Fed Up!

Aeroplane

Greg not making the obvious gem match to only cascade to a better match is the only time I see some fishiness. But Greg being so bad with the abundance of perfect scores shows the game imbalance. Gem matching is pretty basic once you understand sky fall, deck building for the dedicated player doesn't have a lot of variety in Standard with only 1-2 cards from common and uncommon being useful .Legacy is just loopy. There is the good old bug to kick you to the curb after a lot of work, too.

There should be more of variety of mechanics that Greg should be able to use to manipulate the board as I believe they can only improve his gem matching and not his card play. Only losing once shouldn't be the downfall of an event. There should be a high risk reward of using only rare or below cards with a couple of bonus pts. Once the scores start to tier more you can give higher rewards placing in the top 100.  This would also shake up the rankings a bit too.

Nyarlathotep

The game feels more like seeded than completely random. And I am totally fine with that. 


Here are some things that directly point towards that:

- Different events use different base 'difficulties'
And all of them have a completely different feel to them. 

- Coalition pvp means, you make one wrong choice, just a single one, and you're done, it's either a complete loss, or at the very least a loss of an objective. 
Plenty of ppl say this every single week. 

- Other events, like non coalition stuff, means that you basically can't lose a single match ever. 
There may be some events, like FtH for example, that sit somewhere in between. 

When you are in a 'low difficulty environment', Greg will pose no threat and it in general way more cascades prefer the player over greg.

But no one ever complains about those matches, because ppl win and a second later forget about it. 
A loss is not easy forgotten about, esp when it's in a coalition event.

I think they should not really change the randomness of the game at all, because that is a very delicate thing to do. And as far as my player based knowledge of this game tells me, it's working really well.

What they could do is this: restructure rankings/scores 
Make a loss of a single objective or a single loss be less impactful. 

Ppl just complain about being cascaded and stomped by Greg, because they are used to win and win perfect, every single time. 
Magic was never created to be that one sided, and I think neither was mtgpq. 

Bubbles_CS

FindingHeart8 said:

But my only question in the randomness factor is whether or not Greg makes matches knowing what gems will fall after.

I know there's been a lot of speculation and arguments on this in the past.  I'm not making a claim for or against it, but it would be interesting to know *how* Greg chooses what match to make, if there are multiple options available with similar perceived board-state outcomes.

I have frequently found that when I don’t take the “recommended” match and something obviously better doesn’t fall onto the board that Greg will make that same recommended match, so I would hypothesize that Greg uses this same logic for choosing a match. Why build matching logic twice?

I have experimented some with the recommended match and found that it doesn’t seem to have a high rate of sky fall matches, so if it is true that Greg uses the recommended matching logic and that logic isn’t predicting sky fall with some frequency, this would contraindicate the theory that Greg predicts sky fall and uses that to match. If his sky fall rate is really above average, this would suggest another mechanism for boosting his matches. My feeling is that his rate is the same as ours, though. I wish I had the time to properly investigate this, but I don’t and will leave it as theory for now.

Endbringer

Greg designs what gem fall will occur in the moment the match is made. It's controlled to drop gems that will benefit Greg's best outcome. 

FindingHeart8

Tremayne said:

FindingHeart8 said:

Tremayne said:

New MTGPQ - no use of randomness in game, unless proof of correctly implemented randomness is published for public peer review.

As someone who had to edit peer-reviewed journals, I can tell you there's many who are still a ways off from unbiased

Many implementations of randomness are unbiased or many peer reviewers are unbiased? 🤪

Many scholars and reviewers are biased, so subsequently are the articles.

But biased randomness sounds mtgpq applicable alright haha.

Bil

Endbringer said:

Greg designs what gem fall will occur in the moment the match is made. It's controlled to drop gems that will benefit Greg's best outcome. 

That sounds like pure speculation ... And if it is true and Works like a matching rule... We Can Wonder why Greg doesn't get Infinite cascades on each swap.

Larz70

I get what @Endbringer is saying.  It's not about how RNG or stats/probabilities is supposed to behave, it's about how Greg wins.  It just feels a lot like cheating.  Also, Greg doesn't try to beat everyone, he picks unsuspecting victims (/sarcasm but with a tinge of truth).

We may never know exactly how Greg does it but one thing is for sure, when I play a match against Greg, I sometimes get the stupid Greg, and I sometimes get the smart Greg ... or both Greg's in the same match.

