Why Matchmaking Sucks... and How It Could Be Fixed. 

You can obtain epic levels of shaving with this weeks sponsor Henson Shaving. 🪒 Just go to hensonshaving.com/extracredits to get 100 free blades with your purchase, just enter "extracredits" at checkout. Thanks for Watching!

It's so sad that because of Namco Bandai patent on mini games during loading screens so very few games implement it. Even though it got lifted in 2015 everybody got so used to it that we're still very rarely see it implemented.

This is something a number of fighting games already do. When waiting in matchmaking, you can be in training mode honing your skills. An extra minute or two practicing perfect parries could be the difference between a win and a loss.

Yup, loved it when they added mini games in Splatoon match making, sad when they removed it in Splatoon 2, and was excited to see the physical lobby in Splatoon 3

4:09 you ranked them ping, queue time, skill verbally, but the graphic showed ping, skill, queue time.

Most of the big fighting games that have come out in the last couple years allow "training mode matchmaking" where you sit in training mode while you wait for a match. You can even set some up to use the stage and opponent from your last match so you can try labbing things from the fight you just had. Its *great*.

What if there was a way to combine different types of matches? So while one side wins points from kills, the other wins points from terrain control. Or King of the Hill vs Capture the Flag, where one team wants to man a specific location while the other keeps running back and forth across the map. Could be interesting...

One idea for dealing with the loneliness of high skilled players is having an asymmetric mode where it’s 1 high level vs a few of low. The high has experience and map knowledge to make the fight balanced. Plus it would encourage players to earn the high level to play as the boss.

I think this is a very surface level analysis or at least an analysis of just one genre of games. And not only genre but also scale. Stand by queue is the absolute standard right now, if the games you play don't have that, it's either stuck in 2015 or the devs just don't know how to do it. And people are willing to wait quite a few minutes especially in smaller scale games. But for the big games, where queue times are already basically instant, people are also very willing to wait just that bit longer for a higher quality match And there are games too small to have a match making system and they have to rely on public lobbies and organized events And also Elo is not an anagram, it's the name of the dude who invented it

Huh! I never noticed that about elevators! Mind blown!! 🤯

Psychological perception of queuing is a well studied field with applications from coffee shops to theme parks. It has a lot of existing ways to make you feel happier while waiting. Distractions are one of them, but so is grounding your expectations (and often artificially beating it). I think providing an estimate is really useful for some games, especially if they have high variance in queue times. Knowing whether I can get up to take a quick break is great, but often I’ll worry a game might start while I’m out and maybe I’ll get AFK kicked. Solid estimates help respect the player’s time and let them find their own entertainment… maybe watching a RU-vid video that just fits under the wait time.

I miss when you could join Valve tf2 servers, from the server browser. Rather have a higher ping, then being stuck with bots swarming the local servers. _Goes to the nightly active community server._

