Feels like AI is cheating?

Mburn7

Volrak said:

madwren said:

my best approach (both during deckbuilding and actual play) involved the mitigation of bad luck

This, more than anything else, is how to reach the top levels of play, and I like the succinctness with which you've put it.

I agree that good deckbuilding can mitigate bad luck, but sometimes there is just nothing you can do.  I don't think there is a single player here who hasn't lost a game to the AI getting a stupid turn 1 cascade that wins them the game because they get stuck with 4 copies of a non-helpful creature and no good matches.

Those losses feel awful, and the more often they happen the more rigged the game feels.  All the (true) explanations of bias and sample size won't eliminate that unfortunately.

Endbringer

SuperCoolGeezer said:

There are a few theories I have about the actual AI difficulty. These are repeatedly occurring trends/observations I have noticed.

Let me preface by saying there is definitely a "hard mode" AI and a "easy mode" AI. Lately, I have observed a trend of "flexible" or hard/easy combination mode AIs. Furthermore, the AI for my account has been getting progressively more difficult (which to me is a huge annoyance and burden, as it elongates my games). NOTE: I play this game to relax; I don't want a constant challenge; if I wanted a challenge or stress out, I can go play Arena.

Another thing before we start is... there are a TON of variables:
- Does player lvl affect AI difficulty and/or occurrence?
- Does VIP vs non-VIP affect AI difficulty and/or occurrence?
- Was the most recent, "no cards work" bug linked to AI difficulty? (based on its random occurrence, no relation to specific cards, only player-side affected, etc)
- Is there a specific pattern/rotation that we face "hard mode" AI?
- Is the AI programmed to specific events?
- Player PW vs AI PW selection (certain things will be more noticeable depending on how much overlap or opposition there is with on-color/off-color mana bonuses)
- What exactly is "hard mode" AI? Is it really the AI is hard or is it other surrounding factors?
- Actual deck AI is playing
- etc, etc

1. The AI difficulty setting appears to be different from event to event. Perhaps we should call it behaviors. AKH AIs seem to behave a bit more aggressively in their gem matching that show a noticeable difference when compared to other AIs. Perhaps we could say the AI is programmed an extremely specific way and is less flexible. However, I have a feeling some AIs are locked to events while others may not be; I cannot find a definitive trend.

2. If you watch closely at the AI casting gate lands; you will notice the AI has higher success in getting matches from fewer gates/reinforcements than if you were to have more gates/reinforcements and/or other land converters. HOWEVER, this is based on the AI difficulty settings. What I mean is that if your AI is on "hard" mode you will notice this; however, if your AI is on "easy" mode, you wont notice this.

3. If you encounter a "hard mode" AI and you don't have any gem converters, you will see a stark difference in your potential gem matches vs the AI gem matches. The AI will have more favorable matches. (Honestly, just playing without some gem conversion is usually pretty miserable)

4. If you encounter a "hard mode" AI. There comes a time when you have a gem board layout with very very few potential options. At most you will literally have only 1 or 2 potential gem matches. In many cases, these gems will not be very favorable to you, i.e. not your colors, provides an extremely good set up for the AI (this one is more pronounced when you literally only have 1 gem match option, period). This is more rare and you really have to be more observant to notice it; I also believe it is more prevalent during coalition events.

5. Flexible/Combo AIs. There is a off the bat "hard mode" AI who will cascade like crazy and be very aggressive. Strangely this AI will noticeable switch to an "easy mode" AI after a set number of lethal turns. If you can ride out this wave, you will win.

6. Flexible/Combo AIs #2: This Al will only switch to "hard mode" under certain circumstances. One of the more consistent circumstances is low AI life total. Almost like a last burst of glory, the AI will literally suddenly ramp up and hand dump if its life total is very low (This is a very rough guess but from about 30-40% life remaining). This is more pronounced if you have a very bad, slow deck.

7. The starting planeswalkers (gid1, jace1, lil1, nissa1, chandra1) appear to cascade more than other planeswalkers. It appears to be a handicap system? The good thing about this is that it appeared to be working both ways. If I play or if the AI plays, there appears to be more cascading.

8. Very recently, the AI has been getting very odd 5-match cascades. It has become much more prevalent than in the past. I can't even explain it; but, I just couldn't believe how many times I have seen the AI cascade out of nowhere. The funny thing about this instance is.. it's usually an off-color match (depends on what colors your PW and the AI PW is) and if you take it; you will see the most ridiculous mana cascading ever. And its usually one of those matches, where you definitely have matches in your colors available but no matter which combination of many on-color matches you choose, you can only get a single match - it is near impossible to match 2 sets of gems (the gem board is laid out like that). However, on a whim, if you take what you believe your AI opponent will take.. bam.. rewarded.

