Smurfs: The Good, the Bad & the Ugly - Q&A

General Discussion
72

API REPORTS METHODS

I’ll keep there a trace of the methods I’ve used to sample and detect smurfs in other leagues than my own.

V1.1

Research Q°: How many smurfs are there at gold MMR ?
= What’s the proportion of smurfs are encountered by gold players ?

  • Randomized MMR sample (within the MMR boundaries of studied tier) for every player).
  • The players from a same MMR are randomized once again, depending on the numbers of players, to determine which one will be screened, from the top to the bottom (for example if the random number is 5, the 5th player from the top will be selected).

Players with less than 8 1v1 in recent game history are discarded. If a player is discarded, then only the player is randomized once more, not the MMR.

For example for gold 1 EU :
→ Randomizing and MMR from 2720 to 2839 : 2727
→ if 10 players at 2727, randomizing from 1 to 10 : 2
→ second 2727 MMR player’s profile from the top analyzed
→ repeated for each player.

  • Profile analysis

    Criteria for non freelosing smurfs :

    Career games < 40 gold, < 80 plat, < 160 diam (MMR reached)
    More than 70,0% ratio with ≥ 20 games played
    ≥ 2 leagues compared to current ≥ 2 times with same race
    ≤ 2 leagues of opponents with ≥ 25 games played (≈ equivalent of provisional placement)

  • Games history analysis :

    Criteria for freelosing smurfs :

    ≥ 3 freeloses in recent history (24 games for web API, 48-80 games for ingame profiles).

  • FL if ≤ 150s from previous match for web API checks(estimated waiting (queuing + loading time) ≈60s, FL if left ≤ 1:30 + waiting time).
  • FL if left before opponent attack for ingame checks.

Players without freeloses nor profile inconsistencies => regular players
Players with minor profile inconsistencies => undefined

Discussion :

  • I had to define a minimal career amount for reaching gold league MMR, as I had done for platinum and diamond.
  • Slightly tweaked the values of those minimal career games, and of the maximum unsuspicious winrate, as those weren’t satisfactory. I partly based myself on a Reddit thread where players exchanged about after how many games they reached X league, and mad sure to stay below those while providing values that seemed likely to me for extremely gifted + coached players. Those are still comparable to the ones I used in my player-experience report in P1-D3.
  • Rephrased the definition about deliberately playing with a displayed league lesser than the opponent, since that kind of smurfs often makes in fact sure never to reach the 25 games ending the provisional MMR state.

For future versions, I’m considering randomize up to two players by MMR sample (instead of one), to speed things up while keeping the randomization procedure.

V1.1b

Same than 1.1, but for small details :

  • Users with < 8 1v1 played are only discarded if match history is needed to define their category.
    If a player is randomized twice the same day, the second randomization is also discarded.
  • Players whom usable data is too scarce to assess with acceptable confidence are categorized undefined (same than those with minor inconsistencies).

V1.1c

Same than 1.1b, but for some details, mainly defining procedures regarding extreme MMR zones (where there are less players than MMR points) :

  • If a player is randomized twice the same week, the second randomization is also discarded (=> better screening power).
  • If no suitable user can be found at the randomized MMR (all discarded, no players at that MMR), the MMR is randomized again.
  • For extreme leagues (≥ M1 and ≤ B3) if there are no users at the randomized MMR, the closest MMR where there are players is selected instead. If there are two equally close MMRs available, the choice between the two is picked at random. For example if there are no users at 1500 MMR, but there are at 1499 and 1501, the choice between the two will be random.

Websites & tools used :

73

SMURFS REPORT W30-32

GOLD LEAGUE

Alright folks ! So, contrary to the usual, this is a fully randomized report about smurfing within gold league. It may not seem like it, but it’s a tiny revolution compared to previously, as the results are now completely independent from me as a player, and also completely randomized (MMR sampled, players sampled) to ensure maximum reduction of selection bias.

Anyway, here are the results :

How to read : cf. first post (Part II).
Method used : cf. above post.

Quickie :
Over 105 users at gold MMR :

  • 23.81% of confirmed smurfs
    (11.43% of freelosers + 12.38% of non freelosing). :cowboy_hat_face:

  • 76.19% to 42.86% of regular players. :slight_smile:

Several remarks about this data :

  • There is a certain degree of uncertainty due to the small size of the sample. It could be appreciated by providing a confidence interval, but I haven’t learned yet how to calculate it and, hey, we’re just talking about an RTS game. However it’s quite close to what I’ve reported within my league, which suggest both reports provide acceptably accurate results.
    .

  • Though certainly theoretically less biased than my player-experience reports, this method has a number of shortcomings, as its wayyy less sensitive when it comes to freeloses detection : I’ve got only 24 games to work with, minus the last game played (can’t deduce duration), minus any custom/co-op/teamgame/arcade map. And if the games are played apart from each other, they will look like normal games into the API. Hopefully, if a player is in a clan, I can switch back to his ingame history which is much more detailed. And… well I’ve got an easier access to their league progression, which is in fact also a solid hint about non freelosing smurfs. That’s why this method probably underestimates the amount of freelosers, but compensates by better detection of non freelosing smurfs (which reflect the original use of smurfs accounts).
    .