There's another factor to consider. Not "what makes a good match" but "what makes a human BELIEVE that had a good match". For a very large number of players, a 'good match' is one where they fairly effortlessly dominated their opponent. They find 'close matches' stressful and any loss at all is a fault of the game and 'poor matchmaking' that 'makes them lose'. Now, while you have some segment that plays your competitive PvP mode to be, well, competitive in the true sense of the word - to get better, to test themselves against others who are a real challenge to them, and to 'be the best like no one ever was' - real competitive players are a minority next to faux competitive players. These players want to be told they're really good, and get all the dings, and crush enemies, but they mentally wilt instantly when things don't go well, when someone else wins. They SAY they want the hardcore PvP experience but what they WANT is to be the big fish in a small pond, to be the best player amidst a group that won't actually challenge them in a way that takes effort. There's big overlap here with ragelords who become toxic and disruptive when they don't get to be MVP every match, every time, but I'd actually like to drive towards a different issue: The people from the above paragraph DO have a point, even if they rarely know it to say themselves. Giving 100% effort every match, like you need to in a true competitive environment, is less "having fun at a game" and more "working hard at a job." Part of why matchmaking gets so much hate to it is that the goal of making every game as close a competitive match as possible, isn't necessarily the same as making the match "relaxing and fun". BUT, conversely, a match lopsided to where one player can relax and dominate the scoreboard without putting in their best effort... is a match where the losers are having a really bad time getting kicked around by someone they never should have been matched with. Streamers want to show off getting 10x killstreaks - you can't get a 10x killstreak if every duel is 50/50 with an equal opponent. Joe Gamer wants to grab a beer and unwind after work by winning some rounds of PvPFite - he can't unwind if he has to put his best effort forward, every game, every time. And yet if we DON'T match like every match is true-competitive, then while the streamer and Joe Gamer are having their fun, Bob Gamie is getting his face wrecked in an obviously unfair match. There's probably more factors than I can count that go into why this is so hard to do in online games compared to casual sports offline. Casual sports games, you're usually playing with friends or at least coworkers, where 'fun' is a shared concept. If my team is up five touchdowns on the other team in a game with friends, I won't feel great, I'll feel guilty. I'll WANT the game to be close because part of my fun is all my friends there having fun too. An online PvP match is a hoard of complete strangers, and its very hard to get the 'average person' to emotionally connect to an anonymous opponent they don't know and will never see again. Offline casual sports I can just quickly swap a player, "Hey Joe, go swap with Jim" and they trade jersies. Online obviously not. And if I was tired and exhausted and already grumpy, I would excuse myself from a competitive game with friends because I know I am not in the best spot to interactive competitively with people, yet, a lot of people don't emotionally engage with PvP games as if there are real people on the other side.

Why not add a visualisation of the matchmaking progress? So players have a idea _why_ it takes so long. Maybe add player preferences for what metric to drop first?

If an issue with matchmaking is that people who are very different from most players, like someone with a high level or maybe someone who has much higher ping than most players or something, then developers could make it so that the longer yoy wait in a queue, the faster the range of acceptable level or whatever else expands, so that if you're like level 1000 it's not taking 10 seconds for each level it expands or something like that, but for normal people things still take a regular amount of time

Honestly i really like TF2's complete lack of matchmaking by skill- it's a weird thing to say but like I feel as though the game actually works much better when theres a very random mix of skill levels, something about how it can sorta feel like everyone's off in their own little world (as long as its evenly-ish distributed between teams which is usually is cause 24 players is a lot)

4:05 ping, queue time, and then player skill huh? Weird way to spell them, but whatever.

Being able to join a game in progress seems like it would eliminate one of those huge splits in the population. In a sense, you make the game its own mini-game. This could require some gameplay changes to make work better. For instance, a match could support a variable player size, like by expanding or contracting the play area to allow adding and removing players to an existing match. Another gameplay possibility would be that players added during an existing match had an onboarding process where their ability to affect the match gradually grew over time to both prevent them from changing the match balance and let them get a feel for the match before being tossed in. This onboarding could be something like a ghost mode where players had to perform some task like activating a portal to walk through or to take over existing weak NPC bots before being able to interact with players in the world.

I personally really like it when this is solved by having a computer/AI player of a skill level similar to yours stand in for missing players when your elo gap reaches a certain threshold, so that you can always have a somewhat fair match. Nothing is worse to me than being stuck with only very unfair matchups just because only very different leveled players are available

One thing I saw a bit of correlation is how this works in larps especially the more sport like boffer larps. You are drawing on a relatively small pool of people with widely differentiating skill levels. Now when you have a small group like I do you can separate and everyone has to play everyone. But it feels less unfair playing with a mix of skills than it does in a game. I have a few theories as to why and how to implement them into games. Balance Teams So when we set up teams we try to balance them as evenly as possible. We balance the classes first and then skill levels. This can be achieved by searching for players by classes and pooling them together and then averaging the two teams player scores. This can still cause imbalance team like one really good player stuck with a team full of newbies but sometimes in larp we have to do that. Prestige When your playing a larp you really don't get any credibility stomping new character unless you take out a bunch at once. Usually its more strategic to take on people more of your skill or team up against higher level players. Now in Larps you are playing the same people over and over so you know who is good and who is not. So if you have teams with with skill levels you should have an indicator of how high or low they are so you'll know whether to engage or retreat. This could be done with colors like people who are lower than you are green, people around the same level are yellow, and higher ones are red. You get more points taking out red players but not much of anything for green. This will make it harder for better players and now beginner players to be able to fly under the radar because the others going after your better teammates. Teaching Now this I'm not sure exactly how to implement this into a game but in Larps one of the ways people get betters is from the higher level players teaching the lower level ones. Now I've done this in the middle of matches but outside of combat is where I can show them stuff discuss on what they did well and where to improve. This builds community and friendship and though not everyone does it I think its vital to a sustainable game. The only thing I can think of on how to implement this in games is have a training lobby where players can meet and work together on tactics and drills. I hope that's some ideas but I'm sure other larpers or people who play similar games could add to this.