9. The gem board has specific rotations. There are definitely specific games where the gem board is not in your favor or the gems dropping from the "sky" are definitely the colors of your opponent. Much more pronounced when you play a PW that directly opposes the AI PW. So if you are Bolas, and the AI is Huatli2; it may be hard to notice if you get a gem converter out; but, you will notice there are very few gems in your colors and there appears to be a higher drop rate of your AI's PW colors. Another way to see this similar kind of thing is to be playing matching PW colors (so both you and the AI play Bolas1), there are games where you will notice the gems are shifting away from the Bolas primary colors and a bunch of white and green gems are dropping. In this instance, you win if you get the last on-color match; however, if you don't you could be in a world or trouble and you must immediately prioritize off-color mana cascading to speed up the off color gem dropping phase.

10. The AI gets more difficult during specific times, I have found the difficulty more pronounced the day preceding and during coalition events. But this was more of a "bleeding" difficulty, as events surrounding the coalition event suddenly appeared to get more difficult. The problem with this is I can't tell if its actually a rotating system (i.e. you play x games, then on your nth game you face the "hard mode" AI) or if its actually slotted for specific events. Very hard to isolate this trend

11. In coalition events, you will face the "hard mode" AI in at least 1 game. If you survive and win this, you will get a perfect score. This was in the past when I played competitively; lately, I'm not so sure its only 1 game anymore. It appears they increased its frequency a little.

etc etc

Well put. I agree entirely and have had many of the same experiences and knew something fishy was going on. There are still skeptics on here that will argue the fact, but they simply mustn't be paying enough attention.

Volrak

Anything that's really a pattern and not noise is measurable.  If your goal is really to prove anything to anybody, measure it.  Collect the data.

If the goal is just to vent, you can be most likely to vent in peace if you use what has somehow become the longest running thread on the mtgpq forums: https://forums.d3go.com/discussion/70877/the-safe-rant-thread-for-letting-off-steam#latest

But in either case I would suggest to critique ideas, not people.

My own comment about the above thoughts is some of them read like "I've noticed the dice roll more odd numbers when it's rainy outside" - maybe absolutely true in some individual's experience, but unlikely to be repeatable.  Others are like "Sometimes the dice switch to hard mode and roll lots of 5s and 6s" - such trends can certainly be observed in randomness, even when there is no underlying structure.

SuperCoolGeezer

madwren said:

jtwood said:

madwren said:

People feel it's cheating because the skill-to-luck ratio is different than what they expected--especially for anyone who's played paper Magic, where the game proceeds in a very orderly fashion.  

You can make perfect decisions on your side, but if Greg gets a 60-mana cascade, dumps his hand, and leaves you with three turns of loyalty matches, then it can be hard to overcome. 

Imagine that in Standard (not Vintage) paper Magic.  "Play an island, tap for 1 mana, summon merfolk, go." "Play a mountain. Tap for 40 mana, summon EVERYTHING." 

I think this is a good summary of the perception issue. I lost a match yesterday where Greg dumped his hand, which included Quartwood and Bolas Vanguard. I was using Song and discarding down to 1 card each turn. Bolas exiled it. I never got the Song train going again and lost in a couple turns, because Quartzwood will do that.

The reason I lost is in some part due to Greg getting a lucky cascade (the deck had no gem changers in play or in hand). But the other reason I lost - and the reason I was mad about that loss - is because I failed to play the best version of the deck (Song with either Tamiyo or Bone Miser to mitigate the discard). It was an irritating reminder that the best deck is the best deck for a reason. Playing sub-optimal decks exposes us up to Greg's randomness and gives him an opening.

tl;dr - We never lose to the AI. We lose when poor planning on our part coincides with a bout of extreme luck in favor of the AI.

The bolded text pretty much nails it.  

Honestly, when I first started this game, I took losses a lot harder. Some of it was being in a highly competitive coalition, but also because I'm a highly competitive person, and came to PQ with several years of paper experience.

Everything seemed terribly unfair, but that was a mistake; I was conflating randomness and fairness.  When I started accepting how much luck was going to factor into things, and how my best approach (both during deckbuilding and actual play) involved the mitigation of bad luck,  my play became much more consistent and perfect event scores became the norm instead of the exception. 

Anecdotally, I think this is why some people really hate the Timeless Voyage bottom node; essentially, you're giving Greg twice as many chances to bust out a magnificant cascade.  Bad feelings ensue because it exacerbates bad luck. 

I dont really disagree with anything being said and a lot of it is relateable.

