Bug #531 , "routinely misuses is FUD" and AI priorities

[lujo]

Ok, so after checking out the responses to my reports today I was irked by the contrast between two things - the nonchalant accusation of my "consistent misuse" reports as "marketing FUD", which I take to mean as nebulous/unfounded/slanderous/lying , and at the same time "can't be reproduced" on a bug which seems to affect cretures with both soulshift and sacrifice abilities (I reproduced it in a matter of minutes and resubmitted a ticket with all but a video of it).

Sure, I did submit a bunch of things on the off chance that some routine misuse might be easily fixed, but there are too many things wrong with the dismissive attitude towards something like "Diminish is routinely misused". Just put the damned card in a deck, give it to the AI and you'll see it cast on 1/1 creatures, and generally cast at anything at all at the first opportunity. It's completely unusable.

Now, this wouldn't bother me if at the same time there weren't cards which aren't in the pool because "the AI would be bad with them". What's Diminish doing in the pool then? If I took some time to find a screen recorder which worked with XP I could just tape endless hours of cards which shouldn't be in the pool on those same grounds. Yeah, maintenance on certain cards that don't work right would be a pain in the ass, but there's not that much maintanance done on cards that don't work for the AI but are in the pool.

I don't even personally care that this or that isn't in the pool, what gets on my nerves is inconsistency and what gets on my nerves even more is that there's a semi nebulous (on the user/tester end) bug #531 everything gets pinned on.

I don't understand bug #531 and from what I've read about it neither does Korath. But my reports can get labeled as bulls**t , and I can give you video proof that they aren't , while just about anything labeled as "due to untraceable and unfixable bug" can actually be anything from "someone messed up the AI card evaluation system either when originally coding it or later" to "AI is using a ridicuolus system for card draw which messess up card draw" or or as the ticket on that bug itself says "If I'm wrong, I've probably just made the combat AI even worse."

What if you have? What if your card drawing system (or the original one if that one's still there) is behind systemic AI incompetence with almost anything related to card draw? Because it actually looks that way. What if your (or original) exclusive decisions on element evaluation schemes for the AI are behind most or at least some of the many systemic nebulous misplays? Because last time around something like that was exactly the thing, everything was set to be played after combat which made a huge number of things set to always not work.

How about we clear that up? Because I or anyone else, will just keep running into things which warrant reporting, and we will not be telling bullshit, but it'll just get shrugged away. This isn't about any concrete report or complaint, it's about honesty and nor being asshats.

What is the Bug #531, exactly?

I've noticed a consisten pattern of the AI using (some?) sacrifice outlets, specifically Homarid Spawning Bed and Stitcher's Apprentice , wrong. The AI sacrifices things to them in it's own turn, most often during combat before damage is dealt, meaning it routinely wastes damage , opportunity and god knows what else.

In short it attacks and then sacrifices the attacker mid-attack before blockers are even declared. Will 3 seconds of thought be given to how to fix these cards or will it be handwaved away with nonsense?

Because I can handwave it with nonsense myself - bug #531 hur desync dur sacrificing and tokens and combat oh my. But it doesn't have to be that, when creatures have summoning sickness it tends to sacrifice them before combat. It's just trigger happy with sacrificing when it's proffitable + the ability is likely set to be usable before blockers in order to facilitate chump-block-into-sacrifice but the AI doesn't know it's only supposed to do that on the other guy's turn and only sacrifice it's stuff post combat in it's own turn. Way more likely than boogey-bugs.

How do I report this in a way which might result in it getting fixed? I report a bug I've seen a 100 times (literally, I have 3 decks with soulshift in the rotating pool at the moment), piss-easy to reproduce, happens to affect a number of creatures - "can't be reproduced". Not even "give a decklist if you want me to check it out". Another report later "all his routine misuse reports are just for bug #531".

No they're not. They might be clumsy and unhelpfully phrased or even stupid, but they're legit reports for legit problems. And if they are all by some miracle connected to some bogus desync or bug or whatever, then that damned bug is causing pages of reports. If you can't be bothered with or don't have the time to fix it you probably know someone who could be persuaded to do it. Let's put some money together and hire a guy to fix that thing.

[Aswan jaguar]

Most of the bugs you mention have to do with the bad targeting of AI which already is known,known for many many years.It is totally in vain to post bugs about different cards that are miss-targeted by AI it's still the miss target issue.Original developers have made targeting AI intentionally to make stupid decisions in lower difficulties.
It is also the AI de-syncing which Korath has told us about and already has made efforts to fix it.
1- If you like less AI miss targetting use always the top difficulty. AI at top difficulty does a lot less miss targeting than lower difficulty levels.
2- Also you can use the full AI thinking it also helps making a little better AI but making gameplay really,really slow.