I’ve recently been playing this VR game named RUMBLE. It has no player base, I’m talking 25 players in the USA at 3:00 PM, but what makes it bearable is that when match making, you stay in your main lobby so you can train as much as you want, you can choose where you want your target area to be in matchmaking, and that you can choose what rank you want to duel against. The community is also good enough that most players with high ranks will only use abilities that you, with a lower rank, have unlocked. All these paired together gives the player almost complete control over the balance of wait time verses higher quality duels.

OK thought: idle style game in the lobby while matchmaking. So while you're in a match your idle game is accruing widgets in the background for you to spend when you get back to it

What I often see is also that they use "1v1" rating algorithms to calculate player ranks in multiplayer games. As if chess and Call of Duty were comparable in their stucture -.-

4:08 in player skill vs que prioritere, there is a mismatch between what mat Said, and what the screen showet!

I'm a grown ass adult. I have a job, I have bills, I have a house that constantly needs something doing. I get, a best, a couple hours a week to play games. I shouldn't have to play against people who game more in one day than I do in a month. I would love to take up rainbow 6 siege or tarkov, but I won't because I know I'll just get decimated by people who have all the time in world. I gave up on Warzone because I was sick of running into people who had the latest "meta" weapons they'd spent the who day grinding.

so the only multiplayer game with "matches" that i regularly play nowadays is Starcraft 2 (and maybe Stormgate later). in SC2, matches are found incredibly fast so long as you aren't in Master's or higher. while there can indeed be people who smurf or other issues with ELO on its own, most of the time, i am in fact placed with people that are around similar skill rates (or at least win rates) to me. benefit of 1v1 games, i guess. it's something that might be worth considering, how different sizes of matches can promote faster gameplay. i'd have to imagine 3v3 gets found a lot faster than 5v5 after all.

I know what I'm about to suggest is more to implement. What if we let players have some control over these factors. Instead of the computer deciding, we let the player say "I'm ok fighting people with a much higher elo." Or "I'm ok with the ping going 20ms higher". If the game thinks "that won't make a big difference to reducing your wait time." then it can tell the player, allowing them to choose to wait much longer, or if they're really keen to just play, to then further increasing parameters. Or perhaps letting the player know from the get go "hey, there just aren't many people playing in your area at this time. If you want a game fast, up your acceptable ping to play people in different countries." I think this could increase player patience (though not always), as they're choosing what they're ok with, and therefore, might feel ownership of the consequences for waiting longer.

I think what would be great is to have machine learning get to the point where bots could be good enough that they could play like a player at your skill range. That way, if the queue is taking too long, bots are put in instead to fill the ranks, but they are adapted to the range that is set by the developer so you don't have super easy or super hard bots that ruin the experience. I personally think that's the biggest killer of matchmaking/online games is that you have to have a big enough community to keep the game going. Eventually people move on or get bored and then the games tank and they get shut down. Also having tools made as a game ends their support cycle so people can make their own servers would be top-notch. Let the game survive for as long as people play it, that's what I say.

A note, Elo isn't pronounced letter-by-letter. It's named after its creator Arpad Elo, and pronounced "ee-low".

I don't know exactly how much this ties into the main discussion versus being a tangent, but something I'd like to see is more games where the emphasis is on something other than winning a given match. I wonder if someone could tackle the ELO/balance issue from the other direction, de-emphasizing victory or defeat as a driving purpose of a match and in doing so making it less important that players are as evenly skill matched as possible. It might also make some of the other concerns a bit less impactful. I don't expect it would work for all sorts of games of course, but I think there are places it would be interesting to try.

