Why did the AI need a nerf?
Comments
-
Steeme said:babar3355 said:
I don't know what the optimal win rate is to keep people hooked on a game.. but I do feel like it is something less than 95%+.
I know in Magic Duels a win rate of around 70% is very strong, although that certainly goes up against the AI. I also know that I am very glad the AI acts quickly in PQ rather than the slow motion moves in Duels.
But anyway, maybe they are worried about making the AI too hard, so they don't prioritize match 5s? Or are they so dense that they think landfalls are better?
Well, given that JC states with absolute certainty that the AI does not have an advantage over the player in net mana gain, that means they are using some type of in-house simulation to gauge the rate at which the AI progresses in a match.Perhaps setting the 5-swap as a top priority lead to an imbalance in net mana gain in favour of the AI. I mean, we still get periodic "cascade" threads.
The solution was to set something else as the priority. But that didn't work out as planned.
So now they re-jigged the priorities.
Jenga Jenga, J-J-J-Jenga
Only donkeys think the game has an advantage in mana gains. Of course it doesn't.
Anyway, if you read my OP I was speculating that they did it intentionally to try and get a higher retention rate for new players, who might fall victim to the "AI cheating" mentality. I just contend that it's a poor choice in terms of actually leading to better retention rates and more profits for D3. Give them an advantage in cards early in the game, not an advantage in gem swapping.
1 -
I guess one way to determine what the intent of the dev team is is to see if they 'fix' this change in the next update.
0 -
shteev said:Do you think they were trying to make a popular change, and accidentally made an unpopular one?
I think someone made a change that inadvertently caused the reduced match-5s by the AI. They have a number of new employees working on this game and I think someone screwed up and didn't realize it until now.0 -
I wonder if they will attack it with the same vim and vigor as the floodwaters bug?1
-
DumasAG said:shteev said:Ohboy said:Before I go to the trouble of digging. Please state for the record that you do not recall constantly complaining about the incompetence of the coding.
Go!
0 -
Gunmix25 said:DumasAG said:shteev said:Ohboy said:Before I go to the trouble of digging. Please state for the record that you do not recall constantly complaining about the incompetence of the coding.
Go!
//Removed Inappropriate Content -Brigby
0 -
The new AI only goes after your activate gems. It won't destroy your supports or go after 5 matches0
-
It seems that the AI cares less about mana colors now too. In my TG matches against Koth it didn't seem to care at all about red matches, even tho I was playing red too.0
-
babar3355 said:
Only donkeys think the game has an advantage in mana gains. Of course it doesn't.
I think you're incorrect here. The AI has an advantage in mana gains:
1. It's optimized.
2. It doesn't make mistakes.
The AI is certainly not cheating. But it does have an advantage. Humans make mistakes, use emotion, "miss" stuff because they're relying on their eyes. Computers do not have those quirks. Why else would they be adjusting the algorithm?
The player is able to steamroll the AI for other reasons, like card prioritization, anticipation, replacing creatures, buffing enemies / killing allies, loyalty gain/usage, etc.
0 -
-
It is only optimized with how it is programmed. It doesn't make a mistake in the program's eye. But it does not not , in my eye, make the optimized choice and therefore can not have an advantage in mana gain. Even if was programmed hierarchically how you or the majority feel it should, it still will choose the way it is geographically programmed if there are two matches that meets its top priority. I think, previously, if not still currently, it would choose from top left then work its way down and to the right in some fashion.
Therefore, if there are two top priority matches it will pick geographically instead of looking at the cascade that will happen after the match. I have never seen a match that is made specifically to complete a cascade. So, it does not have a mana advantage.
I would say that 95% (probably closer but not 100%) of the time a match 5 is the best for mana gain. Since it seems the game does not currently prioritize match 5, it does not have a mana advantage over me.1 -
UweTellkampf said:
No, it's common sense.Skiglass6 said:It is only optimized with how it is programmed. It doesn't make a mistake in the program's eye. But it does not not , in my eye, make the optimized choice and therefore can not have an advantage in mana gain. Even if was programmed hierarchically how you or the majority feel it should, it still will choose the way it is geographically programmed if there are two matches that meets its top priority. I think, previously, if not still currently, it would choose from top left then work its way down and to the right in some fashion.
Therefore, if there are two top priority matches it will pick geographically instead of looking at the cascade that will happen after the match. I have never seen a match that is made specifically to complete a cascade. So, it does not have a mana advantage.
I would say that 95% (probably closer but not 100%) of the time a match 5 is the best for mana gain. Since it seems the game does not currently prioritize match 5, it does not have a mana advantage over me.
I should have been more clear that the AI can be optimized, but currently it's intentionally nerfed. Sure you could argue that taking a lesser swap might chance you something better, but the only reason the AI currently chooses something "worse" is because they have **** it.Anything that you can think of, can be programmed. It is a finite board with a finite amount of possibilities. The only thing it cannot do is "know" what will drop down from the sky. All it can do is calculate probabilities. The human player does the exact same thing.
0 -
Steeme, there is no single case that I can remember, where I thought, great, what a great matching move by the AI. Why are we at all talking about AI's talent as if it was something unique and needs its own perspective? It's not comparable to our own - but as much capable as they make it to be, and they have made it incapable in the past and now have even managed to completely screw it.
