Gem matching unsatisfying.

[Deleted User]

So this is a gem matching game; but, it feels highly unsatisfying now that the AI cascades a billion while i sit there matching 3 gems... oh but wait sometimes i get a double match cascade! Have you guys changed the formula of gems that fall from the unseen area? It feels highly favorable to the AI.

However perhaps this is due to the lack of efficiently costed gem changing supports that are actually easy to obtain in lower rarities? Those mythics are really some of the few gem changers that are worth it, i feel. 

The new changes have slowed the meta down. Okay good. But were you being conservative because it feels a little too slow. 

FindingHeart8

The pterodactyl elder-dino boss actually was able to cascade consistently enough to where he was able to activate his ability (give target creature you control double strike, vigilance, prevent damage) 5 times in a row.  No exaggeration, that's a 9-cost loyalty ability, and the AI activates it every turn it can so it had less than 9 loyalty before the 1st cascade.

I know it was probably super improbable shoddy luck but still...it's a low rank boss, very demoralizing to have 5 waves of creatures slaughtered by a double-strike, cascade-sponsored Runaway Carriage.

ouch.

Mburn7

This has been debated fiercely here, and so far nobody can prove that the AI is cheating.

That being said it feels like its cheating, and that is just as bad.

DumasAG

I still firmly believe these mega-cascades are a product of observation bias. The game isn't cascading any more than usual (in fact, less now that that the AI was dumbed down to not make optimal matches). Yesterday I hit 3 big cascades on turn 1, so I know it can be done.

The problem with the gem changers is that they're not even the most efficient way to produce mana anymore EVEN IF you could cast them turn 1. If you're not running storm the vault (and thus, you don'y want land supports anyway) then you're doing it wrong.

Unknown

This content has been removed.

Mburn7

Dusty said:

jimpark said:

However perhaps this is due to the lack of efficiently costed gem changing supports that are actually easy to obtain in lower rarities? Those mythics are really some of the few gem changers that are worth it, i feel. 

But I'm constantly gobsmacked by the fact the nobody will start a thread about how broken Storm the Vault is.

Shhhhh.  We're hiding it from the hammer.  Maybe if nobody mentions it, nobody will notice.

Laeuftbeidir

Mburn7 said:

Dusty said:

jimpark said:

However perhaps this is due to the lack of efficiently costed gem changing supports that are actually easy to obtain in lower rarities? Those mythics are really some of the few gem changers that are worth it, i feel. 

But I'm constantly gobsmacked by the fact the nobody will start a thread about how broken Storm the Vault is.

Shhhhh.  We're hiding it from the hammer.  Maybe if nobody mentions it, nobody will notice.

It's
 kind of an unspoken, silent agreement, since everybody loves the card

Kinesia

Whether or not this is really happening, it's a HUGE perception outside the forums. I run into people thinking things are so extreme the AI is cheating all the time and it's so frustrating because I can't prove them wrong, I can't give them anything to reassure them...

This needs an _official_ article posted in the game message thing to WIDESPREAD innoculate against these conspiracy theories. This kind of broad perception should not be up to players to try and fight against.

wereotter

This weekend I had Elenda kill me in 4 rounds thanks to massive AI cascades. It was maddening.

The topic is indeed controversial, but I think at the moment the AI is smarter at predicting cascades than the average player, and not as smart as an expert player leading many in the latter group to complain that the game isn't challenging enough while others complain that the AI is too smart.

There's no chance everyone will ever be happy about the difficulty, unless maybe they let you set the AI difficulty level, in which case they'd have to remove all forms of ranking from the game, which would in turn upset another group.

Volrak

There's been a small amount of data collected on cascades (24 games so far) here.  The trends so far seem to be:

* Players tend to get more matches/cascades per turn than the AI

* Individual playing style and colour balance of your chosen PWs seem to be factors

If anyone feels like shedding more light on this question then adding data of your own would be a great way to do that.

Kinesia

I thought I was pretty good at matching things (and definitely better than when I started!) but watching Babar's videos it's _very_ clear that better matching on the human side leads to more cascades.

It's also a question of priorities... Sometimes you need to sacrifice the highest guaranteed mana for a higher chance of a cascade. Greg's priorities include mana-screwing the player so he doesn't match the same as most beginner players to start with, he's not aiming for the same things.

Kinesia

But, yeah... I think the _perception_ of the problem is a far bigger issue and risk than the problem itself...

MTG_Mage

One major factor is 

The Negativity Bias: refers to the notion that, even when of equal intensity, things of a more negative nature (e.g. unpleasant thoughts, emotions, or social interactions; harmful/traumatic events) have a greater effect on one's psychological state and processes than neutral or positive things.In other words, something very positive will generally have less of an impact on a person's behavior and cognition than something equally emotional but negative.

(definition taken from wikipedia)

So people will always remember when the AI gets cascades and does better as a result, and completely disregard how often it happens to their benefit.

Unfortunately gem matching in this game directly relies how the AI generates the gems falling from above, and the selection of which color gem is chosen is done by this games ability to use RNG (random number generation).

Anyone who has been playing this game for any length of time can absolutely agree that there is nothing random about the RNG in MtGPQ.

I have mentioned how obviously flawed the RNG is in posts dating back to the first few months after release.

