Jump to content
Sign in to follow this  
Vetoed

Potential Halo 5 Skill Rating Exploit

Recommended Posts

I'm not surprised at all that you were the one to figure this stuff out lol!

 

But yea these problems need to be fixed asap before the game gets released.

  • Upvote (+1) 1

Share this post


Link to post

Josh Menke usually responds on Twitter, but I don't doubt you're on to something, even without his clarification.

 

Anyway, @@Sal1ent.

Unfortunately, I think Menke quit working at 343 and on the ranking system. He might be able to offer them information to work on however... I'll see if I can reach out to him on Twitter.

Share this post


Link to post

Great catch @@Vetoed!

 

This exploit was also in Halo 4's CSR system; searching with low ranked teammates and beating decently ranked opponents boosts your rank significantly.

 

I hope 343i gets some advice from Joshua Menke on how to resolve this issue and implements those resolutions in the final game.

  • Upvote (+1) 1

Share this post


Link to post

There also appears to be an issue where ranks from different playlists carry over onto the next one after placement matches, but that's another problem altogether.

This is probably an intended feature, not a bug. It's pretty unlikely that a top-ranked player in one playlist would be a low-ranked player in another playlist. Using previous performance as a starting point removes the "noobstomping" factor from switching playlists, and also means your placement matches will be more evenly-matched.

 

The downside, of course, is a relatively unskilled player could get their first rank in an "easy" playlist, switch to a "hard" playlist, and sit on their rank. This can be handled nicely with either a season system or a per-playlist inactivity penalty.

Share this post


Link to post

This is probably an intended feature, not a bug. It's pretty unlikely that a top-ranked player in one playlist would be a low-ranked player in another playlist. Using previous performance as a starting point removes the "noobstomping" factor from switching playlists, and also means your placement matches will be more evenly-matched.

 

The downside, of course, is a relatively unskilled player could get their first rank in an "easy" playlist, switch to a "hard" playlist, and sit on their rank. This can be handled nicely with either a season system or an inactivity penalty.

IMO if this is by design, it needs to be changed. I can be at good regular slayer, but that doesn't mean my ranking should carry onto the HCS playlist. Beating noobs is just part of the placement process and skill attribution. It can't be avoided with a ranking system like this. 

 

Ranks transferring over don't prove actual skill in a playlist, as all it takes is 10 placement games to obtain the same rank again or so.

 

As for seasonal ranks, it's a different beast. It could be easy to tie in with actual HCS seasons, but I think it would be interesting to have leaderboards to promote grinding for ranks before the season ends. Starting over from one season to another should factor in your previous division after placement matches before re-assigning you to a new one. That's when ranks should carry over, as opposed to right now, and it should be kept within the same playlist.

Share this post


Link to post

IMO if this is by design, it needs to be changed. I can be at good regular slayer, but that doesn't mean my ranking should carry onto the HCS playlist. Beating noobs is just part of the placement process and skill attribution. It can't be avoided with a ranking system like this. 

 

Ranks transferring over don't prove actual skill in a playlist, as all it takes is 10 placement games to obtain the same rank again or so.

I absolutely agree that a rank in one playlist isn't proof of skill in another playlist, but it is at the very least an indicator of skill. When using a previous rank in a new playlist, the system could place players very conservatively (i.e. one or two ranks below their previous-playlist rank), but start the uncertainty very high (meaning players will quickly move up or down if they were not placed correctly).

 

Starting over from one season to another should factor in your previous division after placement matches before re-assigning you to a new one. That's when ranks should carry over, as opposed to right now, and it should be kept within the same playlist.

Yes, this is what I mean when I say seasons. In Menke's previous work with Starcraft 2, your hidden skill rating carried over between seasons, so you'd usually be placed in the same division in the next season. The seasons were essentially there to keep the inactive players off the ladder ranking list.

  • Upvote (+1) 1

Share this post


Link to post

I absolutely agree that a rank in one playlist isn't proof of skill in another playlist, but it is at the very least an indicator of skill. When using a previous rank in a new playlist, the system could place players very conservatively (i.e. one or two ranks below their previous-playlist rank), but start the uncertainty very high (meaning players will quickly move up or down if they were not placed correctly).

 

Yes, this is what I mean when I say seasons. In Menke's previous work with Starcraft 2, your hidden skill rating carried over between seasons, so you'd usually be placed in the same division in the next season. The seasons were essentially there to keep the inactive players off the ladder ranking list.

