Say you have four friends. Three are skilled slayers, lets say Onyx 1500 on the old scale. And one who is average - Gold 900.
They play together. The Gold player gets to hold the oddball while the other’s get busy on the KPM.
As a team they go through placement and lets say they all come out as Diamond 4. Under the system that people are proposing, everyone gets the same CSR gains, so all four of the players are now Diamond 4.
Is that right?
Now the players separate and go into match making solo. They are all Diamond 4 and the matchmaking is using their CSR to find their next games.
The objective player gets chucked into a game with three other D4 players and they play a team of four other D4.
It’s going to be a massacre - because the player is either going to have to suddenly fight 1v1 at D4 level - or his three team mates are going to have to step up to Onyx level to compensate. Chances are they are going to let their team mates down over and over again until their CSR drifts back to Gold.
And on the flipside the Onyx players are going smash their games.
One of the things KPM did very well was isolate the rankings of individual squad members. You would still find players who were under, or over, rated if they spent a lot of time in a squad - but they were at least ball parked - and their MMR could adjust quickly when they branched out solo.
The big difference is that the skilled slayer can hold the oddball. They just choose not to (either tactically, for ego, or under the false pretence that it stops them from ranking up). But the objective player can’t slay. They can only contribute by holding the oddball.
Josh Menke summed it up best when he said anyone can run the flag. The real skill is killing people to win the flag and clear the path.
And God help any team who gets matched with two objective players.
I think what 343i have chosen to do is a giant mistake; though to be fair, I don’t have all the details, so we’ll see I guess. But, the statement that they’ll be matchmaking “primarily” on CSR as opposed to MMR introduces all kinds of concerns.
To me, the solution is simple. Keep CSR in its proper lane as a visual reflection of a player’s mode/playlist TrueSkill MMR that doesn’t oscillate dramatically, adjusts strictly in accordance to match outcome, and resets seasonally. In fact, I think they should remove the extended grind component that they’ve incorporated into it at the higher skill levels (above 1400 MMR). It should definitely not be use it to match-make when you have a far superior metric in MMR that’s designed precisely for that purpose.
Instead of forcing CSR to try itching that progressive grind aspect they should introduce a secondary visual indicator for people to invest their time grinding. Have it based on achieving playlist wins. For example; +1 for achieving an expected win, +2 for an unexpected win, +3 for a serious upset, 0 for a loss, -1 for a quit, -2 for being the 1st quitter. Tie the point accumulations to playlist military rank symbols. For the ranked playlists, allow these new visual indicators to feature more unique representations in comparison to the social mode/playlist military indicators. Visually tie them to a player’s corresponding CSR tier level.
This way the CSRs can finally fulfill their intended role in a more pure fashion. And people get access to a visual indicator that provides them with a more tangible reward for their invested time toward achieving victories in a particular mode/playlist.
Add on top of this an XP progression for Spartan characters. Spartan Rank [SR]. And players will be happy to see that their general time playing the game is being tracked, recognized, and hopefully rewarded by earning store credits and/or unique customization items.
They only “feel” like they don’t because people keep referring to them as “hidden” and coming up with crack pot theories about them. They especially become an easy excuse for people who are frustrated they aren’t ranked as high as they think they should be.
The average punter naturally starts to distrust them.
If 343 were more open with the communication everyone would see that your MMR goes up and down (with form) and how your CSR follows along, kind of like one of those lines of best fit you see in scientific papers.
You would see that your MMR dropped that night you were “off” and had a bit of bad luck with team mates - but that your CSR only dipped a bit because your MMR was starting off high because you had that great run over the weekend.
I’m convinced if we could see our MMRs on Waypoint. Graphed over time. All the different modes, ranked aggregates, and even global. People would be a lot more understanding of how this all works.
You would be able to apply context to see what is happening.
It’s a pity Josh isn’t here any more. He was happy to print peoples MMRs out for them.
i dont think thats right. placement matches willl not be based on wins only, i believe they will be the same as they are now. After the placement matches is when this new system with your csr will take place I believe.
A lot is going to hinge on what they mean by that word.
The simplest fix would have been to reduce the scale. It’s not precise enough for 1-1800+ anyway. And I agree whole heartedly. Cap it.
1 to 117 is the best suggestion I’ve heard so far.
I’ve called for something similar a few times. More of a substitute to the champion tag.
Great minds and such
Mine was to allocate points at the rank you are. And then the military rank on top. So you could be a Season 3 Diamond Three Star General. They could also have regional and global leader boards to help people show off the points they earned.
The key is to attract players to keep coming back.
At the moment a lot of players hit the rank they want and then stop playing.
Yep.
And they said that people were getting upset with the CSR make up of the teams at the end of the match.
No. They were getting upset with the unexplained gains and losses of CSR.
