Duels of the Planeswalkers for XBox, PC, and PS3

[Oversoul]

Oversoul: I realize that you or no one else has a superhuman memory, but I was merely responding to the example you gave, not future examples that you might give because you thought of them just then. I think we all know the AI sucked in Shandalar; some were probably avoidable and some probably weren't.

I wouldn't expect you to respond to stuff I hadn't said yet. I don't know why you're saying this.

All *I* was saying was that for something as simple as coding a landwalk ability that is pretty much non-interactive or conditional, unlike the examples you were giving, having it not work in 2009 is almost inexcusable.

Yeah, it is pretty silly.

So hopefully it IS a "glitch" (whatever that means) and hopefully that means a fix will come out, although fixes for the few XBox games that I'm familiar with (like Puzzle Quest) seem to take a really long time.

What I mean by a glitch, although again there's the caveat that I couldn't program my way out of, well, anything, is that it could possibly be a case where, rather than mistakenly coding the rules so that landwalk does not work, they coded the rules so that landwalk did work, but another part of the coding is interfering with it. I don't know think it would matter much for the purposes of patching it. A patch should be possible either way, right? But it does make such an egregious error more understandable, as rather than them being sloppy with something to simple as landwalk, they were sloppy in a more subtle way and it messed landwalk up.

Do all landwalk abilities always fail? Is it just mountainwalk or just in certain situations?

Also, I kind of forgot about the possibility of patches because back when I played on consoles (other than the handheld DS and Gameboy, which I still have), it would have been completely impossible. A couple of friends got an earlier, glitched version of a wrestling game on the N64 and I got the fixed version. Theirs remained messed up in whatever way it was and mine was fine. But that's no longer the case now. I wonder if that doesn't cause the makers of games to be at least a bit reckless sometimes. If something is wrong, they can fix it with a patch later. That doesn't mean they won't test thoroughly, but maybe they're slightly more likely to let stuff like this get by them?

 

[Spiderman]

I wouldn't expect you to respond to stuff I hadn't said yet. I don't know why you're saying this.

<sigh> Bear with me...

Because originally you said

The AI will sometimes giant growth my creatures during combat. Stuff like that.

(and I took "stuff like that" meaning the AI casting beneficial spells on your own creatures) and I replied with

Yeah, but that was more than 10 years ago. This is now; and something like Mountainwalk should have been easy to code.

, meaning 10 years in coding is like lightyears in the computer world, and you replied with

Shouldn't it have been easy back then to not have the AI kill itself?

, which I took to mean that even for 10 years ago, it shouldn't have been hard to code a basic directive as to not kill oneself, and I replied with

Yes, but you didn't mention that example. Just the Giant Growth.

, meaning yes, I agree with you, but that didn't fall into the original observation of the AI casting beneficial spells on your creatures, and you replied with

Well now, I didn't mention it quite simply because I hadn't thought of it.

(and more, which mentioned the memory), which basically expanded the whole AI problem from one specific subset of cases where the AI cast beneficial spells on your creatures to the entire AI problem, such as not paying to untap Mana Vault or getting out from being under Black Vise no matter what, and I responded with the responding with the example you just gave and not future ones.

So basically, I see the breakdown in the third quote where you asked whether it should have been easy for the AI not to kill itself, because that's where you expanded the AI problems from one particular set of examples to another, whereas my initial reply was only responding to that subset and not all AI deficiencies.

As for the possibility of a glitch, that could be true, only Mountainwalk was mentioned, so not sure if it applied to all landwalks or just that one.

About patching, I think it's more a case of the publishers and the need to meet deadlines (like an upcoming holiday season). Bugs are more apt to be forgiven because they can be "patched" later. And this is not only for consoles, but "regular" computer games as well.

 

[theorgg]

Landwalk fails in the two-headed giant version of the game unless both of the players have the kind of land. That doesn't mean that just the person without the landtype can block, though-- both players can block if one of them doesn't have the land. It's really weird.

What I think happened was in the 'toggle' in the program for "Can block? / Can't block?" Mountainwalk gets triggered by one mountain as "Can't," then the other players' lack of mountain triggers the 'Can,' and they let 'Can' win instead of "Can't."

 

[Spiderman]

Wait, so the example you gave way back in post #9 was for THG?