However, isn't this "Bout of extreme luck in favor of the AI / randomness" actually more related to "unfair-ness" than meets the eye?

"Luck/randomness" is programmed - it must have specific conditions and constraints. Therefore, it is rarely true randomness but rather an altered randomness. Although we don't know the exact specifics of this altered randomness; we know that part of it is designed specifically to give the AI an edge. Because the AI has not been specifically programmed use cards effectively; this AI edge has been amplified to accommodate that [In other words, instead of a smarter AI; we have been given a "lucky" cascading AI in conjunction with a resource (mana, card draws) inhibiting environment for the player side].

Just food for thought

SuperCoolGeezer

Volrak said:

...

My own comment about the above thoughts is some of them read like "I've noticed the dice roll more odd numbers when it's rainy outside" - maybe absolutely true in some individual's experience, but unlikely to be repeatable.  Others are like "Sometimes the dice switch to hard mode and roll lots of 5s and 6s" - such trends can certainly be observed in randomness, even when there is no underlying structure.

What I have listed are personal observations. Regardless of how objective I try to look at things, most likely there will be some biased associated. Despite this, these are the most repeated trends I have observed. Furthermore, this would be the starting point in which to develop a potential hypothesis; but, by no means is this a full blown research study. The problem is that there are too many variables that are far too difficult to isolate. The sheer amount of data required and the number of research studies required to even start to isolate individual pieces of this puzzle are tremendous. Plus, it would make for some very time-consuming and boring game-play - just imagining hundreds of games without gem converters in order to prove gem board related phenomena is a noose away from a hanging. And I can't imagine myself doing it, as the fruits of such labor will yield, well, very little fruit for a game that already absorbs a lot of time. Thus, taking the lack of actual substantial data into account; I would imagine these are rather acceptable for unpolished observations aimed at opening the non-observant eyes of others.

But what you state isn't wrong; however, I do find it just a tad inflexible given the extreme difficulty associated with really proving these (no offense intended). If you notice "...the dice roll more odd numbers when it's rainy outside" happens once; then, yes, it's probably not significant. However, if you notice it happens in addition to "dice rolling more even numbers when its sunny outside," repeatedly over a long course of time, I would imagine its worth looking into. Perhaps the even portions of the dice are loaded with something that triggers based on atmospheric pressure? Everything in life can be spun; but the question is, how many sides are you willing to try to understand.

And perhaps I'm crazy but what about people who can sense weather changes with joint aches? mother's intuition? weird things among twins? Hard to prove but doesn't always mean its wrong. And there are trade professionals that have been doing what they are doing for so long that they can just sense small changes. I don't imagine myself as a professional MTGPQ player or anything; but, I imagine I've had a relatively good amount of experience in the game to notice some relatively obvious trends.

LoboX01

Few thoughts here.

It seems a lot of people see "AI cheating" a fact, but where is the background to show to other people? We talk about Greg cascades, turn 1, turn 4, 5, etc. What about our cascade, when we get it in turn 1? I suppose those are non-existent for people supporting this "AI cheater" theory/fact.

What about matches, when we hang on barely by the thread, but we know we have the card in our deck to save our back, and next turn we pull it, and get even enough mana match to cast and boom, turn the table, win the match.

Like it or not, with the gem matching concept, there always will be a randomness in this game. Sometimes it goes our way, sometimes to Greg's way. Is it more to Greg? Well, until someone takes the time and effort, and measure it, it will be in the beholder's eyes.

Sure enough, this weekend I had one match in RaW, where Greg got cascade after cascade until like turn 5-6, luckily for me, it was a light deck, so was able to come back and win. Now, did Greg cheat, and chose this deck to go with the crazy cascades, so was really being nice to me, because I could still win? No idea.

I faced on Leafkin - Azuza loop deck, still won, ended with perfect score. Again maybe Greg was nicely cheating, and did not pick the Leafkin-Azuza deck for the cascades.

On the other hand, in StD, I lost a match, because Greg got a turn 1 cascade able to cast his hand, strong cards, was never able to recover from that. Yes, I remember these... do I remember the cascades I got? No, probably because I did not care... hey it goes my way, I win the match, so why should I, right? Does it mean I did not got any? Hell no, and I don't have the time and dedication to start to make a statistics about my matches and see how many I get, how many Greg gets, and what is my win/lose ratio compared to that. 

Also, I don't understand a few things here. 

If someone is really playing for pure fun, and not for competing, why does it matter so much that he/she loses a match here and there? If rewards don't count, if no real competition, what difference does it make really? I just make up a "fun deck" and go with it.

Real frustration could come, when someone plays for competition, and/or the coalition events for a competitive coalition. 