This is why I'm fine being a single-player chad. No matchmaking or waiting on single-player games, my friends. No cheaters, no bot abusers, no mod menu users, and none of this crap where you need to wait to play a game.

My favorite game Guns of Icarus had very long lobby wait times, that I am certain I spent more time in the lobby pregame screen than in the actual game. The play experience was worth it and my time in lobby was spent chatting with people about random stuff.

It’s funny because this is actually what gave me the time to really dig into fighting games. Because they let you sit in the lab. And I mean if you’re in there anyway it makes menu bloat and options less scary because - What else do you have to focus on lmao. It’s actually a huge boon for someone like me who has adhd and some pretty severe executive dysfunction. Because when we run into stuff that’s intimidating or difficult - we’re a lot more likely to go. “Nope too complicated” and tune out.

Lol Walpole has become the where's Waldo of extra credit. Spotted you this episode!

I thank this channel for explaining something that has become way too complicated in a way that is clear and concise despite that the topic itself is not

Although I have nothing against a minigame, I particularly love when, during matchmaking or loading screen, you can play the relevant parts of the gameplay loop of the game. Like walking and running in Assassin's Creed, doing combos in Bayonetta, moving around in Fortnite... Players, specially newer ones, can better get used to the controls and how the game works without the stress of the competition or time frames and whatnot

TF2 used to have a good matchmaking system... it was called manually selecting a server. No waiting time and you could see the map, amount of people, and the region the server was on. Now we have to wait quite some time just to enter a half full server that may or may not have aimbots on it...

Love these midweek videos! They alre always great! God bless You 😊😊😊😊❤❤❤❤❤

As a potential idea: what if there was an endless minigame where you tried to get as high of a score as you could before your match was found. So then even if your match takes a really long time, at least you get a new high score on the minigame

4:25 that should be easily fixed by giving the player a choice to wait longer so he can get into a better match... Using the elevator example it would be the choice between taking the stairs and taking the elevator, well that might be not the correct comparison, since something like this would be more close to letting players manually choose which lobby they want to join if they decide that using the auto-matching system is taking too long...

Why can't they just do a layered queue where people are grouped into multiple qualities of matches with the same subset of people that expand and branch off within a fixed lobby time, and then accept the best matches within that grouping at the end of the fixed time?

One important factor in matchmaking that you left out is parties. They introduce a whole new set of problems along with complicating the existing ones on top of being very common. Especially in relation to matching players in relation to their skill for its own sleuth of reasons, most predominantly is that if a really good player and really bad player queue together one (most likely both) of them will be having an unfair match and there is nothing you can do about it.

Learned the mirror story in college over 30 years ago, it's amazing how many software performance issues are solved using that basic concept.

