on paper application. You'll note that I used a 25 point difference, which by your measurement the ai would sit at 12.5 from either. X would still significantly stump the ai while Y is so useless it's likely not even applicable to SP*.

*that's not even accounting for x>y>ai situations of similar caliber.

 

[NINJEW]

I had real life stuff to attend to.
And it's enlightning that you waste no time showing your character so publicly. I was assuming we had a good faith argument, but evidently not. I wont bother to discuss with you again.

Hey remember this?

I will reply to this and then stop engaging this contention, because frankly, I do not know how else to explain it anymore. I don't know how to tell you how somethign looks from a different angle when you seem inccapable of moving.

Lmao someone really wanted to get the last word in. You know this gets like 8 times more pathetic the second time you say it

And you still can't respond to actual analysis of, once again, the balance topic that you chose. I would love to hear an actual analysis of how celestial MC plays out and a proper defense for how that isn't broken, though I doubt you can given how cagey you've been about, like, what actually defines "broken" or "OP" for SP, all you seem capable of doing is saying "it's not the same criteria that those terms are judged on in MP"

If you do, I'd love to have a real discussion with you on whether your definitions for those terms are or are not reasonable, though first I'd have to finish laughing at someone saying twice that they're going to stop responding and still responding a third time.

 

[Ferrus Animus]

The ai isnt a differently weighted challenge. Game settings might create a differently weighted challenge(see:conquered worlds, 1v11ais), but the ai player is a set of codes designed to play with the system with variable cheat factors included to attempt to make it a competitive oppenent.]

You know that these cheat factors influence the game? That is why they're there.
The AI competetetiveness cheat tries to amek the AI a better challenge by giving it more stuff. That more stuff needs to be fought. (Which is not the case in PvP, where every palyer has roughly the same amount of stuff).

To go back to the Celestial MC example:
Extreme AI gets +40 food/turn and +100% to its income. A simple starting capitol has 1 food worker and +20 base food, so the AI has a comparative advantage of +65 food per turn.
Depending on how elaborate you want to do this, you can calculate the food values the MC adds via colonists and find the break even point.
But by turn 8 (assuming you get MC online then) that is 520 food advantage (which causes further resource growth via colonists). And the AI keeps getting resources too.
So while the ability has an absolute gain, that is significant in a 1v1 Duel PvP environment and is a relative advantage from the turn it works, for the big map PvAI environment the point where that absolute gain turns into a relative advantage much, much later, if the player can manage to keep it up.

You've also showcased variables well within margin of error for on paper application. You'll note that I used a 25 point difference, which by your measurement the ai would sit at 12.5 from either. X would still significantly stump the ai while Y is so useless it's likely not even applicable to SP*.

*that's not even accounting for x>y>ai situations of similar caliber.

Look, I really have to question if you get what I'm saying here.

The values were an arbitrary example. They don't matter for the deeper point. Do you recognize that point?
If it is easier, make the values X= 100; Y= 1 and AI= 50. Does that help?

 

[NINJEW]

imagine a world where ferrus is capable of naming something that was ever broken in SP

As for an example, I do consider battlefield autopsies (the assembly doctrine) to have been much stronger in SP and it might even still be on bigger maps. Where it gave a handful of potentially deciding techs in duels, on bigger map it allows doomsday rushing rather quickly before the AI can react to it.

maybe the closest example and if you perform that same analysis i highly doubt battlefield autopsies ends up coming out as looking better than celestial MC or the +tech cheats the AI gets

i do not believe it is possible for ferrus to believe that any mechanic is broken in SP. everything is balanced in SP unless it is somehow wildly more valuable than the AI's cheats, which just isn't possible outside of a dev having a stroke while editing an ability

 

[Sinsling]

The values were an arbitrary example. They don't matter for the deeper point. Do you recognize that point?
If it is easier, make the values X= 100; Y= 1 and AI= 50. Does that help?

I could just as easily use 1, 500000, and 1000000 as well. I used numbers that I felt provided a reasonable enough gap that noone would misconstrue the two as being anywhere near each other. When you attempted to use a smaller gap as justification, I simply pointed out the difference in value between the arbitrary values we both used.

Those resource cheats also aren't amounting to much. With a simple food sector and pop in food, byebye ai lead;

I was originally going to go until turn 8, but a food event I knew how to use and the ai didnt came up. I easily could of flown past the ai's pop just by taking that biodome or, better yet, the food landmark.

If you want to continue claiming it takes anywhere near optimized play to catch the ai, please start providing counter-evidence, because it clearly isn't true.

 

[coodav]

Guys, could you please summarize what you think each other is saying? I have to say, I've completely lost the thread. This seems to be a real odd round-robin of semi-insults with a few arbitrary numbers thrown in. Help me get back on the train, or perhaps consider if it is a ride worth taking? Who believes what about this exactly?