But again, if someone is pay-to-win, then probably would have all the cards anyways with a VIP, or with some other extra cash in, so 80 yellow, 40 pinkies for a first place does not make a huge difference.

For a FTP player, it would make a difference over time, for sure. This can be offset somewhat, if you are playing in say a top 25 coalition, as you can drop a match, lose the perfect score, but a top 25 place would still give better rewards for the coalition, and hence to you.

ambrosio191

SuperCoolGeezer said:

madwren said:

jtwood said:

madwren said:

People feel it's cheating because the skill-to-luck ratio is different than what they expected--especially for anyone who's played paper Magic, where the game proceeds in a very orderly fashion.  

You can make perfect decisions on your side, but if Greg gets a 60-mana cascade, dumps his hand, and leaves you with three turns of loyalty matches, then it can be hard to overcome. 

Imagine that in Standard (not Vintage) paper Magic.  "Play an island, tap for 1 mana, summon merfolk, go." "Play a mountain. Tap for 40 mana, summon EVERYTHING." 

I think this is a good summary of the perception issue. I lost a match yesterday where Greg dumped his hand, which included Quartwood and Bolas Vanguard. I was using Song and discarding down to 1 card each turn. Bolas exiled it. I never got the Song train going again and lost in a couple turns, because Quartzwood will do that.

The reason I lost is in some part due to Greg getting a lucky cascade (the deck had no gem changers in play or in hand). But the other reason I lost - and the reason I was mad about that loss - is because I failed to play the best version of the deck (Song with either Tamiyo or Bone Miser to mitigate the discard). It was an irritating reminder that the best deck is the best deck for a reason. Playing sub-optimal decks exposes us up to Greg's randomness and gives him an opening.

tl;dr - We never lose to the AI. We lose when poor planning on our part coincides with a bout of extreme luck in favor of the AI.

The bolded text pretty much nails it.  

Honestly, when I first started this game, I took losses a lot harder. Some of it was being in a highly competitive coalition, but also because I'm a highly competitive person, and came to PQ with several years of paper experience.

Everything seemed terribly unfair, but that was a mistake; I was conflating randomness and fairness.  When I started accepting how much luck was going to factor into things, and how my best approach (both during deckbuilding and actual play) involved the mitigation of bad luck,  my play became much more consistent and perfect event scores became the norm instead of the exception. 

Anecdotally, I think this is why some people really hate the Timeless Voyage bottom node; essentially, you're giving Greg twice as many chances to bust out a magnificant cascade.  Bad feelings ensue because it exacerbates bad luck. 

I dont really disagree with anything being said and a lot of it is relateable.

However, isn't this "Bout of extreme luck in favor of the AI / randomness" actually more related to "unfair-ness" than meets the eye?

"Luck/randomness" is programmed - it must have specific conditions and constraints. Therefore, it is rarely true randomness but rather an altered randomness. Although we don't know the exact specifics of this altered randomness; we know that part of it is designed specifically to give the AI an edge. Because the AI has not been specifically programmed use cards effectively; this AI edge has been amplified to accommodate that [In other words, instead of a smarter AI; we have been given a "lucky" cascading AI in conjunction with a resource (mana, card draws) inhibiting environment for the player side].

Just food for thought

How do we know part of it is designed specifically to give the AI an edge? No one has provided anything other than anecdotal evidence of what they feel is happening.  No one has provided links to patch notes or question threads from developers, so we cant go by their word. Unless someone wants to gather the proper statistically significant data for everyone to analyze we have to assume it's random. I noticed @Volrak has a link in his signature for collection cascade data, I suggest people contribute to it in earnest if they truly care about what is happening. Wild speculation is just that, speculation.

JamesGam

LoboX01 said:

Few thoughts here.

It seems a lot of people see "AI cheating" a fact, but where is the background to show to other people? We talk about Greg cascades, turn 1, turn 4, 5, etc. What about our cascade, when we get it in turn 1? I suppose those are non-existent for people supporting this "AI cheater" theory/fact.

What about matches, when we hang on barely by the thread, but we know we have the card in our deck to save our back, and next turn we pull it, and get even enough mana match to cast and boom, turn the table, win the match.

Like it or not, with the gem matching concept, there always will be a randomness in this game. Sometimes it goes our way, sometimes to Greg's way. Is it more to Greg? Well, until someone takes the time and effort, and measure it, it will be in the beholder's eyes.

Sure enough, this weekend I had one match in RaW, where Greg got cascade after cascade until like turn 5-6, luckily for me, it was a light deck, so was able to come back and win. Now, did Greg cheat, and chose this deck to go with the crazy cascades, so was really being nice to me, because I could still win? No idea.