That's why there should be a high skill displayed on the service record. If the visibility of a person's high skill is clear, there's no need for ranks to carry between playlists. It's easy to estimate someone's skill, but the grind in other playlists is still something to strive for instead of starting at the top.

Share this post


Link to post

I had a feeling it worked like this a few days after release and forgot. Some kid on my friends list I met in MCC was playing with us and he got placed in Silver because he played his placement games alone. Me and my 3 friends were all Onyx II or III and we hit semi pro in like 10-15 points playing nothing but silvers, bronzes, unrankeds, and maybe one gold. I was joking that playing with him was like playing with a booster in H3, I guess I was kinda right.

Share this post


Link to post

I thought this warranted its own thread because not a lot of people pushed Halo 5's ranking system to its limit, and this is deserving of attention before the game launches.

 

Right now I don't have access to an Xbox to test it out, but after investigating a bit and talking to a few people, I've come to the conclusion that similar to previous ranking system (Halo 3 and Reach Arena), it is possible to progress faster through ranks while searching with much lower ranked accounts. I'm also speculating that searching with weaker ranked teammates provides a form of safety net when losing, as the game assumes it shouldn't be as punishing of a loss because of the lower ranked teammates.

 

Look below at this person's service record and searching habits.

 

B7caoFyCUAAy4P0.png

 

f81087c00ae1fe7fba31b19935b4e2b9.png

 

Of course, this would require rigorous testing to prove how significant the impact is, but I figured I'd put it out there. Arsenal himself told me sometimes, searching without the booster account results in +0 wins on the rating, and the main idea behind it was not to lose a lot of points in case of a loss.

 

There also appears to be an issue where ranks from different playlists carry over onto the next one after placement matches, but that's another problem altogether.

 

Thanks for reading.

 

@@Sal1ent

Thank you for finally exposing this exploit vetoed. Hopefully this gets patched before the release of the game or its going to be Reach arena all over again. I Played ssyphons team the other night won 5 games, and lost 5 games against them. Lost over a days worth of ranked points (50 or so points) due to the opposing team iron boosting. Not only does it benefit the team boosting, but it also effects the team they're playing against drastically.

  • Upvote (+1) 2

Share this post


Link to post

Only solution to this would be to making the amount of points each win or loss as a static number, which will be the same for anyone no matter the rank.  Of course, this will introduce the new crybabies of crying that they carried a lower level team against a team of semi-pros and didn't really rank up.

 

In other worse, no matter what the change is crybabies will still exist.

Share this post


Link to post

If this doesn't get fixed, and it is true, it would be out of control as new accounts are gold now. Crazy how the same problems appear 7 years later.

  • Upvote (+1) 1

Share this post


Link to post

Unfortunately, I think Menke quit working at 343 and on the ranking system. He might be able to offer them information to work on however... I'll see if I can reach out to him on Twitter.

Where did you hear that? was it confirmed?

Share this post


Link to post

Where did you hear that? was it confirmed?

Yeah it's confirmed--he has said it himself on Twitter.

 

He has been answering questions on Twitter about the ranking system still, though

  • Upvote (+1) 1

Share this post


Link to post

Where did you hear that? was it confirmed?

 

 

So update on the situation.

 

I've been talking to Menke for a bit and he explained to me that this is known and was done by design, to allow higher players to play with weaker teammates without being punished. He then said that 343 would fix it if it became the main way to level up, which is inevitable given that we're dealing with the Halo community here, and this exploit can effect opponents' ranks in a drastic manner in case of a loss. You can open up spots in the pro division by knocking people out of their rating, so this will turn into a problem at one point or another.

 

I explained that social playlists are the best way to allow high skilled players to search with weaker teammates without being punished, and therefore, catering to weaker teammates in ranked was a bit superfluous on top of offering social hoppers. He also mentioned that in Starcraft 2, players also have a choice to just search an unranked version of available ranked playlists, but I always thought that offering an unranked version of a playlist that's ranked to begin with will just hurt the population and sometimes impact search times or match-up accuracy, so it should be avoided. Social hoppers should offer a distinct experience (E.G. ranked slayer is 4v4 on symmetric maps, unranked slayer is 5v5 on whatever maps).

 

In conclusion from our exchange, I think the best approach is to negate the impact that low level teammates have when searching at the top level. For instance, in a pro tier match-up, a bronze level account (or ANY rank outside of the usual Matchmaking skill range) should, by default, have the same impact on the end game result as the best players on their team do. The system should also be changed to always to progress a minimal set value, so sometimes, wins don't result in +0 rating.