The elevator intermezzo was fun :D

I live in western australia. Whenever I hear about game matchmaking thats using skill based matchmaking, separating platforms, etc, I shake my head. Ping. Ping is the only thing they should be using. Specifically, the regions game devs use are woefully inadequate. I used to think my ping issues were simply poor download speed. I had dial up tier internet in the 2010's after all. But no, it wasn't download speed. Geography and developer netcode means the best case scenario for me is like, 50-70 ping. I'd consider that exceptional, many other players would consider this borderline unplayable. I understand western australia is a relatively low population area, but I don't believe for a second that there aren't enough people playing games with large playercounts to scrape together a match of only WA players. What I've started to suspect is that the game isn't even trying to do that. It's slapping an arbitrary region on things, and that region is likely lumping all of Oceania and south-east asia into one region and then moving on. I watch clips of other players gameplay and the things they can do because they don't have my ping is so disheartening. Its like they're playing a completely different game. The fact that I *could* have that experience but don't get to because game devs don't consider ping a real priority is infuriating. Or rather, they don't realise their region based solution to the ping problem does not actually solve it for the people impacted the most by it. Dedicated servers don't even help, because the dedicated servers are always in Sydney. Theres a miniscule chance I'll get an all WA match in a game without dedicated servers, but with, its the same awful ping every time. If you've not experienced what this constant poor ping is like, its hard to understand just how crippling it is. The way games compensate for lag is, in simplified terms, if you see yourself hit your target, the hit goes through. This means for WA players, it is impossible to dodge anything. You cannot peek around a corner. You cannot use melee without being hit back. Enemy AOE weapons have larger AOE's than yours. Enemy range is increased across the board. Enemy snipers are impossible to counter, as you cannot flinch them off target before they shoot you. It fundamentally worsens every single aspect of the multiplayer shooter experience. I dread to think what a fighting game would be like. The only ones I've tried were so awful it went from figuratively unplayable to literally. I don't want to call it lazyness, its more like inertia and lack of awareness of the problem in the first place. Most game devs are from America, Europe, and Asia, all regions that don't deal with this issue. Developers implemented the current region based system, saw that worked pretty much fine for most players, and don't really think about it any further. All the while the people who are most affected by ping are also basically missed by the ping compensation method being used. Nobody is going to try and reinvent the wheel on the basics of matchmaking, so its never been given a second look, and the lack of game devs in places like western australia means most devs don't even know the issue exists. And it is an issue. Its very telling that the red section of your graphic for worsening ping quality, the worst case scenario, is literally the best ping I will ever see in a game. I'd gladly take longer queue times if it meant good ping, but as you said, most players aren't willing to wait that long. Though I think a factor that needs to be considered is matchmaking errors. When matchmaking takes longer than a few minutes, its usually because the matchmaking has bugged out under the hood and is, presumably, spinning its wheels doing nothing. So what players learn is that if matchmaking takes more than a minute, they should quit out and start matchmaking again. Matchmaking that has visualisations of the match thats being found can help with this. If I can see that the game has in fact found 7 out of 8 players, I'm more likely to stick it out. Though that also raises the question of whether the matchmaking is even being conducted in a way that such a visualisation makes sense. Because it probably shouldn't be. If its found 7/8 players and a group of three people joins the queue who would be a perfect fit for the match, it can't bring those three in. Its a complex topic but sadly its not going to get any better until some tech development completely replaced the infrastructure our internet is based on. Simply because the developers don't know their current solution is leaving certain populations behind. I'm just frustrated that something as banal as geography has left me dealing with a 90's problem while everyone else is talking about things like skill based matchmaking or matching based on roles, acting like they've actually fixed the ping problem when they haven't.

I like the idea of giving the player some interesting activity to do during wait times. I would like to add that the minigames or activities players get should be based on the wait time. A game of pong is good for killing a few minutes, but a practice skirmish is better for anything longer.

I remember the original Splatoon had a mini-game while waiting for matchmaking which was surprisingly fun. Kinda surprised such features aren’t standard by now.

Also I hate it when matchmaking screws me over for playing at certain times, where you have to play in the morning for better matchmaking and if you play during noon , afternoon and nighttime the game makes you wanna break your controller out of frustration

This video doesn't touch on Variance, an issue with customizable games like World of Warcraft, Hearthstone, or Magic: the Gathering. Then add in a draw-smoother to make for fewer non-games and you get even less variance, as the draws of those decks become more consistent. To me, the failure of matchmaking in this one aspect is the single biggest thing that keeps me from playing these platforms, even more than the game mechanics and microtransactions. While many of the platforms take in account player skill and/or ladder rank, they have a significant component of "will this match be challenging or engaging". The idea is to challenge and/or frustrate the player to grind more games. Little attention is paid to how repetitious those matches are or how frequently a low-competitive deck gets matched with a high-performing, metagame all-star. Since it's rarely possible to opt out of this queue, it's entirely too easy to get pidgenholed into the same 3-4 matches time and again to the point where you want to throw your electronic device out the window. Changing what you're playing only works so well in this context, unfortunately.

I usually just skip over the ads, but I wanted to support the channel so I gave it a look, and the Henson razor looks exactly what I need, as I just bought a Gillette razor which turned out to be very crappy

You said queue time is second @4:08 but in the animation it's the player skill

I play Crossout, a game from Gajin, a Russian game company. In this game they added some solutions to this problem: - Adding bots to complete the count of 8 players per team. Of course bots are not as good as human but that's better than wasting more time waiting for humans. - Allowing you to use most functionalities of the games (Market, Crafting, Settings, Store...) while you are waiting. The only functionality that is disabed is the building because you should not be able to change anything to your machine while looking for a battle, otherwise the match making will break.