 

[Oversoul]

So basically, I see the breakdown in the third quote where you asked whether it should have been easy for the AI not to kill itself, because that's where you expanded the AI problems from one particular set of examples to another, whereas my initial reply was only responding to that subset and not all AI deficiencies.

Huh? No seriously, I cannot understand this. I've read this paragraph three times and three times and I have yet to figure out what, if anything, it means. The rest of what you said made sense though, although I don't think anyone on the planet uses the word "stuff" when the intended meaning is as limited as possible. "Stuff" is always inclusive. I can't even picture it being used exclusively without it seeming comical. But that takes things off on a huge tangent.

About patching, I think it's more a case of the publishers and the need to meet deadlines (like an upcoming holiday season). Bugs are more apt to be forgiven because they can be "patched" later. And this is not only for consoles, but "regular" computer games as well.

Yes. I didn't mention PC games because I cannot remember when patches became available for them. I mean, I would assume it was within my lifetime, but I never noticed it happening. It's been that way for quite a while, though. Console games, though, have only been patchable relatively recently. Sure, there have always been deadlines and plain old mistakes and things, but I can't help suspecting that maybe being able to patch a game has led to developers being slightly less thorough than they would be if any errors in the final product would be irreparable.

 

[theorgg]

Yes, but the rules should still cover two headed giant, as it doesn't let you choose the player you attack-- you attack just the team opposite you.

 

[Oversoul]

You know, one would think they would take some extra precaution with THG, as it seems like a part of the game prone to that sort of problem. Programming that might normally work elsewhere could have issues when the rules are modified for THG and this would have to be accounted for. But maybe it is only obvious in retrospect or they really did try, but messed up. I don't know...

 

[Spiderman]

Huh? No seriously, I cannot understand this. I've read this paragraph three times and three times and I have yet to figure out what, if anything, it means.

That paragraph was an attempt to sum up everything before. Again, you started out with an AI problem of casting spells on your creatures. I agreed and said that compared to that problem (AI interacting with your permanents), coding a working landwalk ability ten years later should have been easy. You responded saying that shouldn't it have been easy 10 years ago for the AI not to kill itself, which totally switched the original AI problem of interactivity (and specifically casting spells on creatures) to a different set of situation where the AI would kill itself (of which no specific example were mentioned, so I assumed manaburn, although taking damage from Mana Vault came up later).

The paragraph was just summing up what was going on in the quotes preceding it and where the breakdown occurred.

 

[Spiderman]

More info on Duels of the Planeswalkers Expansion Pack 1

Duel the XBox Developers Oct 24

 

[Oversoul]

That paragraph was an attempt to sum up everything before. Again, you started out with an AI problem of casting spells on your creatures. I agreed and said that compared to that problem (AI interacting with your permanents), coding a working landwalk ability ten years later should have been easy. You responded saying that shouldn't it have been easy 10 years ago for the AI not to kill itself, which totally switched the original AI problem of interactivity (and specifically casting spells on creatures) to a different set of situation where the AI would kill itself (of which no specific example were mentioned, so I assumed manaburn, although taking damage from Mana Vault came up later).

Huh? You had me at first this time. But let's see. I guess the first thing I mentioned as far as the Microprose game goes was the Giant Growth example. So yes. And you did then compare that to properly coding landwalk, saying that the one should be easier, which I suppose you did say and I disagree with. I mean how many lines of code would it take to specify that Giant Growth is only for your own creatures? Anyway, then you say I responded by saying that it shouldn't have been easy 10 years ago for the AI not to kill itself. Wait, did I say that? Because that's the opposite of what I meant...

 

[Spiderman]

See, I disagree about the Giant Growth. Maybe there IS a time to cast it on your opponent's creatures - getting out from under Black Vise, for example, and you're stuck at 1 green mana.

You misread the part about the AI killing itself. It was part of paraphrasing your question; "Shouldn't it have been easy 10 years ago to code the AI so it didn't kill itself?".

 

[Oversoul]

See, I disagree about the Giant Growth. Maybe there IS a time to cast it on your opponent's creatures - getting out from under Black Vise, for example, and you're stuck at 1 green mana.

Do you disagree? Because for you to disagree with me and think that there is a time to cast Giant Growth on your opponent's creatures must mean that I think there is not. And I don't.