  • If we sum-up, that’s about one fourth (25%) of smurfs at gold MMR, compared to roughly one fifth (20%) at platinum. That’s quite a lot, and as I expected, the proportion of smurfs seems to increase while the level lowers. The next reports will confirm or not.

PLATINUM 1

The usual report at P1-D3 MMR :

Quickie : 20,90% of smurfs (including 14.93% of freelosers), 79.10% of regular players, over a 67 users sample.

With this report, we’ve reached a 500 players sample, which I consider is enough, so I will stop my platinum player-experience reports there.

I’d like to thank all the smurfs who unwillingly participated to this study by trying to steamroll (a less and less) unsuspecting player ; to those who almost fell off their chair discovering that I had guessed right away ; and to those who allowed me to report them by stooping to lowly insults.

I’ve met a few smurfs that were actually pretty chill guys, and I’m sorry if I didn’t extend the conversation for overly long, but you guys should understand that what you’re doing is far from chill/good. It would be a pleasure to meet again the two or three of you, in another league, once you’ve stopped smurfing. :star2:

FUN FACTS

  • The girly pseudos being associated to smurf accounts seem to be a thing even in gold league. But in fact, it also applies to more generally lousy, dumb, or somewhat trolling names that are linked to smurfing, as some of them can’t resist the idea of starting to troll right away. It verified more than once through this gold report. If you see someone with a joke-like name (for example OmegaPoop, Rosie, or IreallyLoveU (I’m making those up, not quoting real users)), or a barcode, start considering the possibility that someone is trying to smurf you. I may make a best of of all the hints you can get in the future.

Once upon a time, there was a platinum player, that we will call Obiwan. He was matched against an unranked barcode, that we’ll refer as Anakin. He thought to himself « Damn, I’m 80% sure this is another one. I’m gonna cheese him, at least the game will be short ». Obiwan was wise, but he hadn’t foreseen at what point he would be true, until this happened :

We both had the same idea. Only, Anakin the Barcode committed to his plan despite being scouted right away, while the gentle platinum decided to cancel and build back rax and bunkers at home.

And though minute, the defender’s advantage of having the high ground was decisive, as though losing a lot of SCVs, it allowed the platinum player to snipe the enemy’s repeatedly sent SCVs, and then to finish his own bunker. And that is how the highground, once again, ended up being fatal to Anakin. :laughing:

That’s all folks, see you in three weaks for silver league ! :smiley:

1 Like
75

I would say that freelosing, is one sign that is a smurf, but there are more ways to find them. Today I played like 6-7 games all of them where smurfs, why I said that 1) low amount of games. This game is 10 year old, I would say that having less than 3k or 2k games in your account is a smurf
2)nicknames, I think this a great clue, as smurfs does not care about the account, they always put silly names.

I think if you follow those 2 behaviours you will find more smurfs than just checking for freelosing.

76

This is in essence stratified sampling, so you need to make sure that the sample sizes for each mmr range are reflective of percentage of the population in gold league, also your sample size is ridiculously small, you should have at least 20.

77

So at least I can put a name over what I’m doing. Thanks for the suggestion of using SC2 API btw, originally it came from an idea of yours in another thread. :+1:

My « sample » size is of 35 users per tier, which makes 105 users current total. There are roughly about 78 [56-101] players per MMR point, which makes around 28 000 players currently playing at gold MMR in EU. It seemed they were equally spread through the whole league though, there weren’t significantly more players per point in G3 than in G1.

So of course 105/28 000 is small, and would I make it 500, it’d still be small. The thing is it takes a ton of time to do (around 1h20 for 10 players), as the double randomization slows things down, and the amount of players barely playing 1v1 (thus being discarded) seems higher as the league lowers. And let’s not forget that, at the root, I’m still a player. So I also want to use part of my freetime to play. :wink:

Would those reports consistently attract attention from the community or from the devs though, I might be willing to increase progressively the total sample size over time, just like I did with my previous player-experience report.

Maybe regular players from platinum and leagues below do. I also imagine that if you’re a bronze/silver beginner and you get repeatedly crushed by seemingly much higher level users, it might discourage you from continue the game altogether.

Why should intermediate to high level users care about lower level players will you ask me ? I think the answer is the same than when any user made something beneficial for the community : some do create and maintain exceptional websites (such as rankedFTW), others do tutorials on youtube, and some threads on forums. :slightly_smiling_face:

Also, from what I’ve seen, most smurfs seem to just hate it when they suspect the other user is also smurfing. Some would leave right away, others will start writing walls of text while the very game hasn’t ended, and some even insult you. It’s funny how while being smurfs they don’t have any idea how to make sure. Anyway, if even smurfs hate getting smurfed, why shouldn’t regular players ? And what if you’re a gold 3 player, and that’s not one game but one over four or five ?

The thread is quite long so I understand if people don’t read it completely. Searching for freeloses is my primary criteria for finding smurfs, as it allows to find those who elect to smurf while remaining at one same MMR.