I faced on Leafkin - Azuza loop deck, still won, ended with perfect score. Again maybe Greg was nicely cheating, and did not pick the Leafkin-Azuza deck for the cascades.

On the other hand, in StD, I lost a match, because Greg got a turn 1 cascade able to cast his hand, strong cards, was never able to recover from that. Yes, I remember these... do I remember the cascades I got? No, probably because I did not care... hey it goes my way, I win the match, so why should I, right? Does it mean I did not got any? Hell no, and I don't have the time and dedication to start to make a statistics about my matches and see how many I get, how many Greg gets, and what is my win/lose ratio compared to that. 

Also, I don't understand a few things here. 

If someone is really playing for pure fun, and not for competing, why does it matter so much that he/she loses a match here and there? If rewards don't count, if no real competition, what difference does it make really? I just make up a "fun deck" and go with it.

Real frustration could come, when someone plays for competition, and/or the coalition events for a competitive coalition. 

But again, if someone is pay-to-win, then probably would have all the cards anyways with a VIP, or with some other extra cash in, so 80 yellow, 40 pinkies for a first place does not make a huge difference.

For a FTP player, it would make a difference over time, for sure. This can be offset somewhat, if you are playing in say a top 25 coalition, as you can drop a match, lose the perfect score, but a top 25 place would still give better rewards for the coalition, and hence to you.

So the player side is another variable; however, I don't quite understand why that is being considered a comparison point? Clearly the two sides are different -  one is the player and the other is the AI. Although the player side may influence the AI in that the player side gem matches create a completely new instance based on the whichever gem match the player takes.. I don't believe it is that relevant to the current case of looking at AI and its behaviors or environmental factors (gem board layout, cascade, etc). And unless someone knows a way to cheat the gem cascade system, pretty sure player side is not considered cheating.

And more importantly, perhaps the term "cheating" is a bit inappropriate. Honestly, all that is being presented here is a case for the AI potentially having assistance; ultimately, classifying it as the "hard mode" AI. The things that are being presented are possibilities in an attempt to understand and clarify. And perhaps some loving empathy but we can see that's not happening.

But what perplexes me beyond belief is how strongly people are objecting to even the possibility of what is being presented - as if stating the world is flat. What is even funnier is that the argument for the RNG or randomness is also something you also cannot prove without tons of data. Without the actual programming code, we would not know for either side of the argument. Perhaps I am playing devil's advocate but even seemingly random things or occurrences can be fabricated to appear that way - usually this may be more in the realm of manipulation, deceit, cons, espionage, framing, etc

JamesGam

ambrosio191 said:

SuperCoolGeezer said:

madwren said:

jtwood said:

madwren said:

People feel it's cheating because the skill-to-luck ratio is different than what they expected--especially for anyone who's played paper Magic, where the game proceeds in a very orderly fashion.  

You can make perfect decisions on your side, but if Greg gets a 60-mana cascade, dumps his hand, and leaves you with three turns of loyalty matches, then it can be hard to overcome. 

Imagine that in Standard (not Vintage) paper Magic.  "Play an island, tap for 1 mana, summon merfolk, go." "Play a mountain. Tap for 40 mana, summon EVERYTHING." 

I think this is a good summary of the perception issue. I lost a match yesterday where Greg dumped his hand, which included Quartwood and Bolas Vanguard. I was using Song and discarding down to 1 card each turn. Bolas exiled it. I never got the Song train going again and lost in a couple turns, because Quartzwood will do that.

The reason I lost is in some part due to Greg getting a lucky cascade (the deck had no gem changers in play or in hand). But the other reason I lost - and the reason I was mad about that loss - is because I failed to play the best version of the deck (Song with either Tamiyo or Bone Miser to mitigate the discard). It was an irritating reminder that the best deck is the best deck for a reason. Playing sub-optimal decks exposes us up to Greg's randomness and gives him an opening.

tl;dr - We never lose to the AI. We lose when poor planning on our part coincides with a bout of extreme luck in favor of the AI.

The bolded text pretty much nails it.  

Honestly, when I first started this game, I took losses a lot harder. Some of it was being in a highly competitive coalition, but also because I'm a highly competitive person, and came to PQ with several years of paper experience.

Everything seemed terribly unfair, but that was a mistake; I was conflating randomness and fairness.  When I started accepting how much luck was going to factor into things, and how my best approach (both during deckbuilding and actual play) involved the mitigation of bad luck,  my play became much more consistent and perfect event scores became the norm instead of the exception. 

Anecdotally, I think this is why some people really hate the Timeless Voyage bottom node; essentially, you're giving Greg twice as many chances to bust out a magnificant cascade.  Bad feelings ensue because it exacerbates bad luck. 