Now if you already do use the top difficulty and you are experiencing more miss targeting than you used to in contrast to earlier korath releases or the original game then that is a behaviour Korath would like to know as mentioned in his releases.

The same goes to AI sacrifice decisions there are already some reports about different cards about it but still the root problem is that AI makes poor decisions on that.It is futile to post all the cards with sacrifice as bugged.

About the soulsift bug you mention maybe Korath has fixed it already while fixing something else that is why he can't reproduce it.Don't get frustrated in such a hurry.You will have the chance to report it again if there is still an issue.I didn't test it myself and I don't know.

[lujo]

It is totally in vain to post bugs about different cards that are miss-targeted by AI it's still the miss target issue.

First of all it isn't. That IS bullshit, and whoever is saying that is lying either to themselves or to other people or just plain doesn't understand the game well enough to be able to tell what is going on and what is supposed to be going on.

- A bunch of misplays I report in general (and way more that I don't report) come directly from the card/ability/drawing/life/other element evaluation system which causes the AI do do wrong things at wrong times.
- Even more come from the absoultly absurd method the AI uses when determining whether or not to draw a card or scry. Not the pririty it places on drawing but the abusrd, ridicuolus and obviously wrong method of evaluating card draw by the value of the cards about to be drawn
- Also, many of them come from there not being proper enough directives about when to use a certain card or not, whether it's the correct time in a turn (so wrong or insufficient phases) or the desired result (don't throw this away unless it kils something, for example)

And if I report mistargeted cards I report them because, yes, there is a chance that it might be some kind of nebulous AI quirk, but there's also a chance that there is a bug with this card and in this case I have to make a note of it to not bloody put it in decks as it's going to be misused. Or all it takes is really just re-flagging the phases when it can be used/activated.

This forum seriosly ought to take a cue from the Forge board and have both a "known issues" list and a "AI improvement requests" one. The most important things about building decks for the AI in Shandalar is knowing which cards to not put in, and which cards you must put in because the AI plays in patently retarded ways unless you give it the exact right things. Why the devil are we forced to pretend it's otherwise? Bugs and cancerously godawful mishandlings are all over the place and they don't go away just because someone won't or can't acknowledge them.

Original developers have made targeting AI intentionally to make stupid decisions in lower difficulties.

And this was a monumentally stupid decision on their part. And if for some reason this is still in place I'm not aware of there being an option to set the difficulty in infinite duels mode or even checking which one it is.

And another thing about this is - I make decks for the AI, but I have no bloody idea what the people playing them are going to pick as their difficulty because it affects seiges and starting pool and whatnot. There is no point trying to make decks at less than max thinking because the AI needs all the thinking it can get, but why the hell would someone play my decks if the difficulty choice is going to ruin them? I've played MtG against a good number of obviously stupid, like very low IQ, people but even they didn't mistarget their stuff.

It is also the AI de-syncing which Korath has told us about and already has made efforts to fix it.

Let's keep that de-syncing away from my reports. I'm reporting stuff because it doesn't work in hope it gets fixed. If there was a proper known issues sticky I'd be posting stuff that Korath won't or can't fix (or fix at the moment) there because then it's even more important to be aware of it.

If desyncing is causing problems so regularly then prove me wrong and solve the problems I'm having by fixing it. I'm somewhat sure it won't, because I say that Korath for all the things I could and do praise him for is dead wrong about some things he did seem to fiddle with. And he really should stop painting me as god knows what and being childish about it.

"Routinely misuses" is what it is, "Routinely Misuses". Energy Tap is cast, all the time, on the first possible target. If Energy Tap is cast on a creature which costs 1 mana, that is complete lunacy and Energy Tap should not be in the pool because the AI criminally mishandles it. Or it should have a clause which tells the AI to bloody well cast it on stuff that cost more than 1 mana. There would still be problems with it, namely the one of madly spamming it, but that would actually be workable as it wouldn't, as a rule, be cast to do nothing at every opportunity. Should this be reported? Hell yes, because if "AI can't handle it" is a big enough thing to not implement cards, then wtf is this doing in the pool? And if this can be fixed, then I'd be mad to not ask if it will be, meaning report it.

There's no f***g desyncing going on anywhere near that, it's "AI can't handle spells wihtout very specfic and limited casting conditions" combined with not well thought out or even just hasty impelmentation. If it worked any better it + Spire Golem would be in at least 1 deck in any Shandalar configuration because it's a rare bit of blue ramp which opens up a million possibilities. Whether Korath thinks that's worth fiddling with, or even just changing the schedule, that's his thing - but I'd be an idiot for not reporting it.