Whats the point if you get quick matches if they are not fun because your skill rating is so VASTLY diferent? This is the main reason why i quit playing PvP games that dont use a somewhat strict ELO matchmaking. Sure i got into the game in under 60sec, but i went 0-40. I didnt have fun, i would have rather waited 3 min and went 20-20.

This video just reminds me of Wayfinders recent release debacle. The actual player count apparently massively exceeded their estimates, this crashed the servers. Many hours later they put a rough patch in place to limit player count to 10k, no word on if that was per platform or total. People waited so long in queues with the same problems games had in the early 2000's where you got booted from the queue due to a minor hiccup in Internet and now they find themselves back at the end again. This lasted several days, many people that decided to try to wait it out ended up frustrated and gave up. After they gave up it turns out they had exceeded the Steam rules for refund and most couldn't do so. This got so bad that the developers released a quide for players to follow to refund. One of the major things that made communication near impossible was the fact that the devs were only releasing info to their Discord, which wasn't openly shown as an option.

Could be a very nice solution for Hunger Games/Battle Royale type games. Make set of "Free for all" or similar servers (squad based perhaps). You can have a very action packed game mode available instantly, with the tense, slow, but extremely satisfying game available after a few minutes. Also, it would be interesting to have a Battlefield or Arma type game, but with each player only spawning once at the start of the match. Perhaps with the option of having "reinforcements" arriving at a later point.

Possible Great Fix(es): Lower total game mode count, Have strong proxy servers to connect to main servers with good ping rates, limit to your game either being only on PC or only on console (if console, combine different ones so that Xbox and Playstation are in the same ques due to very similar controls)

Cooking in the lab while I wait for a street fighter 6 match always makes the wait feel good

Playing For Honor with my friend as duo. We get matched against full pre-made team of 4 and usually our two randoms are way lower in skill level, not always but I can count with one hand times where one of the randoms did more than feed. This is a big issue I've seen before, shocked this wasn't mentioned

A huge part of why I love Battlebit is that it straight up has a server browser. I know that's not really feasible in a lot of games but if you can add it you really really should

Splitgate's really good at this. The lobby is basically a fiesta FFA game on a small map.

Multiple game modes seems like the easiest complication to cut. Presumably the player counts per mode would tend more towards a power law than even distribution, so the matchmaking in the least popular modes would be consistently atrocious anyway. Force everyone into the same bucket to improve the experience across the board, and no one will miss a game mode they never saw.

Bug report at 407-4:10 - Audio does not match video / chart.

Was the intro a Jackson Galaxy cat dictionary parody?

Hey, maybe an alternate solution to fixing the problem is making the quality of the game take precedent, then ping, then the matchmaking. Just throwing that out there.

Okay but why do we keep getting kicked out of the queue and thrown to error screens?

I feel like we should let players decide how they prefer matchmaking as settings. Give them 50 matchmaking points, and they start perfectly balanced weight on queue time, skill match, and ping. Players can decrease a weight by 2 to gain 1 more matchmaking point. All weights start at 100, and can go to 0 or to 200. So if you want the absolute fairest games, you can have a weight of 25 on queue time which may mean you'll wait up to an hour for a match. And ping is lower such that they maybe playing with up to 500 ping. But they know that all players in the match are "very" close in skill to them. Essentially just slows how quickly that expanding skill window for matchmaking will expand.

Thanks again

I don't know what to tell you. I'm still traumatized from my league days, where every single game was 5 pros vs 5 noobs and there was no such thing as a hope for winning when it was clear i got the bad team. The game was just unfun because i had 0 impact, and would either get carried or turbo stomped. 0/5 stars would never recomend that game to anyone

4:09 is your graph wrong with your text? I thnk you flipped "Queue time" and "play skill"

I'm kinda surprised people aren't willing to wait, when I Q up for a match in something I just play a youtube video on my phone while I wait, so I'm not bored or anything and can wait a while without caring

I have generally found I actually prefer swiss tournament style match making to elo ladders. Its less stressful, since not every game is going to be close. And it keeps those close games in context (whether I win or lose against this opponent seems like a smaller deal when both lost to the best already). This type of match making will require more people with good ping to match up, but skill matching is basically not necissary.