I dont really disagree with anything being said and a lot of it is relateable.

However, isn't this "Bout of extreme luck in favor of the AI / randomness" actually more related to "unfair-ness" than meets the eye?

"Luck/randomness" is programmed - it must have specific conditions and constraints. Therefore, it is rarely true randomness but rather an altered randomness. Although we don't know the exact specifics of this altered randomness; we know that part of it is designed specifically to give the AI an edge. Because the AI has not been specifically programmed use cards effectively; this AI edge has been amplified to accommodate that [In other words, instead of a smarter AI; we have been given a "lucky" cascading AI in conjunction with a resource (mana, card draws) inhibiting environment for the player side].

Just food for thought

How do we know part of it is designed specifically to give the AI an edge? No one has provided anything other than anecdotal evidence of what they feel is happening.  No one has provided links to patch notes or question threads from developers, so we cant go by their word. Unless someone wants to gather the proper statistically significant data for everyone to analyze we have to assume it's random. I noticed @Volrak has a link in his signature for collection cascade data, I suggest people contribute to it in earnest if they truly care about what is happening. Wild speculation is just that, speculation.

You are right. We don't know anything. However, the same can be said for the opposite argument: "How do we know part of it is designed specifically to not give the AI an edge?"

This will be "speculation" but here goes:

1. We know that there is definitely a super? hard mode AI that exists. Apparently, its name is Manu.

> If Manu exists, then at minimum his opposite the "easy mode" AI must exist. This is based on our gaming experience since we are exposed to both "easy mode" and "hard mode" AIs. From here, we can speculate the potential possibility of other AIs with varying levels of difficulty. Unproven; but, logically, very possible, as not every AI shows the same static level of difficulty. But, lots of variables to filter out...

2. For the most part, the AI lacks a lot of skill (i.e. can't use card mechanics / orders cards in hand as creature > support > spell) **AI Cycling is a new recent feature and will not be considered for now.

2.a. In the past, the AI has self targeted for spells that clearly should be used on the opponent. **I believe quite a bit of this has been resolved

2.b. Aside from cards that require certain criteria [fraying omnipotence / beacon bolt(no spells in yard)], the AI will use cards as they fill with mana. **Sometimes the AI cannot cast card draw/fetch cards unless a slot in their hand is freed.

> Based on this information, we know that the AI does not have complex capacities. In other words, each card is not programmed to be used a specific way - there is no algorithmic pathway charts (google adult cardiac arrest algorithm). Therefore, since there is very little AI "skill", another approach must be used in order to enhance the AI's "skill" to attain "hard mode". But before that, we have to define what is considered "hard mode" or more importantly what is considered a "hard" match. We don't know what the devs used as a definition; but, i imagine it would be probably be along the lines of "getting more threats out" than the opponent (player). Therefore, find a way for the AI to cast more cards in order to increase difficulty since the AI and individual cards do not have complex capacities.

3. The AI never chooses the cards of his deck.

3.a. Cards have been worded as "opposing" instead of "target" in a lot of cases to prevent the AI from targeting their own creatures or self.

> Since the AI is incapable of choosing his own deck, the "hard mode" AI needs to be able to use any deck. Therefore, a method needs to be employed that allows the AI to use all potential cards in the MTGPQ collection. And since the complex individual programming of cards (see 3.a.) will take far too many resources, including time; use a broad-range approach.

Based off of this short list, logically, it makes sense to employ a gem cascading AI as the "hard mode" AI. The massive mana generation will allow the AI to create enough threats to mimic a challenging match. Also, this broad range approach will self compensate for variables such as PW, deck, etc. Despite this, there are so many moving pieces and variables that it will be near impossible for the players to actually isolate. And as is human nature for most: the path of least resistance will be the one chosen, i.e. it's all RNG. Whoops, this last part was a bit off topic; the point is there are tons of variables that need to be isolated.

However, the question you need to ask yourself is, "how do they do this?" Well since it's so hard to manipulate the AIs programming. What if they manipulated the formula associated with the gem board layout and sky-falling gems? What if they used the environment surrounding the AI to create the illusion of a "hard mode" AI? After all, if the player can't get mana to cast their cards, while the AI gets tons of mana; well, we have ourselves a challenge. But it's not really a fair challenge, is it? You can't really blame someone for feeling cheated for situations as below:

madwren said:

People feel it's cheating because the skill-to-luck ratio is different than what they expected--especially for anyone who's played paper Magic, where the game proceeds in a very orderly fashion.  

You can make perfect decisions on your side, but if Greg gets a 60-mana cascade, dumps his hand, and leaves you with three turns of loyalty matches, then it can be hard to overcome. 

