You can literally see this game trying to find you throwers as a reward for winning a few games.
8 min queues for four games straight. All wins.
Suddenly 12 minute queues to find us the two worst Reins of all time two games in a row?
I see how it works. This is only the 1000th time seeing this over 18 seasons. So obvious.
I think the losses have triggered you and are clouding your judgement.
If they give people rewards for queuing in off roles, quite often you will come up against/or along side, people who don’t really understand those roles.
You mean it’s almost like the matchmaking puts good players with the worst players, in order to try and elevate the worst?
Weird.
Guess I’ll just post this whenever I see a “matchmaker doesn’t do anything iffy” post:
Recently it was slipped that there are legacy values. Then take this Blizzard Activision patent into account:
“[0016] In another example, if a player has been performing poorly (e.g., getting killed at a rate higher than the player’s historical rate), the scoring engine may dynamically adjust one or more coefficients to match the player in a game that will improve the player’s performance. For example, the player may be matched with easier opponents, matched with better teammates, and/or placed in a game that is more tailored to the player’s preferences (e.g., players that play in games more closely aligned with their preferences tend to perform better).”
From: `https://patents.google.com/patent/US20160005270A1/en
No words are really minced there. We need this quote to be unequivocally rejected that nothing of this sort occurs in Overwatch.
Or that the abstract for this patent describes matchmaker specifically placing people with unique skins in games with people without them, to encourage purchase of those skins.
I’m pretty sure that’s just the matchmaker questioning if you belong in a higher rank by testing your skill after an unusual spike in performance.
Well known that dps blame the tanks, tanks blame the healers and healers blame the dps. Seems you are sticking to form.
So, it’ll decrease your rank and place you in games with lower ranked players?
How dastardly of them.
Ok, obviously you have read the patent and clearly see the whole picture very easily. Can I ask you two questions?
Why would they need a patent for that? (I’ll give you a hint, what you suggest couldn’t even be patented, as it lacks the key tests applied to a patent, as I’m sure you’re aware)
Why would they compare you to a historical rate to do so?
Sorry, 3rd question I need to ask (something something Spanish Inquisition):
OP says he won 4 games in a row and is bothered by the even longer queue times because of it
this man, a scholar, claims that OP is upset due to losing
Por que?
Well, here’s a tidbit from an argument I already had with someone on the forums a few weeks ago about that very same patent.
Sadly the thread was closed and then unlisted so I can’t link you directly to it.
As to your questions.
They need a patent for it because it’s a process they created and they want to prevent other people from implementing it without paying them. It absolutely can be patented for what I claimed, and that’s what it’s used for - to use deep learning to improve matchmaking algorithms. Processes can in fact be patented, I’m not sure why you think it can’t be.
I’m not sure what you’re intimating? The historical rate is called an anchor, and is used in all forms of complex matchmaking. You can see it in CSGO with their derivation of the Glicko-2 model, it’s used in basically everything except basic ELO (which is really only used for chess and trivia). Anchor is used to help determine volatility (deviation and uncertainty).
Giving better teammates is just one of the facets of the process, and is pretty much directly stated to only occur when there’s determined certainty of a player’s rank and they’re underperforming considering their own past personal stats.
This basically just says that it’s using a determined deviation, and the player fell out of that deviation. It essentially is an anti-smurfing tool. This is basic matchmaking information.
All uncertainty about this patent, which people have attempted and failed to use on these forums multiple times, is purely because of 2 things: People don’t understand legal jargon, and people don’t understand matchmaking.
There’s nothing out of the ordinary with the patent except for matching players with cool skins.
Everything else you said melts away with this comment. Congratulations on posting a lot of babble, when you actually agree with my post.
You cannot match with “better players” in a lower SR match, unless the matchmaker knows they are better players. The reason for doing this, as described in the patent, is to give you a win.
Question 4:
What happens to the other team when the system “gives you a win”?
It never “gives you a win”. That’s not what it’s describing. Literally basic matchmaking principle - you have an expected rating, and depending on your performance within this expected rating your matchmaking is going to change.