It is quite frustrating how often in the very same pack there are two of the exact same card pulled.
I recently opened my monthly reward super pack (3 packs of 5 cards) and there was no less than 3 sets of 3 of the same card! The RNG bug is everywhere in this game including card draw. What are the chances of playing several long games in a row and not drawing two or three of the cards in your ten card deck! Or getting the same exact card from an elite pack several times in a row. 

I am certain that the programmers coded the RNG with a dependance on the last item selected (or a very small pool of last few items) so the likelyness of that item being picked is like a 90% more bias towards the last selection than from the whole pool to choose 'randomly' from. 

Please note that I am completely aware of negativity bias and have been playing this game since release.

This should be the highest priority on the 'bugs and technical issues' since it affects all aspects of the game. 

It may be too far gone as the bug is likely in the source code and compounds and gets worse with each new release.

Theros

From my own experiment and what some have suggested in the past, never match gems suggested by the AI unless there are no other options. If you do so, AI cascade next turn.
My theory is that we randomly match gems the AI would've suggested.
just an observation you can try in TG

ZW2007-

I'm not sure about AI cheating but I am very convinced of a rubber band effect. If I hit a huge cascade, the AI is more likely to hit one on it's next turn and vice versa. You just need to watch more closely for them because after the whole board has changed, it's easier to miss the optimal move.

Unknown

This content has been removed.

Stormbringer0

I notice the same problem with way too less matches in my PW colors for weeks now. The problem isn't that the AI cascades like a maniac, because if you can do so too, it's a lesser problem and you can keep up. The gap between you matching three gems of a color which isn't your PWs and the AI matching colors of their PWs (not even cacading out) is too big at some time of the game. I've lost more games to that problem rather than the AI cascading and steamrolling me.

Another problem is that the players are playing way more gem convertion which limitates them in deck building to some extend.

span_argoman

wereotter said:

This weekend I had Elenda kill me in 4 rounds thanks to massive AI cascades. It was maddening.

The topic is indeed controversial, but I think at the moment the AI is smarter at predicting cascades than the average player, and not as smart as an expert player leading many in the latter group to complain that the game isn't challenging enough while others complain that the AI is too smart.

There's no chance everyone will ever be happy about the difficulty, unless maybe they let you set the AI difficulty level, in which case they'd have to remove all forms of ranking from the game, which would in turn upset another group.

So is this a matter of player skill (or lack thereof) in gem matching leading to the perception that the AI is cheating? Is there any good way to dispel that notion if it is? Cause it seems like every once in a while someone comes along to make the same claims.

No doubt there are the standalone matches where either side (player or AI) gets massively more cascades than the other side even without gem converters or board destroyers. But people seem to be insinuating that there is more to this.

On a side note, deck design also helps to mitigate AI cascades by filling your deck with cards designed to give you a card advantage.

TIMEWARP

This is why we need data.

The Battle Log would be an ideal place for this. the Devs would greatly help us and take away any thought on AI "cheating" if they could give us tallies of 'total mana gained' 'total loyalty gained, 'total cards drawn'. The AI needs a little leg-up because it is not as 'smart' as a player, so PvE events generally have characters with high starting life and high mana bonuses to compensate. This is fine. The AI should not however need proportionally better gem-matching results. RNG is part of the game, granted, but any hint of bias should be able to be disproven.

Volrak

MTG_Mage said:

Anyone who has been playing this game for any length of time can absolutely agree that there is nothing random about the RNG in MtGPQ.

The underlying RNG algorithm is sound according to folks who've looked at the code; images have been posted representing hundreds of thousands of random samples generated by the code which look like noise (as they should); this contrasts with visible patterns in dodgy RNGs.  That doesn't guarantee that the usage of those underlying random numbers must necessarily be unbiased, or that the RNG is being seeded correctly, but it rules out one possible problem.

Nobody's exhaustively studied each way the mtgpq RNG is used, but the data I've looked at (mostly to do with card drops) hasn't revealed signs of unexplained RNG behaviour.  If there's a real issue that's truly perceptible above and beyond the effects of negativity bias, it follows that we should be able to collect representative data from which the issue can be demonstrated statistically.

MTG_Mage said:

It is quite frustrating how often in the very same pack there are two of the exact same card pulled.

Even though I think how often it happens in the long run matches how often it's expected to happen, I agree that it can be quite frustrating!

Volrak

Stormbringer0 said:

I notice the same problem with way too less matches in my PW colors for weeks now. The problem isn't that the AI cascades like a maniac, because if you can do so too, it's a lesser problem and you can keep up. The gap between you matching three gems of a color which isn't your PWs and the AI matching colors of their PWs (not even cacading out) is too big at some time of the game. I've lost more games to that problem rather than the AI cascading and steamrolling me.

A tip on gem matches is that you can totally expect colour issues over the course of a game unless you take active steps to prevent it, especially if you run PWs which have strong gains on one particular colour.  That's simply because the gems dropping down to replenish the board aren't weighted towards your colour, whereas the gems being destroyed by you and the AI are (as has been said, the AI does make matching your colour a priority).

Apart from gem changing effects, steps you can take to keep your colour healthy include preferring moves which make two off-colour matches rather than one on-colour match; avoiding matches which would leave your on-colour gems lined up for the AI to match; and preferring matches with higher chance of cascade (e.g. go for matches which drop new gems into positions where there are already 2 adjacent gems in a row).  In all cases except making two matches in the same move (and that excludes the case where the second match is a visible cascade), the AI ignores these tactics.  So they are all ways in which we can outplay the AI, and reasons why we are able to get more matches per move on average than the AI.