So..Rip off meta players? (and one tricks)

General Discussion
21

It doesn’t matter if we both agree on it or not, the reality won’t shape itself to fit what we agreed upon.

I would like it to not ban off meta heroes.
I would like it to not ban one tricks.
And i don’t know if any of that will, or not, happen.

Basically, i don’t know, and we will have to wait and see.

22

BELIVE ME

i have tried

they report you anyways

people report me for playing bastion EVEN IN QUICKPLAY

want a real example? in comp, a few hours ago,
we won thanks to a sym player who placed a clutch tp, the sym player iswhat made us hold the point

team answered with “report sym”

also, the enemy team had a bastion in the same match that completly rekt out team, they lost anyways but still

what they said? “report bastion”

23

I think skepticism is fair and natural.

I just think people are really too skeptical. We don’t know anything of the system yet, or how it would work. But historically, machine learning detects “good” patterns and “bad” patterns. So I imagine a lot of focus will be put into learning from the good patterns from a good source.

We’ll see, but I think the skepticism is absolutely fair.

24

You can’t be too skeptical. There are only 2 options:
You are either skeptical, or you are not. There is no in between or any specter.

What I THINK you meant by skepticism being on a specter, where you can be a bit skeptical or a lot skeptical, that implies, i think, that apply skepticism, but not here. Which isn’t really a skeptical approach to not apply skepticism.

25

There are other threads where people are already bashing the system.

So there is a “too” skeptical IMO. It’s being hostile and overly defensive, and adamantly planting your feet. :stuck_out_tongue:

1 Like
26

Well that means there is a lot of skeptical people, not that they are too skeptical.

27

Not always. If you want the system to make a decision on if it thinks something is right or wrong you need humans to label examples, but if you want the system to try and uncover patterns hidden in the data you don’t. In ML jargon this is supervised vs unsupervised learning.

28

…I don’t understand?

If someone is skeptical and saying, “I have concerns”, I think that is fair.

There are other people who are saying, “This system is broken, we don’t need it. It’s just going to punish people. You’re stupid to do so” or something along those lines. I think that is overly skeptical and drawing too many conclusions when we don’t know anything.

29

Read my thread where i raised this question and got a dev response, and by thread i mean only the first post and the paragraph where i linked an article, i was talking about a chatbot that learner from humans some obscene things to say. It was funny, but machine learning doesn’t mean it won’t adopt bad things.

30

But machine learning can always have human assistance. I don’t think they’d create a system for detecting bad behavior through mechanics without human assistance :stuck_out_tongue:

31

The chatbot you’re talking about was obviously loosely monitored and guided - Probably by design.

1 Like
32

Unsupervised learning is when you don’t care about accuracy but looking for pattern recognition. That isn’t relevant in this case. In a reporting system accuracy is extremely important as you want to make sure no one gets banned without justification.

33

You are completely right. They will try their best to do what they intend to do.

The question is, can they not let the false reporting data in, as some valid data? if they can, there will be no problems.

I guess people don’t have faith because a lot of them are being penalized for just playing the game on offmeta heroes and they feel its extremely unjust.

34

It could not possibly be worse, really. So, I am not too concerned.

35

Well just report them then, although i highly doubt that story is true, there is no reason any of your teamates would report the bastion or symm unless they were tilted.

36

Fair point. Let’s hope they will monitor this trained agent closely.

37

Here’s hoping. It’s got potential to be VASTLY superior to the existing autoban system. It’s also got the potential to be catastrophically terrible.

38

didnt machine learning end up being racist before ?? why are we trying this again ?

39

I don’t understand what this machine can bring to the table, are troll picks safe or not? By troll picks I meant, not the troll picks who throw games, the one tricks!

40

If you dont belive this story you can literally google any of this stories from anywhere…

they happen very often…

an example you may belive…

Steevo, a gm sym player had this happend to him where he got a warning for “griefing” despite winning 8/10 placements

Or this hanzo main that got banned a short while ago i cant remember his name