So, no, the AI is not optimized even a single bit, and no, it makes mistakes most always. Or to say it from the perspective of the game it is supposed to be winning: The developers have not managed well (willingly or unwillingly) to program it in a way that it would consider all its opportunities to find the best one(s) that enable it to do such things as
1. Get to their 2nd/3rd ability (for example via the extra gain from match-5s or, you know, saving up the first)
2. To scan the board for possible second or third degree matches resulting from their own
3. And now even to destroy our own supports, even those hurting their next move badly.
(and much more)
I also can't really believe this move has been made to please the new players. We have so many discussions in the gaming communities about the difficulty of games and challenge they pose to us. This here is actually a slap in our face in sense of "facing a difficulty". The game would be just so much more fun if there was a no feeling that we're playing a potato.0 -
UweTellkampf said:Steeme, there is no single case that I can remember, where I thought, great, what a great matching move by the AI. Why are we at all talking about AI's talent as if it was something unique and needs its own perspective? It's not comparable to our own - but as much capable as they make it to be, and they have made it incapable in the past and now have even managed to completely screw it.
So, no, the AI is not optimized even a single bit, and no, it makes mistakes most always. Or to say it from the perspective of the game it is supposed to be winning: The developers have not managed well (willingly or unwillingly) to program it in a way that it would consider all its opportunities to find the best one(s) that enable it to do such things as
1. Get to their 2nd/3rd ability (for example via the extra gain from match-5s or, you know, saving up the first)
2. To scan the board for possible second or third degree matches resulting from their own
3. And now even to destroy our own supports, even those hurting their next move badly.
(and much more)
I also can't really believe this move has been made to please the new players. We have so many discussions in the gaming communities about the difficulty of games and challenge they pose to us. This here is actually a slap in our face in sense of "facing a difficulty". The game would be just so much more fun if there was a no feeling that we're playing a potato.Ok, I see what you meant now. It's quite difficult when you give a one line response.
We are all in agreement that the AI didn't need a nerf. But I disagree that the AI makes mistakes. It does exactly as it is programmed.
0 -
UweTellkampf said:Steeme, there is no single case that I can remember, where I thought, great, what a great matching move by the AI. Why are we at all talking about AI's talent as if it was something unique and needs its own perspective? It's not comparable to our own - but as much capable as they make it to be, and they have made it incapable in the past and now have even managed to completely screw it.
So, no, the AI is not optimized even a single bit, and no, it makes mistakes most always. Or to say it from the perspective of the game it is supposed to be winning: The developers have not managed well (willingly or unwillingly) to program it in a way that it would consider all its opportunities to find the best one(s) that enable it to do such things as
1. Get to their 2nd/3rd ability (for example via the extra gain from match-5s or, you know, saving up the first)
2. To scan the board for possible second or third degree matches resulting from their own
3. And now even to destroy our own supports, even those hurting their next move badly.
(and much more)
I also can't really believe this move has been made to please the new players. We have so many discussions in the gaming communities about the difficulty of games and challenge they pose to us. This here is actually a slap in our face in sense of "facing a difficulty". The game would be just so much more fun if there was a no feeling that we're playing a potato.3 -
Steeme said:
Ok, I see what you meant now. It's quite difficult when you give a one line response.
We are all in agreement that the AI didn't need a nerf. But I disagree that the AI makes mistakes. It does exactly as it is programmed.
1 -
Hi. I'm here to clear up some misconceptions in this thread.
First, to confirm: yes, the AI has been changed to deprioritize L-shaped match-5s. I can't tell you why it has, only that it has. I have a reliable source of this information.Mburn7 said:Its funny, this happens every update. They fix all AI bugs, and then when they release new cards and mechanics it breaks again.fiirst said:my AI (sorin) match 5 white gems as L shape. i think it depend pw color too.Pcell777 said:It bothers me more, now, that AI deliberately avoids 5-match gems than them having crazy cascades and extra turns.tfg76 said:The weird thing is that gem optimization seems like the simplest thing to implement!OhBoy said:As far as I could tell, the game has a defence and aggro modeCorn Noodles said:I think someone made a change that inadvertently caused the reduced match-5s by the AI. They have a number of new employees working on this game and I think someone screwed up and didn't realize it until now.Steeme said:1. It's optimized.
2. It doesn't make mistakes.
2) true, in the sense that it is not bugged. it makes mistakes when it comes to decision making, but it makes exactly the decisions it is meant to.5 -
octal9 said:fiirst said:my AI (sorin) match 5 white gems as L shape. i think it depend pw color too.
[edit]
@octal9 ooh, I nearly forgot... the AI seems to be avoiding popping my supports too. I played for ages with a Cast Out on 1 shield the other day. What's the word on that?
0 -
Thanks for the insights, Octal. It's very interesting.
As Shteev pointed out, we've definitely noticed that our one-off supports--Claustrophobia, Hixus, et al.---seem to be lasting a bit longer, as if the AI has deprioritized them as well.
0
Categories
- All Categories
- 44.8K Marvel Puzzle Quest
- 1.5K MPQ News and Announcements
- 20.2K MPQ General Discussion
- 3K MPQ Tips and Guides
- 2K MPQ Character Discussion
- 171 MPQ Supports Discussion
- 2.5K MPQ Events, Tournaments, and Missions
- 2.8K MPQ Alliances
- 6.3K MPQ Suggestions and Feedback
- 6.2K MPQ Bugs and Technical Issues
- 13.6K Magic: The Gathering - Puzzle Quest
- 503 MtGPQ News & Announcements
- 5.4K MtGPQ General Discussion
- 99 MtGPQ Tips & Guides
- 421 MtGPQ Deck Strategy & Planeswalker Discussion
- 298 MtGPQ Events
- 60 MtGPQ Coalitions
- 1.2K MtGPQ Suggestions & Feedback
- 5.6K MtGPQ Bugs & Technical Issues
- 548 Other 505 Go Inc. Games
- 21 Puzzle Quest: The Legend Returns
- 5 Adventure Gnome
- 6 Word Designer: Country Home
- 381 Other Games
- 142 General Discussion
- 239 Off Topic
- 7 505 Go Inc. Forum Rules
- 7 Forum Rules and Site Announcements