Custom AI decks UPDATED 2013/08/27

[Psyringe]

Assuming that you mean the second deck mentioned on that page (the first one seems to be mentioned only or reference), for this specific deck I'll have to say no. The author says "The deck is extraordinarily hard to play" and then needs 468 words to explain how to get the timing right. Basically each of these words is another nail in the coffin that we'd build the AI with such a deck. The current AI has no concept of timing. Currently, a good AI deck is one that requires 0 words of explanation to get it right.

Second Sunrise probably won't work for the AI unless it gets hardcoded. There are several problems culminating here:
- The AI is incapable of intentionally building up a situation where it has to discard many cards.
- The AI knows nothing about the effect of "Second Sunrise" when it plays the card (or considers to do so)
- The AI has no idea how many cards were put into the graveyard this turn.

So, to make it work, you'd probably need to:
- make sure that the AI has to discard lots of cards most of the time, so that *if* it plays Second Sunrise, there'd actually be cards in the graveyard to put into play.
make sure that the opponent usually doesn't have to discard many cards, the AI will not check this.
- make sure that most of the cards that get discarded would actually be useful if Second Sunrise gets cast.
- and still rely on luck, because the AI will happily shoot itself in the foot by discarding Second Sunrise itself.

I think it'll be very difficult, if at all possible, to build a deck that meets these conditions. It may be possible to write an AI that can play such a deck, following instructions given by the deck builder, in the future. I certainly hope that Wagic's AI will reach that stage. For the time being though, the AI wants much simpler decks to play with.

[jeremy]

guess for now we can only kick the AI with this deck on our own, i'm currently making a variation of that deck, with limitations due to no 2nd sunrise. Open the vaults works but just not as well. Guess i'll take out more lands in order to use one of artifacts with the cantrips like lotus petal.

[jeremy]

Made an updated version of the open the vaults combo deck later in the thread.

[Psyringe]

other than having to add a {s} instead of {t} for 5th dawns Krark-Clan Ironworks to fix it

Shouldn't that be simply nothing instead of either {s} or {t}? (In any case, thanks for spotting the bug!)

[jeremy]

ah, i was a bit confused because I was coding in a black lotus for my own version of open the vaults use. Still trying to figure out how to code in suspending.
theres still a bug bugging me tho, Mogg Sentry, I can't seem to make the +2/+2 temporary and only spells.
instead it activates permanently and whenever an opponent plays spells n puts down lands.
it becomes too broken for play pretty quickly at 1 red mana.

[Psyringe]

Still trying to figure out how to code in suspending.

Not possible afaik until somebody programs it into the engine.

theres still a bug bugging me tho, Mogg Sentry, I can't seem to make the +2/+2 temporary and only spells.

I haven't looked at the code (or the card), but that sounds like 2 known bugs in Wagic 0.8.1:
a) lands use the stack (and therefore count as spells)
b) some "until end of turn" effects are permanent

Both are fixed in SVN already, so the bug should be gone in the next release (which unfortunately may stil ltake a while due to other bugs cropping up). You may be able to fix (b) temporarily by adding " ueot" to the end of the auto= line.

[jeremy]

Made a proxy Lotus bloom, by only using it 3 turns after I drew it. and tested the deck out with bloom inside with some changes like adding 2 fire becons to prevent accidentally topdecking yourself while sacrificing stuff to ironworks, which still trigger their draw effect. Most times I get the combo out n draw my whole deck on the 4th turn, usually on average 5th turn seeing as sometimes your mana scr..ed. After you get the combo started its pretty easy to sling a 125 damage blaze to the computers face or if you want to humiliate it, you can pull it off a few times n get a hundred life or so by comboing till all 8 open the vault type spells are used up then Reminisee and do it all over again. You can go nearly 300 damage if your careful not to deck yourself, never really tried tho.

[jeremy]

do you think we can handle suspend and cascade by
writing it in a card that it has to be tapped 3 times before it activates, like maybe it comes into play with 2 +1/+1 counters on it, is unable to attack or block and is not targetable by effects till it activates. Like when it has no more +1/+1 counters and everytime you tap it , it removes one counter, then it triggers as comes into play?

As for cascade how bout having the card make like a random version of demonic tutor that casts a random card in deck that fits the mana amount.
Like being able to cast Bloodbraid Elf which costs 4 mana and it randomly searches the library and hits say a 3 mana Imperious Perfect and casts that into play.

Do you think thats codeable with our current system?

[Psyringe]

(Btw, we should ask Niegen if he'd prefer these matters to be discussed in separate threads, our latest mails could be seen as a thread hijack.)

"Suspend": Your idea is an interesting approximation of the Suspend effect, but not without problems. You can't change the fact that the "quasi-suspended" card is in the wrong the zone (it'll be in the battlefield while it *should* be in exile according to the rules. This has several unwanted effects:

- A permanent with a rule like "You get +1/+1 for each white permanent on the battlefield" (or similar) will receive a bonus for the quasi-suspended card. Or (different example) a quasi-suspended creature will keep Pestilence in play although it shouldn't.
- The card would use the stack while you quasi-suspend it, so any triggers associated with that would activate. Example: "Whenever X comes into play, -this- gets +2/+2 until end of turn."
- Non-targeting cards like Wrath of God, Armageddon, Nevinyrral's Disk etc. will affect the quasi-suspended card even though it *should* be safe in exile. This is also true for other non-targeting cards, e.g. Heartmender or Harbinger of Night.
- *If* any of the cards that affect cards in Exile gets implemented, you'll have the same problem there, just the other way round.

Your suggestion also gives the player a *choice* when to play the suspended card (he could stall indefinitely by simply never tapping it), which real Suspend does not.

I'm not sure whether the tap / counter-removal / trigger when no counter left workaround is codeable with the current card code ... others know more about this than I do.


"Cascade": I don't think there's currently a way to determine the mana cost of a card in the library. Your workaround also misses the part of the cards that get first exiled, then shuffled, and then put back under the library.


In short, as said above, both of your suggestions are interesting ideas that approximate the effects of the real thing. Whether this approximation is close enough to justify the work of coding these cards, is the decision of the person who wants to code them.

From what I see from the devs, I think the approximation is too far off the mark for something like this to be included among the officially supported cards - they'd rather implement the "real" feature instead of spending time on potentially problematic workarounds. But that doesn't have to stop you from coding these cards nevertheless if you think otherwise and find a way to do so.

[Raziel017]

hi jeremy,

i checkd out the decks works fine.new challenge.i like it!also i was wondering can you also provide avatars for them?for maximum satisfaction thanks in advance.