Since you mention it, the AI is laughably bad about Black Vise. It will do anything to get out from under it. If it has no creatures, it will play Giant STRENGTH (or Growth, but that one can work with no consequences) on your creature even though that would kill it faster. But that isn't what I was talking about. After all, it was, as you pointed out, over a decade ago that this game was made and getting an AI to know subtle details about prioritization is, I would imagine, rather tricky.

What I was saying was so bad about this Giant Growth example is that the AI will sometimes cast Giant Growth on your creatures when it actively harms itself by doing so. That's why it was the first thing I remembered. It was memorable when it happened (and it happened several times too). I'd be attacking with some creatures, and the AI would block so as to kill one of mine, then Giant Growth, saving it and killing one or more of the blockers. This is not just a bad play. It's a play that makes no sense.

Consider the alternative to this of just making it so that the AI would never cast Giant Growth on an opponent's creatures. Sure, it's not perfect. There's Black Vise and whatever else. But the exceptional cases are negligible. And for that matter, they could be factored in. Specify that Giant Growth is not to be used by the AI on creatures it does not control unless certain circumstances are met. That still seems like it would be pretty easy, no? But even without such stipulations, just the one limitation would significantly improve the realism of the AI and make it look not so obviously lame.

 

[theorgg]

The version I've been playing (Manalink 2.0 via VMWare Player) has the Giant Growth bug in it. Instead of using it on a creature that would be a GOOD creature to use it on, it hits the creature blocking that creature instead. I never saw that before when I was using the original game and Spells of the Ancients.

 

[Oversoul]

Interesting. So that problem, at least, wasn't always there.

 

[Ransac]

Do you disagree? Because for you to disagree with me and think that there is a time to cast Giant Growth on your opponent's creatures must mean that I think there is not. And I don't.

Hmm... I must disagree with him then. That's fine, because I love disagreeing with Spidey, especially over his choice of spandex when he decides to go out on the town.


Ransac, cpa trash man

 

[Oversoul]

Ransac...

You are at 1 life with an untapped Forest at the end of your opponent's turn. Your opponent has a Black Vise. You have five cards in your hand. You think you will topdeck the card you need to win the game next turn. You have Giant Growth in your hand and no other spells to cast and get your hand size down to four. You also have no creatures in play. But your opponent has a creature. What do you do?

 

[EricBess]

Oversoul...

If you are the computer, apparently, you wait until your opponent attacks you and then cast Giant Growth on one of his attacking creatures.

I don't think anyone was saying that there was never a time that you might cast Giant Growh on an opponent's creature, just that whatever algorythm the software is using to determine when that might be is flawed.

I suspect that much of the AI is based on a percentage chance of taking certain actions, and rather than trying to just figure when is the appropriate time, they just let random chance decide. Even a 0.1% chance will happen every 1000th time.

 

[theorgg]

This in the build I've got (which I have learned is a hacked version), it's not a .1 chance. It's almost a 25% chance if the computer blocks a creature with a smaller creature, they're going to overkill their tiny creature (or giant growth an attacker that was unblocked.

I don't know how often the Giant growth is in hand, so 25% might be too low.

 

[Oversoul]

I don't think anyone was saying that there was never a time that you might cast Giant Growh on an opponent's creature, just that whatever algorythm the software is using to determine when that might be is flawed.

Well, I think the consensus, including myself, is with you on the algorithm being flawed, but that sure isn't what I'm getting out of Ransac's post. I mean, I think he's just trying to be a pest, which is fine.

And yeah, as Orgg said, it happens pretty frequently. I mean, I can't say how often because I never really tracked it and it stands out when it happens, which might distort my perception of it, but it's certainly not a rare event. I don't know what triggers it. Maybe it happens randomly or maybe certain things cause it to mess up. Beats me.

Another bad thing about the AI, although I think we'd all (except maybe Ransac) agree that it's a trickier one, is dealing with attacks in general whenever its own creatures can die. I remember it basically being unable to calculate that if it didn't sacrifice some blockers, it would die for sure with my next attack. And even when it had a huge army that could overpower me, it would avoid attacking if I could block and kill a substantial number of attackers while still remaining at more than 0 life. Hopefully this new game isn't so bad about that, or it would be trivially easy to outsmart.

 

[theorgg]

This game ain't too smart, either. If you've got a creature bigger than the ones it could play in its hand, it doesn't play a creature. It just lets you swing in and beat down.

This is on the highest difficulty level.