anyway to make AI play betterr

slidecage

just went against a aunt may emma team

even though they had enough ap to wipe me out with aunt may red the ai would only fire emma red

they need to add something that tells them to fire what power

DAZ0273

The old Devs did a lunchtime experiment where they ramped up the AI ability (things like being able to see non straight line Match 5's and prioritise powers) and apparently it was a bit of a bloodbath in relative terms compared with usual game expectations and it was quickly decided that players would not enjoy this experience at all.

It should be remembered that the Devs reduced the amount of cascades you could suffer on the AI's go in PVE due to player backlash, this was even more player unfriendly.

Maybe they could try again but they know what the metrics look like in terms of expected player wins and the outcry that will follow should that drop by any noticeable amount.

entrailbucket

My favorite AI story is about 5* Gambit. He was a menace on defense, just awful to fight, worse than Chasm ever was. It felt like if you didn't have him, you lost every time.

When they announced that they'd be looking at nerfing him, the devs revealed his defensive win rate. Any guesses???

...

10%. The most broken defensive character ever, who felt like he stomped you every time, was actually winning 10% of the time. If the AI wins 10% of the time in MPQ, it feels broken.

meadowsweet

@entrailbucket said:
10%. The most broken defensive character ever, who felt like he stomped you every time, was actually winning 10% of the time. If the AI wins 10% of the time in MPQ, it feels broken.

I take your point: 10% is much, much less than 100%. But there are several caveats to note:

1. What was the "skip rate"? PVE matches are voluntary. If I had three doors and Gambit was behind one of them, I'd choose between the other two. If all three PVE nodes were Gambit, I would hit the skip button until one of them wasn't. In theory, I never had to fight Gambit if I didn't want to (or knew I'd probably lose.)
2. Were retreats also considered defeats? Most likely they were, but I'm not 100% sure about that.
3. What if I can reliably defeat Gambit's team, but I have to use 1-3 health packs after every single match? Those wouldn't be considered defensive wins, but it's still frustrating and quickly limits my ability to continue playing.
4. If you had a Gambit counter team champed or max-champed, maybe you won almost 100% of the time. If your roster didn't have any effective Gambit counters, maybe you lost almost 100% of the time. You would also get a 10% defensive win rate if you have nine players losing to Gambit only 2% of the time while one player loses to Gambit 82% of the time. That tenth player is going to be upset and vocal about it.
5. Maybe 10% sounds low - players still won 90% of the time, after all! But hypothetically, if the next highest character's defensive win rate was only 2-3% (out of hundreds of characters), then clearly Gambit was a freakish statistical outlier behaving unlike anyone else in the game. But I don't know if that information was included at the time?

DAZ0273

Exactly. Suffering back to back losses in PvP is a major shock to the system for players, if that became more regular there would be murder on the dance floor. All those Shang teams who suddenly know where to place those Deadly Hands after they have been flooded with AP by Jane and then can use red to power up again. Boom.

entrailbucket

@meadowsweet said:

@entrailbucket said:
10%. The most broken defensive character ever, who felt like he stomped you every time, was actually winning 10% of the time. If the AI wins 10% of the time in MPQ, it feels broken.

I take your point: 10% is much, much less than 100%. But there are several caveats to note:

1. What was the "skip rate"? PVE matches are voluntary. If I had three doors and Gambit was behind one of them, I'd choose between the other two. If all three PVE nodes were Gambit, I would hit the skip button until one of them wasn't. In theory, I never had to fight Gambit if I didn't want to (or knew I'd probably lose.)
2. Were retreats also considered defeats? Most likely they were, but I'm not 100% sure about that.
3. What if I can reliably defeat Gambit's team, but I have to use 1-3 health packs after every single match? Those wouldn't be considered defensive wins, but it's still frustrating and quickly limits my ability to continue playing.
4. If you had a Gambit counter team champed or max-champed, maybe you won almost 100% of the time. If your roster didn't have any effective Gambit counters, maybe you lost almost 100% of the time. You would also get a 10% defensive win rate if you have nine players losing to Gambit only 2% of the time while one player loses to Gambit 82% of the time. That tenth player is going to be upset and vocal about it.
5. Maybe 10% sounds low - players still won 90% of the time, after all! But hypothetically, if the next highest character's defensive win rate was only 2-3% (out of hundreds of characters), then clearly Gambit was a freakish statistical outlier behaving unlike anyone else in the game. But I don't know if that information was included at the time?