And they were driving themselves insane chasing shadow metrics like KD that made no difference.
All 343 had to do was explain things better. And give us tools to look up our ranking stats.
I liked, or agreed with, pretty much all of your responses to me… except for this part.
I don’t see anything wrong with the current scale of the CSR system. And what may I ask are you basing your comment on regarding it being imprecise at its current scale?
Also, while I once believed it would be beneficial to remove the Onyx numerals from post-match carnage reports, suggesting that they should only be reflected within a player’s career statistics page, I certainly don’t think the ranking system requires a cap. If people are able to prove that they’re skilled enough to continue pushing the indicator ceiling further and further out, so be it. I also think they should bring back the Champ badges for the top (however many) Onyx players that possess an MMR beyond some minimum level (1600? 1800?).
I’m not really a fan of the 1-117 indicators within the CSR system, but I do like its potential if it were incorporated as part of that secondary mode/playlist progression system or as a profile’s XP progression system similar to Halo 5’s SR152 system.
The scale suggests skill changes on a game by game basis.
It doesn’t.
When we were 1-50, for example, people were more than happy to let their skill evolve over days to weeks. Which is the way it should work.
There wasn’t this hypervigilant focus on 1-2 points. Every game.
On a couple of points.
The way we currently have the MMR bouncing around (responding to form fluctuations) and the CSR travelling a line of best fit. So if your MMR is ostensibly 1300 - it’s bouncing around, say, 1250 to 1350. And your CSR is oscillating between 1295-1305. So you lose the precision inside that. Your CSR at 1298 is really not any different to 1302. You’ve just caught it at a different phase of MMR vs CSR vs the level of opponent MM just found for you.
Just the fact that it’s approaching and never reaching an arbitrary scale that based on probability (your MMR curve) intuitively tells you a single point on that scale is not going to have precision.
The system will tend to rank server populations into their own bell curve. There will be differences between servers. And the further away you get from each other, the more independent those curves behave. You can’t say with any confidence that a 1300 US East is the same as 1300 US West. And you certainly can’t compare those with a 1300 Europe.
Anecdotally, I find that the tiers are pretty accurate. You can say a D3 vs a D3 will be a good contest. But that’s essentially breaking the population into 1-36. That’s a bit too small.
I don’t think you could ever go back to 1-50. There’s too much emotional (nostaligic) baggage there.
Hence the 1-100 (or 117).
I do. If just for the toxic ways people are trying to extend it.
If you see an 1800+, what’s your first thought? Is it wow, they are genuinely in the top 0.1%… or do you think - how did they game the system to grind that out?
And we know a lot of people do it. To the level that they are desperate to reset the population every three months.
Yep. As long as they don’t go down the dark path of top ‘200’. That would create an unholy toxic grind for points.
I would have thought the top 0.5% sounds about right?
i just reaiised that system i just made up is not even based on csr matchmaking. my idea is competely based on mmr that only changes every 10 games. it should only change every 10 games because matchmaking cant always be trusted to start changing mmr rapidly unless your looking for smurfing behaviour.
basically mmr shoulld change every 10 games unless tthe smurfing detection system is activated. when i say change every 10 games, i mean get the average from those games. it wont always be 10 games either, mmr could change in 6 games if you won 6 games in a row. the system will update your matchmaking mmr and give you harder games until it eventually finds your level where you are winning about 5 out of 10 games
The system does take recent games into account. More recent ones having more influence. So it already kind of averages out your last ‘x’ games.
I think it can.
Which is what it already dose. Very quickly. To the point that people actually accuse the algorithm of cheating.
Overall I get what you are wanting to do. But I think it already does it.
The problem is we all look at our last game results and pass judgement. But we have no context. All we can see is the player’s CSR. We don’t know their MMR (higher or lower). That D3 may have underperformed - but was that a one off - or are they on a form slump and that was what the system expected (with an MMR lower than D3).
Similarly - that G3 that did better than expected. Their MMR may be higher. So it wasn’t them that tilted the game result unexpectedly.
Again - a simple solution would be for 343 to a) explain the system more clearly, and b) give us the data we need for context. If we could see a player’s MMR and CSR plotted vs time - I’m sure it would all make sense.
But now we are going to me matched by CSR. People’s form can be well above or below this. So potentially less even matches on form.
thats my point. its a yo yo effect. its like hey you lost this game, let me give you an easier one. what you end up with is a system that is trying to be too perfect that ends up giving you easy wins and easy losses. Only games with huge populations can afford systems like this where minute tweaks to mmr dont matter.
Its better to just wait, relax a bit, find the average from a bunch of games (who knows, you might improve in that time) before changing any mmr. Changing a little mmr too quickly can have adverse effects on a game with low population and a large skilll gap.