Well, that's exactly what RNG means right?  Let's say the RNG in Greg is programmed to pick the best move and crush us 10% of the time (either through crazy cascades or a freeze).  The funny thing with stats and numbers is that 10% does not always mean everyone's win rate is going to be 90%, it's also possible to create a scenario where 9 players get perfect scores while 1 player gets shut out.  That's still 90%.  I've heard this said before in one form or another, "if the probability is something like 50%, it means for half of the population, it's 100% and for the the other half, it's 0%".

Is there a solution?  Maybe a difficulty level that players can choose?  If I want to relax and experiment with decks, let me play against easy Greg who always picks off-color match-3's and lets me have all the match-5's.  For serious events (like coalition) or if I want a challenge, let me play against the evil Greg who seems to know what color gems are dropping from above.

Endbringer

Bil said:

Endbringer said:

Greg designs what gem fall will occur in the moment the match is made. It's controlled to drop gems that will benefit Greg's best outcome. 

That sounds like pure speculation ... And if it is true and Works like a matching rule... We Can Wonder why Greg doesn't get Infinite cascades on each swap.

Their is a goal set up to fill the mana of certain cards to rise to the challenge, and then it slows down. Haven't you noticed more than a lot of times, how a cascade will take place accurately enough to "exactly" fill the high mana cost of a certain card in yours or Greg's hand and then stop? Or perhaps numerous cards, making matches and dropping gems to make matches until 3 cards are full or 4, and then once it fills them (to combat the current threat or create a dominating offense) it stops.

Larz70

Endbringer said:

Bil said:

Endbringer said:

Greg designs what gem fall will occur in the moment the match is made. It's controlled to drop gems that will benefit Greg's best outcome. 

That sounds like pure speculation ... And if it is true and Works like a matching rule... We Can Wonder why Greg doesn't get Infinite cascades on each swap.

Their is a goal set up to fill the mana of certain cards to rise to the challenge, and then it slows down. Haven't you noticed more than a lot of times, how a cascade will take place accurately enough to "exactly" fill the high mana cost of a certain card in yours or Greg's hand and then stop? Or perhaps numerous cards, making matches and dropping gems to make matches until 3 cards are full or 4, and then once it fills them (to combat the current threat or create a dominating offense) it stops.

This one can be explained by the timing in which the animation, and the values/amounts sent to the animation.  When there is a cascade, the game calculates the total mana gained from all the matches.  For example, player matches a gem for 6 mana and it cascades into another match for 3 mana, a total of 9 mana.  The card's mana-filling animation will use a value of 9 mana.  Meanwhile, the cascade animation will shoot 6 mana to the card.  Then it shoots another 3 mana to the card but by then the mana-fill animation already completed the 9 mana fill-up and the +3 mana will fail to make the mana budge since it was already accounted for.

This is similar to opening a booster pack and the split-second before the orbs and cards are displayed.  If you look at your rune count, you'll see that it has already been updated with the runes earned from dup cards so if you opened a 5-booster pack and the rune count jumped up by 50, that means the pack contained 5 dup common cards.

Gunmix25

Endbringer said:

Greg designs what gem fall will occur in the moment the match is made. It's controlled to drop gems that will benefit Greg's best outcome. 

I guess that's why I beat the living tar out of greg every time then, because he makes bad decisions trying to cheat?
  

I disagree because players tend to remember those tiny kitty matches rather than the ones that went their way... in the exact same way they are complaining about what greg supposedly is manipulating. In short, the RNG is working as intended and yes, moments when Greg gets those wonderful cascades is irritating, but I know full well that it happens to me just as much. We just tend to remember those rotten cascades than the good ones.  I am sure Greg is somewhere on the AI forum complaining about how the humans are cheating.

In my coalition we often would present screen shots of a match and ask what our team mates think the next move should be to maximize a likely cascade. This game is not about how fast you can get to the win but how smart you can... which means taking your time and assess the falls before making the swap. 

FindingHeart8

I wish we could gauge how intelligent Greg really was.  I used to frequently play chess against an AI where you could adjust its skill level.   At the highest levels it was near impossible to beat, at lower levels it played like a fblthp match   

Would be nice to see how intelligent Greg overall ranks, separate from the extra variables.