Imagine that in Standard (not Vintage) paper Magic.  "Play an island, tap for 1 mana, summon merfolk, go." "Play a mountain. Tap for 40 mana, summon EVERYTHING." 

SIDE NOTE: *I assume the devs weren't completely insensitive and set some restrictions or parameters in which "easy mode" or "hard mode" clicks on/off. Though another very real theory is that the gem board layout rotates and favors certain colors at certain points in time (if you use a Thousand Year Storm + Stitch deck and get like 50 extra swaps, you can see how the gems kinda rotate towards certain colors)

Thoughts?

JamesGam

Oh! And I will leave here with this final little piece of my thoughts. Why it took me so long to come up with is beyond me but...

If we think about "Manu" as a "hard mode" AI, we have to define this "hard mode". "Hard mode" can only exist if it can consistently portray characteristics that make it "hard mode". In other words, it cannot be a one-pump chump AI since that will automatically disqualify it as "hard mode". Consequently programming must be in place to ensure a relatively steady flow of high pressure (i.e. mana) on the opponent/player. If things were truly random, it would be very difficult for such consistency to be achieved, no? I mean its possible; but, I would imagine it would take the moon and stars to align.

Bubbles_CS

I wish I had even a fraction of the time required to collect and reconcile the data to understand what is actually going on, because I totally would.

Volrak

JamesGam said:

Oh! And I will leave here with this final little piece of my thoughts. Why it took me so long to come up with is beyond me but...

If we think about "Manu" as a "hard mode" AI, we have to define this "hard mode". "Hard mode" can only exist if it can consistently portray characteristics that make it "hard mode". In other words, it cannot be a one-pump chump AI since that will automatically disqualify it as "hard mode". Consequently programming must be in place to ensure a relatively steady flow of high pressure (i.e. mana) on the opponent/player. If things were truly random, it would be very difficult for such consistency to be achieved, no? I mean its possible; but, I would imagine it would take the moon and stars to align.

Hard mode (Manu) was just a change to the evaluation function to choose more optimal matches on the gem board each turn.  Although it still didn't match as well as a skilled human player, the extra mana and loyalty Manu pulled in over Greg made a striking difference to the game in the brief period they used it - so much so that they shelved it due to a flood of complaints that it made the game too hard.

I find myself wondering, if Oktagon's goal was to make the AI stronger, which would they choose?  Invest time and money designing, coding, and testing an elaborate set of "cheating" features they won't ever announce, which tweak the game experience so subtly that although some people get tantalising hints of it, nobody can actually distinguish it from the normal variability of randomness?  Or just flick a switch and bring back Manu (or parts of him)?

jtwood

Volrak said:

Hard mode (Manu) was just a change to the evaluation function to choose more optimal matches on the gem board each turn.  Although it still didn't match as well as a skilled human player, the extra mana and loyalty Manu pulled in over Greg made a striking difference to the game in the brief period they used it - so much so that they shelved it due to a flood of complaints that it made the game too hard.

I call this period in the game's history: "The Good Ol' Day"

JamesGam

Volrak said:

JamesGam said:

Oh! And I will leave here with this final little piece of my thoughts. Why it took me so long to come up with is beyond me but...

If we think about "Manu" as a "hard mode" AI, we have to define this "hard mode". "Hard mode" can only exist if it can consistently portray characteristics that make it "hard mode". In other words, it cannot be a one-pump chump AI since that will automatically disqualify it as "hard mode". Consequently programming must be in place to ensure a relatively steady flow of high pressure (i.e. mana) on the opponent/player. If things were truly random, it would be very difficult for such consistency to be achieved, no? I mean its possible; but, I would imagine it would take the moon and stars to align.

Hard mode (Manu) was just a change to the evaluation function to choose more optimal matches on the gem board each turn.  Although it still didn't match as well as a skilled human player, the extra mana and loyalty Manu pulled in over Greg made a striking difference to the game in the brief period they used it - so much so that they shelved it due to a flood of complaints that it made the game too hard.

I find myself wondering, if Oktagon's goal was to make the AI stronger, which would they choose?  Invest time and money designing, coding, and testing an elaborate set of "cheating" features they won't ever announce, which tweak the game experience so subtly that although some people get tantalising hints of it, nobody can actually distinguish it from the normal variability of randomness?  Or just flick a switch and bring back Manu (or parts of him)?

Good points and a very possible theory.