 

Party size restrictions also need to be implemented to match a team of 4 with another team of 4, to3 with another to3, and 2s / solo as "randoms". Random matches should provide the biggest boost due to the risk of getting bad teammates, in order to make it balanced on the long run. Assuming your average win rate goes from 80-90 in a team of 4, to then 60-70% as a random, the rating boost for searching as random could be about 15% on top of what you would get normal, and as a team of 3, 5-10% higher than what you'd get as a team of 4.

 

Of course, these are significant changes, but I think this is the best way to handle the whole situation. All feedback is welcome here, I haven't done calculations to figure these numbers out, but I think all of this makes sense.

  • Upvote (+1) 6

Share this post


Link to post

I played those guys. Lost 50 points the first game. Beat them the second game and got one point. :(. Refused to play them again.

 

I think they are pussies on the one hand, but on the other hand its good the bullshit gets exposed during the beta.

 

You're right in thinking that this can't be left to the Halo community. Many people will start doing it, unfortunately.

Share this post


Link to post

Unfortunately, I think Menke quit working at 343 and on the ranking system. He might be able to offer them information to work on however... I'll see if I can reach out to him on Twitter.

Where did you hear that? was it confirmed?

 

 

Menke has returned to Blizzard/Activision (likely to work on the rankings for Overwatch)

  • Upvote (+1) 1

Share this post


Link to post

I explained that social playlists are the best way to allow high skilled players to search with weaker teammates without being punished, and therefore, catering to weaker teammates in ranked was a bit superfluous on top of offering social hoppers. He also mentioned that in Starcraft 2, players also have a choice to just search an unranked version of available ranked playlists, but I always thought that offering an unranked version of a playlist that's ranked to begin with will just hurt the population and sometimes impact search times or match-up accuracy, so it should be avoided. Social hoppers should offer a distinct experience (E.G. ranked slayer is 4v4 on symmetric maps, unranked slayer is 5v5 on whatever maps).

As a side note to this point, in SC2 the unranked and ranked playlists both draw from the same player pool, and you have a separate, hidden rating for your unranked games. This means there's no impact to search times or skill matching, but it does come with a whole new set of issues (ranked players still get points from games against unranked players, so there's the possibility of unranked players giving free wins or generally not trying very hard in their games, etc.).

Share this post


Link to post

As a side note to this point, in SC2 the unranked and ranked playlists both draw from the same player pool, and you have a separate, hidden rating for your unranked games. This means there's no impact to search times or skill matching, but it does come with a whole new set of issues (ranked players still get points from games against unranked players, so there's the possibility of unranked players giving free wins or generally not trying very hard in their games, etc.).

Can't say I'm familiar with how SC2 works, but if you have players who are there to screw around with weaker teammates vs. a team going in to win it in the same pool, I'd assume it can lead to mismatched parties... I'm not sure Halo should adopt this system.

 

Social playlists have done it well in the past. You can have a hidden ranking in social for better matches, just these playlists should offer an experience different from ranked playlists to give them both separate player pools that don't overlap each other.

Share this post


Link to post

Lessen the impact by setting artificial bounds on Truskill (or whatever value is determining rank) as x% of the mean. Essentially treat irons as golds in a semipro game. Or reduce the potential gains by a higher ranked player. Remove the incentive by making it ineffective.

 

Why Menke and co. Would put in a system that needs active policing is beyond me- it hasn't happened before and it won't happen now. Shame Strongisde didn't explain to Menke how he won that truck before he left.

 

Noticed the +0 after some games, knew that Reach-like iron boosting would be a problem eventually. Thankfully, I haven't run into any clowns actively exploiting it, rather just poor groupings of randoms by matchmaking (While unacceptable IMO, is another problem entirely).

  • Upvote (+1) 1

Share this post


Link to post

but on the other hand its good the bullshit gets exposed during the beta.

Reach Arena, H4 CSR, H4 Top 200, H4 Truck challenge thing all had the same exploit- which has completely busted ranked Halo MM for years- it's known and not remotely new. Baffled it made it this far, will be incredibly and pleasantly surprised to see it fixed at retail.

  • Upvote (+1) 1

Share this post


Link to post

Reach Arena, H4 CSR, H4 Top 200, H4 Truck challenge thing all had the same exploit- which has completely busted ranked Halo MM for years- it's known and not remotely new. Baffled it made it this far, will be incredibly and pleasantly surprised to see it fixed at retail.

It probably won't be fixed.

  • Upvote (+1) 1

Share this post


Link to post

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use & Privacy Policy.