Machine Learning Systems

Alright so if i look back to a dev post of around 2 years ago we were promised something that could have solved many issues.

Machine Learning Systems that battle against Abusive Chat and Throwing

I’ve seen many comments from players who think that their reports are meaningless, but we want to stress that they are actually very helpful and incredibly important to improving the Overwatch community. The in-game thank you messages we recently added let you know that the time you spend making a report is indeed making a positive contribution. Just this last week, we corrected an issue that prevented many players from receiving these messages, so players should now start seeing even more feedback about their reports. To further improve these systems, we’re doing a lot of exciting work to develop “machine learning” systems to assist in accurately identifying abusive chat and gameplay sabotage. These technologies will work together with player reports, empowering the community to quickly bring attention to bad behavior so that the appropriate steps can be taken to discourage or prevent future bad behavior from ruining others’ experiences playing Overwatch. These same systems will also protect players from false reports.

Source: PLAYER BEHAVIOR IN OVERWATCH Post date: 2018-05-03T00:11:00Z

We haven’t had an update on this after those 2 years, unless i’m missing something. If i am, then i’d be happy if someone can update me on the situation.

Sure automated systems might result in false positives, though it can be a bumpy road towards a better place.

I think these systems can have a good place in present Overwatch since we already have alot of issues regarding rule breakers.
I mostly believe this is due to COVID-19 straining Customer Support resources, though i cannot speak for Blizzard.

These systems might relief Customer Support from handling alot of reports, but then they can use that time for appeals if false positives do happen.
The more this system is online and being tweaked the better it will perform.

I’m kind of disappointed that they never worked out this despite promising it.

Now of course i can be wrong here and that they’ve already put such systems into place, hence why there are that many false postives.

Though i’d love to hear an update on this previous post.

Do you think machine learning systems to battle against rule breakers still have a place in Overwatch?

Anything you would add to these systems to improve them.
Or anything you’d like to see change in our current reporting system?
More info on what happens during the process of a report, maybe even want to change the rules so smurfs can be reported?

Let me know it all in this thread!

2 Likes

I think it was abandoned as part of the laying off of 8% of it’s staff.

1 Like

I’m pretty sure these can be relocated to different departements, and of course this system won’t be perfect meaning they might even stay in the same position.

No I mean they DID lay off 8% of their staff… not relocate to different departments.

Past Tense. Teams closed down, 8% of the staff was given “a don’t come monday.”

I’m sure such a tool can work, if properly trained and revised.
Just look at how precise facebook ads or amazon ads are.

2 Likes

Oh i see what you mean now, sorry for the misunderstanding

1 Like

It’s ok, I would MUCH MUCH MUCH rather they kept the project.

I think it was one of the most important projects which was being done for overwatch (and that workshops).

1 Like

800 people is only 8% and we still get extremely slow content releases? Think maybe it’s high time the share holders fired the president and get someone who can actually run the business.

Blizzard does more than just Overwatch, and the current gap is around OW2, AND Overwatch is more than just Overwatch dev. You got OWL and the like, and they take HR resources as well.

I don’t think I should comment on that.