As far as games with multiple modes go, would it be feasible to include a "join any game" option for casual players? It's an idea I got from conveyoy belt sushi restaurants in Japan. They often have electronic wait lists at the entrance that let you specify booth or counter seating, but selecting "no preference" gets you whichever seat opens up first. In a matchmaking context, maybe players could rank their preferred game types so the game knows what to do in case it matches up multiple players who chose the "any game" option.

Casual Valve servers as they used to exist are dead. Used to be you just open the server browser, think to yourself 'I wanna play on Gorge', and then find yourself a gorge server. You go there and everyone plays on the map they want for multiple games or until the players themselves elect to force an early move. Nowadays I queue up, start going into a map I want, halfway through the map switches to loading another one I don't want to play because the previous match finished. I hate this new map, I requeue. I actually get into a server, harvest this time. It's not gorge but it'll do. But once the game is loaded the round is already over. Everyone is voting for a crappy map like Brazil or something. I decide to stick around and suffer through it. The new map loads, one team is half empty because of the typical mass-departure come the end of a valve casual match. The team with twice the players quickly wins the first round as new players cycle into the losing team, see how bad it is, and leave. By the time the server is actually full, one round is done, and if the same team finishes, we begin all over again. Feels like you spend more time loading up matches than playing them. moral of the story: live in a region with plenty of community servers or uninstall tf2.

The thing that bothers me most if it matches all the really good players in one team, how is that so difficult to make the Teams balanced?

So basically what Namco did with Ridge Racer back in the PS1 days to "alleviate" load screens...Minigame of Galaxian...will eventually be applied to matchmaking screens. That's a good idea, but I wonder if Bandai/Namco will try to block that based on their "patent" on that mechanic, considering a matchmaking queue could be seen as another type of loading screen. Something to ask your resident pro game designer and possibly resident law expert.

I think even with games that give us something to do while we wait like overwatch did with the mess around lobby and practice range we get bored of these activities and people just end up turning it off or sitting afk in them anyways. It was fun for about a month but after i never bothered again. I may be in the minority (idkfs) but i dont think well even do better with them.

Wait so all we need is some sort of crazy fast internet tech that allows us to play without really caring about ping...

Halo 2 did it right. A really revolutionary player scoring system that even modern MOBA games could learn from.

PANR has tuned in.

I 100% agree with you that matchmaking sucks

This explains Overwatch matchmaking a lot...

How about letting the player choose the mix of priorities for matchmaking? If they're more patient, let them reduce the importance of wait time and increase the importance of a similar skill, for instance. Or, hell, give the option to skip it entirely and go with the classic approach of just picking a server from a list or typing in an IP address. Honestly, unless your game is something like an MMO, where you have a persistent world that needs constant developer intervention just to keep it running, you should really just be leaving the running of servers to the players. It's just better for the long-term survival of your game. If your company goes bust and every player who bought the game also got a copy of the server software for it (eg Team Fortress 2), then they can keep playing until the heat death of the universe. If the game was depending on you to run servers and you go bust, the game dies with you and you leave no legacy.

Based on the closing statement am wondering if it will be possible to invent a game where you just enter and don't have a queue at all. Or maybe make the game time so long that you will go in and out of that game. Like you are part of team red and you conquer vast areas of land that take hours to do but you are always part of team red so you still feel engaged when in game. Maybe some games already have something like this?

Yeah this completely skipped over the single most important metric to the people that pay the developers, game revenue, usually expressed as microtransactions. PvP matchmaking is centered around encouraging players to buy / grind whatever OP meta the developers recently released.

I wonder if anyone has tried, asymmetrical matchmaking. Regular versus, but 5x10lv players versus 2x15 + 2x10. Rough balancing. Or, perhaps giving low enough players some interesting buff to level the playing field.

"insert vietnam war PTSD of tf2 casual and mostly competitive matchmaking"

I'm so tired of the match making in Pokemon Unite. They have definitely prioritised queue times of player skill because nowadays you get matched within 3 seconds from queueing up for a ranked match and most of the time, the player skill level is widely different. I just wish I could around 10 seconds instead so that I don't have to feel frustrated that I have to deal with bad team mates for 10 minutes. I guess the player base just want a quick, dumb casual game :/

Your forgetting the money aspect. I ride elevators about 14 times a day and yet to have been on one with a mirror.

