I’m surprised that the devs have reversal/consolation in the game.
It’s quite obvious that the matchmaker has questionable capacity to discern differences in skill even in gross terms, let alone the fine, nuanced deviations between players in the same match that would be required to award reversal/consolation. This isn’t really an opinion either. The devs have conceded that they are only able to give reasonable matchmaking between ranks and that skill deviation in-game is entirely dependent on the circumstances of the given game. So why admit this, yet somehow (contradictorily) purport to be also be capable of discerning the minute MMR differences that would be required to confidently assign the reversal/consolation conditions?
It’s oddly silly logic errors like this that engender low confidence in the game’s development. It seems crazy that no one speaks up in a meeting about obvious mistakes like this.
It’s easy for a machine to tell which number is bigger.
The accuracy of the number is not always great. Since the playets might have gotten derankers/trolls in their team, which lowers their number. Or perhaps their enemy teams have had trollers and thus their number is too high.
The matchmaker would be better if no one trolled or deranked, bought boosting services or pre levelled accounts.
It’s upto the playerbase. The mm itself is simple.
Right, that’s my point. The goal of the MM is to accurately determine differences between players. The accuracy is quite low (understandably), but they implement mechanisms (consolation/reversal) that require exceptionally high accuracy to work as intended. Very odd.
Yes. I’m not pointing out the mechanics of it. I’m fully aware of how they judge the numbers. But the numbers are being used to reflect a characteristic. I’m specifying that it fails to be accurate in this purpose. It rewards or punishes a little bit extra based on the assumption that it can tell one team was advantaged and one disadvantaged. We know that it cannot tell this with a reasonable degree of accuracy, so why implement it? I understand they have to use something for the matchmaking itself. It’s the best job they can do to match games. But if you know your numbers don’t do a reasonable job of telling which team is truly advantaged, then don’t bother implementing. Seems a simple decision.
I’m sorry but if you described a generic IF AND statement as “A.I” then you don’t know what you’re talking about.
The matchmaker takes too long to calibrate your true rank, and is not accurate because they are too lazy to take into account factors other than win/loss. That’s what a lack of imagination does to a company.
The more parameters there are, the more likely their spaghetti code is going to bug out.
And the devs would have to keep updating the “personal performanse” parameters by hand when ever the meta shifts. Because when the system was allowed to do so by itself, it easily boosted silver mercy players into diamond or further because of high healing numbers that were directly comparable to higher level mercy play.
There’s AI, there’s machine learning, and they can always manually tune. Silver mercies won’t get as much healing as higher level mercies because, and I want you to take note of performance metrics here, higher level mercies:
Context switch more between targets; increase overall healing efficiency
Use guardian angel more often to move around, thus
They avoid dying as often
Know when to prioritise damage boosting over healing
Do not commit to stupid resurrects that fail or cause staggers
Pull out the Barbie glock in the right circumstances
Yes it’ll be slightly complicated to track all this efficiently, but it’s possible and could evaluate player performance far better than win and loss alone. The metrics they can go off on are limited by their imagination
I’m pretty sure the issue would be to aggree on acceptable metrics for each rank separately.
How many assists, how much damage done/ amplified, how much healing, how many rezzes, how many elims, how many deaths.
No two matches are alike. As an extreme example from gold quickplay, both on push maps.
One match. 1 death, 6 elims, 3k healing done, 1 rez. Victory.
Another one. 6 deaths, 1 elim, 18k healing, 5 rezzes. Loss
One could argue my overall stats were lower in the first one, but arguably i did better since i won.
In the second one my heals and rezzes were good, but i lost.
That’s where the issues start coming up.
Good or bad stats? Which stat weights more when appraising the amount of progress based on them?
It gets very complicated very fast. And then you need to do the same to every hero for every rank.