1. "... set of 'cheating' features they won't ever announce, which tweak the game experiences so subtly that although some people get tantalising hints of it, nobody can actually distinguish it from the normal variabillity of randomness?" I'm sure you can find the information for such things with other applications/games that did exactly this. Consequently, although what you say is true in an idealistic and completely moral/ethical situation; we cannot trust that it will be so in reality. Ultimately, it will neither prove nor disprove the two sides of the debate - a wash.

2. The investment to a broad range approach feels rather cost effective and time efficient. This is particularly the case when we know to some degree in the past they provided broad-range blanket solutions to certain issues. Also, when we consider the past history or tasks Oktagon has been busy with; they have had a lot of things on their plate (dropping new sets (some of which were delayed), bug resolution, customer service improvements, QoL improvements, VIP, VIP related problem resolution, and other features. Of course, this supports both sides of the argument, as a pre-built Manu could be a cost effective/ time efficient solution. Once again another wash. 

3. Is there information I am missing on Manu (Honestly, I don't remember Manu that well.. it's been a long time.. if someone could jog my memory)? Is there more that you know about Manu that perhaps other do not?

3. With the return of Cycling, were any changes made to the AI to help it use those cards appropriately (or are there plans to do so)?

Answer: Yes. Greg now knows how to use Cycling! Also, we have plans to make it learn how to use other mechanics strategically as well. But don’t worry, we won’t make Greg reach Manu’s level again so soon.

(GD’s Comments: Manu is the nickname we gave to the “Pro-Level Greg”, that we tested for a brief time during the release of Ixalan. Manu was a bit harder than intended, so we had to reduce its intelligence before it started a rebellion against mankind.)

a. Manu can't use other mechanics other than cycling; therefore, Manu cannot be that complex. If Manu cannot be that complex.. then its still possible other RNG related components (gem layout, skyfall, card draw, etc) were utilized. But another wash.

b. If we consider Manu the highest tier "hard mode" AI and Manu is able to make significantly better gem matches than any Greg; then a lower tier "hard mode" Greg can still make some good gem matches... however, none of this information negates the possibility of any of these AIs having additional environmental assistance (gem layout, skyfall, card draw, etc). And these environmental assistance factors could very well be programmed to trigger when facing a certain AI. But another wash.

What bugs me is that people are willing to stand on the side of RNG and adamantly negate other potential possibilities; despite, there being information in other fields/games/applications/companies that specifically manipulate this data in their favor. And it's rather funny to see that for the most part, both sides of this debate are incapable of really proving anything. Nevertheless, for all we know, all of us could be correct - they combined RNG and manipulated RNG.

Anyhow, for the question you raise, point #2 would be my opinion for a potential broad range solution.

ambrosio191

People who are far more curious and far more experienced in programming than myself have verified that new gems that appear are RNG generated and done at the time of creation, i.e. after that gem swap is made.  Greg cannot take into account what is going to come because what's going to come does not exist until after the gem match is made.  Those people have also shown what Greg takes into account when making matches (though that was a few years ago and things have changed, with those changes being announced in various patch notes).

This has been argued time and time again over the years.  It's never been anything other than RNG.

Bubbles_CS

ambrosio191 said:

People who are far more curious and far more experienced in programming than myself have verified that new gems that appear are RNG generated and done at the time of creation, i.e. after that gem swap is made.  Greg cannot take into account what is going to come because what's going to come does not exist until after the gem match is made.  Those people have also shown what Greg takes into account when making matches (though that was a few years ago and things have changed, with those changes being announced in various patch notes).

I would love to know more about this and how it was done.

ArielSira

I remember Octal9 showing solid evidence of the randomness in generating gems.

ambrosio191

Yes, Octal9 has shown evidence in the past.  You can look at his "insightful" posts for some of it.

madwren

JamesGam said:

What bugs me is that people are willing to stand on the side of RNG and adamantly negate other potential possibilities; despite, there being information in other fields/games/applications/companies that specifically manipulate this data in their favor. And it's rather funny to see that for the most part, both sides of this debate are incapable of really proving anything. Nevertheless, for all we know, all of us could be correct - they combined RNG and manipulated RNG.

To pivot off your comment, what bugs me is that we seem to get a post every month alleging that the AI is cheating, yet none of the accusers are willing to provide anything but anecdotal evidence (that often conveniently disregards that players experience the same cascade/gem behavior).  Why wouldn't one be on the side of RNG?

The burden of proof lies on the person who makes the claim, and those claims are always the same.  

Here's a great thread to save people some lookup time. https://forums.d3go.com/discussion/70603/the-lucky-greg/p1

ambrosio191

Thank you for that link @madwren . I remember seeing the images octal posted in that thread but couldn't remember when or where. Tried finding it but as you said, this topic gets brought up again and again and octal has responded time after time to exhaustion. There was more than I wanted to sift through.