2- Also you can use the full AI thinking it also helps making a little better AI but making gameplay really,really slow.

It doesn't make "a little better AI and make gameplay really, really slow", that's bullshit. I put it back to over the original thinking time every time I patch (well I do forget but then remember when really stupid plays keep happening). It doesn't make gameplay much slower unless you go a lot above the original thinking time or have a really old maching, I suppose, but reducing it ten time and then eyeballing the quality to be "acceptable"... it is just hubris.

The same goes to AI sacrifice decisions there are already some reports about different cards about it but still the root problem is that AI makes poor decisions on that. It is futile to post all the cards with sacrifice as bugged.

Again, this is plain wrong. I post about Archfiend of Depravity, somebody who just repeats mantra tells me "AI just sucks what can you do" and I have to explain that "No, the bug is this and this" and lo and behold "the bug is that and that". I have to report every single one where I can tell what's going on somewhat, because I don't even report the ones where it's obvious to me that it's one of the standard retarded things (so much so that I don't even bother using cards which might cause it).

About the soulsift bug you mention maybe Korath has fixed it already while fixing something else that is why he can't reproduce it.Don't get frustrated in such a hurry.You will have the chance to report it again if there is still an issue.I didn't test it myself and I don't know.

I know, but every time I see the "will not fix" I have to wonder if it's a "fuck you personally for reporting something I can't be bothered with" or "allready fixed" or "it's gonna be fixed with a bunch of other things at some other time". I report stuff because it's not working right, but if it's gonna stay broken I have to take it out of decks and either put other things in or come up with other decks. The "can't be reproduced" is particularly insulting if it just closes the ticket, when you KNOW it can be reproduced, how it can be reproduced, you have 3 different decks reproducing it, and if someone payed you for demonstrating it on stage you'd make a steady living off displaying it.

Meh, this is counterproductive, I'm just cheezed off. Just stop with the "Nothing can be done" and "It's all bug #531" nonesense, it can and it's not. And don't put me in the damned gitlog because I asked for cards that someone doesn't feel like implementing because "I'll have to rewrite them when I add splice" if over half the pool is effectively unusable for the AI anyway. Don't add them, fine, or don't fix problems or ignore bug reports, but don't be childish about it. I'm in the right to make the vast majority of these reports, and it takes me more time to write them than it takes people to clear them and my time is actually as valuable as yours is.

Eh, I'm already sorry about this.

[TheKillerNacho]

[lujo]

Btw, believe it or not, this was useful.

A few weeks back I acidentally discovered bug #1169 . It means that infinite duels mode can't start unless you have a Shandalar savegame.

I went and really cranked up the thinking time until it did cause slowdown, but it didn't cause slowdown in infinite duels. It did get it to cause slowdown in Shandalar, so I went and overwrote the savegame with a Wizard difficulty one. This got me the slowdown in the infinite duels mode.

So apparently, and I suppose logically, the savegame affects the parameters under which infinite duels operates. "Infinite duels" is I guess a hack which lets your savegame character fight Arzakon but adjusts your deck and his, that would make sense.

Well, now we know how to do adjust infinite duels difficulty, and previously we didn't.

But then choice of difficulty affects thinking time regardless of what you set it as in the .ini? This is terrible because noone wants to play Wizard at least because of the seige time, while there's no real point playing a difficulty where the AI is set to mess itself up. The difference in play quality is very noticeable.

Also, the way shandalar -e and -p determine difficulty seems to be by "resume game".

Does the fabled desync actually happen at higher AI thinking times? What other invisible differences are there between difficulty levels, and why are they even there? Initial scrambled cardpool for the player and lives are all the difficulty there really needs to be (in fact that's way too much, just starting with a pile of cards is plenty). Making the AI "dumber" just makes cards and decks malfunction.

Also, playing at apprentice is not something you want to do anymore even if it didn't cause enemy decks to malfunction by default. Starting mono-colored means more dead cards because you seem to always get gold cards and nothing is filtering colored-mana artifacts out of the pool (they're not artifacts in the tradition sense, but it hasn't been taken into account), while at the same time you only have one color to rely on. Starting wiht 2 colors and having some shot at using your multicolored cards (which are often strong) makes apprentice unintentionally harder. It can, and often will, screw you out of being able to use your starting rares and uncommons. I've been harping on this a lot and the initial cardpool distribution should've been fixed ages ago because a small child could see what's wrong with it. It's basically one of the things that if it was adjustable would be re-adjusted at every re-patch by everyone who meant to actually play Shandalar becuase they way it is now is just wrong. I'll make a formal request for it, it should've been done when a large number of multicolored cards and artifacts with colored mana costs were introduced to the pool.