Wasnt there a reason they couldn't do mini-games during loading screens or something? Some patent by some game company so monopolizing it and not letting anyone else do that? Quick Googling shows that Namco is the culrpit "Despite the Invade-a-Load prior art, Namco filed a patent in 1995 that prevented other companies from having playable mini-games on their loading screens, which expired in 2015." -Wikipedia

Kind of besides the point but Elo isn't an initialism, it's the name of the system's creator Arpad Elo

Mini-games in queue are a nice touch but it doesn't really solve any of the issues with bad matchmaking. Overwatch 2 for example lets you do a ton of different things while in queue and that game's matchmatching is absolutely awful. At the end of the day match making is a very complicated problem with a ton of variables, too many to account for with a traditional algorithm which makes it a prime candidate for AI. Game developers are assuming what balance of ping, queue times, and player skill are best at a global level but the AI doesn't make assumptions, it learns which is best through training. The AI can run through millions of games and it's ability to to dynamically create matches based on thousands of variables will only improve over time. There may be certain situations where high ping is acceptable and this can lead to higher quality games. They can be instances where a specific player is willing to wait longer and thus the AI could get them a better game. You can't get that level of granularity with a traditional algorithm because it would take decades to program. With an AI you can have an extremely dynamic system that makes changes based on what was effective in the past, it learns. In addition traditional algorithms are very bad at assessing skill. This is down to the fact that traditional algorithms have to program each potential outcome or situation in specifically. On the other hand AI can be trained to handle a nearly infinite combination of scenarios. By extension AI can provide a far quicker and more accurate assessment of skill, meaning that players have to go through less placement games and where the skill matchup is more closely aligned with the player's actual skill level.

Did anyone else think the Sponsors Name was "Handsome Shaving"?

I think id rather have a shorter wait for players. But sometimes i end up in a match with fellow level 10ish plateres and one try hard level 100+ who sweeps everyone (or even worse, im the one L10 in a loby full of L100+ and im doomed!) Not being a full time gamer meens i want quick casual matches rather than wait for uo to an hour to play in a hyper competitive match that has no scope for making a single strategic mistake, or it leaves you un a constant death loop

4:05 Those second and third items got swapped on either the art or the script.

Looking sharp, kiddo!

0:16 What is this game?

hear me out: no matchmaking. just select the game mode and put people on the highest populated server with a ping threshold and make game servers big again, 16-32 players per server. matchmaking will never be perfect, you'll always have smurfs or someone on a losing or winning streak, or a unevenly ranked friend group that will make the matchmaking feel unbalanced. so why promise something that only infuriates players when the promise is not fulfilled? when playing TF2, counterstrike, or battlefield back in the day there were plenty of people who were better than me, but that was something i looked up to, i wanted to be that good, nowadays when i come across a better player it just tilts me and i stop playing the game.

It ain't just video games. Board Game Arena has some flaws too, at least for their arena mode.

Why don't developers give users an option to choose their priority metric for matchmaking (ping, skill, queue ) instead of trying to solve it for everyone in the same way? Maybe that could also be used as an input for algorithm optimisation and set as default settings for new players. Full disclosure, I don't play online games so i might be talking out of my ass :)

Final Fantasy 14 has both the best and the worst match making possible. In the main parts of the game, the bits of content that you need to do, or want to do, you can get into it, do it, and get out of it rather rapidly, usually with a group of people that includes at least one person who knows how it's done.... when it breaks down, you get a group of 8 people against a superboss who all have no clue how to do it, try for half an hour, then give up and que again, only to end up back in the same situation ... for days. And on the other hand, you have the PVP matchmaking, and that's ... both better and worse as well. Good players get matched with people of nearly the same skill, or lower, at most times, and are challenged but never curb stomped all the way up the ladder until they hit a point where the skill they have cannot raise them any higher. ... on the other hand, casual people are matched with people well above the skill level, and simply never rise... it aint fun for them at all.