Yes, you can. You are not matched based off of SR at any point. You’re matched based off of MMR. Your MMR does not strictly decrease by a proportionate value to your SR after a loss. If there’s a random losing streak against a high ranked player the matchmaker already has a decent idea of that player’s anchor and their deviation, they’re going to try to prevent them from leaving that deviation. If the player continues to underperform then their MMR and their expected deviation is going to be adjusted.
That is all that is being described by this process.
And the matchmaker should think or care? I’ve been 2800, and spent most seasons in Plat. Now i’m seeing idiot games at low gold and the MMR. Robot needs to give me hard mode just because i won a few?
Well, that’s rigged.
How about just make random matches and god forbid someone ranks up without a fake challenge from the matchmaker itself.
This kind of thing is exactly why this game is so toxic. It’s on Blizzard.
Sorry, wasn’t wanting to get into the MMR/SR thing, was just behaving as if there was one number of relevance. So for your sake, we will go with MMR, and do an example:
Player A has just lost a match. He lost it so badly his stats were well below his historical rates.
Normal competitive system: He loses MMR. Gets placed with players at new MMR.
Patented system: He apparently gains MMR (according to you) and is placed with higher MMR teammates.
This line also causes great concern. This can only be done by handicapping matches, or by the method described in the patent.
You described handicapping in the exact way a lot of people are complaining about.
Never stated that. It’s far more likely that their MMR will decrease but they’ll be placed into a match where they’re one of the below-average MMR players to balance the team. MMR is based on a +/- 3 system with precision down to 0.01 at the very least (stated by the Devs), it’s difficult to find 6 people for a single team (plus 6 for the other) that have identical MMRs, so it needs to create an average that both teams match, then find players to create the average.
Again, fundamental.
You clearly have no idea how a “normal” competitive system works. In many other systems (such as, again, Glicko-2, the most popular matchmaking system for online games in existence) it’s fully possible to gain rating after a loss.
In Glicko-2 it’s also explicitly described that time plays a major factor in deviation determination. If a losing streak occurs such that all losses are in a single session then their weight in effecting deviation is minimized, as the system understands that streaks occur. This causes the new re-determined deviation to be identical to the prior deviation + anchor.
No, it does not “handicap” matches. It prevents win/loss streaks.
Thus artificially engineering wins/loses. Handicapping.
Cool story.
Why would anyone think this is the purpose of a match maker in a competitive setting.
Everything you describe is artificially engineering wins/loses. Handicapping.
The purpose of such a system is to keep smurfs from stomping low level matches for 50 matches straight until they eventually reach their proper SR. If you’re suddenly performing much better than usual then the system checks to see if you belong higher than where you are.
You clearly just do not understand how matchmaking works. All matchmakers in and of themselves are artificial processes of creating wins and losses. It facilitates a win or a loss by matching players into matches with expected outcomes, then makes changes determined by whether that outcome occurred or not.
Also, I find it hilarious that you would seem to ignore the Glicko-2 model considering it is the most popular baseline matchmaking system used in existence, and nearly every single video game uses some slightly altered form of it.
Including (Confirmed): CSGO, TF2, Dota Underlords (autochess), Guild Wars 2, among many, many others. League of Legends uses a system derived from ELO (which Glicko itself is derived from, and their additions basically turn the model back into Glicko-2). The only popularly used ranking system that isn’t Glicko or ELO (both of which are similar) is TrueSkill, which is often used combined with the aspects of Glicko-2 to create a proprietary ranking system for online games.
As a sidenote, it’s also possible that Overwatch is now largely based off of the TrueSkill 2 model since Role Queue’s implementation, since TrueSkill 2 came out sometime in 2018, shortly before Role Queue began being developed.
This would mean that the process I described is likely even more pronounced, as TrueSkill 2 takes into account factors such as absence from the game, pre-made groups, player statistics and experience (duration played), and other factors including those already present in Glicko-2 and TrueSkill, such as win-streaks and session duration.
For those of you wondering why you’re suddenly dropping/climbing in rank, that could very well explain it.