 

[Sinsling]

Guys, could you please summarize what you think each other is saying? I have to say, I've completely lost the thread. This seems to be a real odd round-robin of semi-insults with a few arbitrary numbers thrown in. Help me get back on the train, or perhaps consider if it is a ride worth taking? Who believes what about this exactly?

There is an attempt to claim that the ai is a competent measure of balance because it has cheats, and there is a counter claim that the ai is not a competent measure of balance because it is an incompetent player.

Or at least that is where it has ended up, after claims that MP and SP balance is different.

 

[NINJEW]

ferrus believes that balance discussions shouldn't be MP oriented because the balance conditions between SP and MP are radically different

i've mostly been saying "ok so what does a SP balance discussion look like? what metrics do you use to judge if something is overpowered or not? do you have an example of something that is overpowered? why do you think that x thing that MP people say is overpowered is not overpowered in SP?" and getting anything resembling an answer to these questions has been like pulling teeth. which is why i've said that despite ferrus calling for SP balance to be more highly considered, he's totally incapable of actually having a discussion in the terms that he's talking about, because he can't put forward anything other than "it's different in SP than it is in MP" but in many more words. it was only in his most recent post that he finally gave anything resembling what an SP-focused analysis of an issue would look like from his perspective, despite being asked to repeatedly, and we are on page 8 and also it's really bad

sinsling has detailed his perspective above

 

[NINJEW]

best i can tell ferrus thinks something isn't broken in SP unless you get more resources out of it than the AI gets through cheats, which uhhhhhhhh

i can see why he would want to avoid analyzing any specific issue or naming an example of an extent balance problem

 

[coodav]

There is an attempt to claim that the ai is a competent measure of balance because it has cheats, and there is a counter claim that the ai is not a competent measure of balance because it is an incompetent player.

The AI is incompetent. I think that is clearly proven. The cheats are apparently about the only thing it has going for it. Can they approximate each other? Well... maybe. Not sure if you won that last game, but if you slant the table against yourself more, I think we could get it there even without you throwing it. So you want me to think of some examples of how that would work? A new 'mission impossible?' I bet you I could do it, if you played by my non-computer-regulated rules.

ferrus believes that balance discussions shouldn't be MP oriented because the balance conditions between SP and MP are radically different

Well... we could break apart the computer behavior, and how to make it function best. And we have all jointly tried a lot of these combinations lately. To the core though, the MP strategies were amazingly successful on the SP opponents.

Neither of you seem to be asking the end-game question though: 'what do you use to minimize the turns before victory - by conquest?' And if you do this, there may be a different strat. Could even be something we try to find out. But the MP strats would probably do fine on this, and may lead to snowballing victory chances beacse of how fast you go out of the gate. So it may be best in both.

I mean, there is a discussion to be had here. And it would be great to dig into the code behind the computer behavior, so maybe we could even help out the AI and spot real solutions to a lot of the problems we are seeing. But I fear I am presuming a lot. So why don't you think the conversation hasn't lifted here? The real discussion seems obvious to me. Can I get you to change to this?

If you did, what I would really like to ask is, what are the biggest technical failures in the AI, and how can Triumph / Paradox fix them? And this is a super-technical question, some of which I have been harping on for the last few months.

I like playing SP. I don't like playing this SP. Beating this SP gives the same satisfaction as beating my toaster. If we could somehow get through, it would be amazing.

 

[Ericridge]

I believe the discussion on how to improve the AI warrants a new thread while this thread is about godmode mod unit combination.

 

[Ferrus Animus]

There is an attempt to claim that the ai is a competent measure of balance because it has cheats, and there is a counter claim that the ai is not a competent measure of balance because it is an incompetent player.

Or at least that is where it has ended up, after claims that MP and SP balance is different.

I think both are true actually.
My attempt is to claim that the sum of the cheats and incompetence provide a difference that needs to be considered as part of balance.

The claim I see arranged against me is that due to the AI being an incompetent palyer it should be disregarded and the balance should only consider the issues coming up in in a player against player comparison.
And while I fully support the result of putting different strategies on equal footing, I consider the SP experience that results from that to be another factor that shouldn't be disregarded like that, as it is the experience I dare say most players of the game indulge in.

 

[Sinsling]

I think both are true actually.
My attempt is to claim that the sum of the cheats and incompetence provide a difference that needs to be considered as part of balance.

The claim I see arranged against me is that due to the AI being an incompetent palyer it should be disregarded and the balance should only consider the issues coming up in in a player against player comparison.
And while I fully support the result of putting different strategies on equal footing, I consider the SP experience that results from that to be another factor that shouldn't be disregarded like that, as it is the experience I dare say most players of the game indulge in.

Fair enough, but as I've already brought up the inclusion of SP more or less just comes in as a factor of "do x and y both work vs ai", hence why ai is the control group of balance. Every build is facing the same cheats, same logic, same abusable movements. Once the control result has been established for whatever is being balanced, it has to be compared to the control results of another method of getting similar outputs. The two methods don't even necessarily have to be used against each other in pvp.