Don't misunderstand. The point of the story is not "see, Gambit wasn't even all that good!" The point of the story is "in MPQ, wiping 10% of the time feels AWFUL."

A guy who wins on defense 10% of the time actually is broken! The normal wipe rate is probably like 1% or less, so Gambit was winning like 10x (or more!) more than anybody else.

AdeptusRevolt

@DAZ0273 said:
The old Devs did a lunchtime experiment where they ramped up the AI ability (things like being able to see non straight line Match 5's and prioritise powers) and apparently it was a bit of a bloodbath in relative terms compared with usual game expectations and it was quickly decided that players would not enjoy this experience at all.

It should be remembered that the Devs reduced the amount of cascades you could suffer on the AI's go in PVE due to player backlash, this was even more player unfriendly.

Maybe they could try again but they know what the metrics look like in terms of expected player wins and the outcry that will follow should that drop by any noticeable amount.

https://sites.google.com/view/mpqhub/home/ai-information

Tony_Foot

@entrailbucket said:

Don't misunderstand. The point of the story is not "see, Gambit wasn't even all that good!" The point of the story is "in MPQ, wiping 10% of the time feels AWFUL."

A guy who wins on defense 10% of the time actually is broken! The normal wipe rate is probably like 1% or less, so Gambit was winning like 10x (or more!) more than anybody else.

Yeah but how many of those recording where players fighting a one cover gambit. Or using a one cover gambit, as has been stated there are so many variables in that stat. I bet with champed gambits the win rate was higher than 10%. I say that as someone who didn’t want him nerfed and would take him back in any of his previous forms right now with no complaints. Mthor would still trash him.

nwahs086

@slidecage said:
just went against a aunt may emma team

even though they had enough ap to wipe me out with aunt may red the ai would only fire emma red

they need to add something that tells them to fire what power

You don't want the AI to be better because losing repeatedly because of cascades and the AI doing stuff out of your control isn't fun.

sambrookjm

@nwahs086 said:

@slidecage said:
just went against a aunt may emma team

even though they had enough ap to wipe me out with aunt may red the ai would only fire emma red

they need to add something that tells them to fire what power

You don't want the AI to be better because losing repeatedly because of cascades and the AI doing stuff out of your control isn't fun.

Having just got obliterated by a bunch of mThor cascades that took out two of my 550+ LVL (with boosts) characters, I approve of this message.

KGB

@sambrookjm said:

@nwahs086 said:

@slidecage said:
just went against a aunt may emma team

even though they had enough ap to wipe me out with aunt may red the ai would only fire emma red

they need to add something that tells them to fire what power

You don't want the AI to be better because losing repeatedly because of cascades and the AI doing stuff out of your control isn't fun.

Having just got obliterated by a bunch of mThor cascades that took out two of my 550+ LVL (with boosts) characters, I approve of this message.

I read the description of what happened in the 'worst moments' thread.

We really need a support that has Teen Jeans stun passive + team damage on enemy match 5. The damage part isn't even really that necessary but would at least give the support 'some difference when leveled up' so it can be something quite minimal. That would curb so much AI cascade nonsense.

KGB

revskip

@KGB said:

@sambrookjm said:

@nwahs086 said:

@slidecage said:
just went against a aunt may emma team

even though they had enough ap to wipe me out with aunt may red the ai would only fire emma red

they need to add something that tells them to fire what power

You don't want the AI to be better because losing repeatedly because of cascades and the AI doing stuff out of your control isn't fun.

Having just got obliterated by a bunch of mThor cascades that took out two of my 550+ LVL (with boosts) characters, I approve of this message.

I read the description of what happened in the 'worst moments' thread.

We really need a support that has Teen Jeans stun passive + team damage on enemy match 5. The damage part isn't even really that necessary but would at least give the support 'some difference when leveled up' so it can be something quite minimal. That would curb so much AI cascade nonsense.

KGB

That would be awesome tbh. A healthy counter to the fantastic 4 car would be dope.