For the other types of smurfs, the one who don’t freelose but prefer to create new accounts and grind the leagues from the bottom to the top, I’ve defined (and even refined) other criteria :

So as you can see, I do take in account the total career games when tracking down smurfs. There are great differences about when players reach some leagues depending on their age, their spontaneous affinities with multitasking, their tendency to cheese and their being coached or not. So the cutoff I’ve set are quite conservative.

Yet it seems the earliest diamond reach in Reddit was 200 games career, so I’ve set for a 160 games cutoff for that league, with 80 for platinum, and 40 for gold league. Any account reaching those MMR with less career games than those is considered a smurf account. And there are. :cowboy_hat_face:

As for the silly names, it’s a thing indeed, but there could be regular users with funny names too, so though it raises suspicion, you can’t say for sure you’re facing a smurf basing yourself on the pseudo alone. :wink:

78

Although I’m currently not done surveying the silver league (only one tier done ! :cowboy_hat_face:), there was two testimonies interesting enough to be mentioned in this thread :

The first one is about the consequences of smurfing on unexperienced players :

This confirms what I supposed about the consequences of smurfing on less experience players : frustration, and for some, a deterrent to even play the game.
That’s why I think smurfing in those leagues slowly kills the game, and why it’d be important for the devs to realize it.

The second, on the contrary, relates to what leads to match-up smurfing :

The interesting part is that it’s a perfect example of what leads some users to become match-up smurfs : despite allegedly not complaining about balance anymore, freelosing entirely one match-up in fact relates to the intimate conviction those match-ups are imbalanced. And that nothing those individuals could do or say would change that : so after complaining, then cheesing, the next step is freelosing some match-ups entirely.

There’s a fix suggested, though, the possibility to veto some match-ups on the ladder :

  • I think that it wouldn’t be fair to have users playing only one or two match-up instead of three, thus discarding the less favorable ones, to be ranked in the same place than those who made the effort to overcome the more difficult ones.
  • However, it might be a good idea for unranked, since unranked doesn’t ranks players, is more meant as a training ground ; and since that may allow for match-up smurfs to stop freelosing.

Match-up smurfs are minority though ; but even for training purposes that suggestion could come in handy. :thinking:

I’ll also use the occasion to underline the first post has been updated and revisited. :slightly_smiling_face:

79

So he posted to say that him and a bunch of other high level players he knows have smurfed a bunch of times, and by doing so have confirmed smurfing is not a problem. Okay… Your well thought out and insightful post is no match for that logic.

1 Like
80

I think blizzard should solve the problem by changing your MMR proportionately to the length of the game. That way Freelosing wouldn’t affect your MMR at all, or at the least would be much more time consuming and difficult to do.

81

That’s an idea. The cons would be that if would affect cheesers and alliners as well. And in a tournament, even if you’ve lost to a 12 pool spine rush or a proxy 5 rax, well if you leave the game, the game is lost and cost you just as much as a 40’ macro game.

So I’m quite fond of queue time fixes : you freelose, you get a 5 minutes queue time penalty ; and that penalty doubles each time you freelose if you do so in a short time interval. That, that would be dissuasive, as it would increase the time required to stay at a wrong MMR exponentially ; and this without affecting regular players in the least.

But either way, the point is that it would be a good thing if the devs implemented freeloses detection. From there, anything can be discussed, but we haven’t made that step yet. :thinking:

I think you might be alluding to D3stryw4p0n or Cheezecake. Actually, a few months later I discussed with the second one, in a more appeased way, and he made a quite useful suggestion. So, despite not having studied himself the topic, in the end he was one of the guys who allowed me to transform my purely observational reports into real randomized surveys. So, however tensed that his word have been, it wouldn’t be fair to reproach him those, as ultimately they contributed to improve my work’s quality. :slightly_smiling_face:

Also, his observations refer to the first iterations of this thread’s first post. Since then, it has been reworked, documented and updated quite a lot of time. So that’s why some of his remarks may look less relevant today than several months ago.

Regarding the substance of the debate though, smurfing frequency ; it appears that smurfing does not make you suddenly more informed on the topic. A good quarter of smurfs do not even know what smurfing means for starters ; and as it’s most of the time concealed, can’t really be aware of the extent of the phenomenon.

So here’s where my small work can get some value: delimiting the extent of smurfing as precisely as possible. :mag:

Thanks for your appreciation, LightWarrior. :slightly_smiling_face:

82

I disagree with this completely. As a master player, it feels exactly the same fighting a gold player and a platinum player. They are both significantly worse than me and I can do anything I want to win. Not to mention the skill differences in these leagues vary so wildly its insane. I’ve seen legit gold players that I thought had plat level macro but were god awful at holding cheeses so they stayed in gold league. You simply cannot know for sure as a master player if who you are facing is actually in that league or not.

Its a bit easier to tell once you get to platinum as diamond players will be fairly competent at the game. So will stick out among the plat players still struggling to remember when and where to place their buildings. Still easy to beat but they are noticeably better than the plats.

1 Like
83

All this being said though I have no idea why anyone argues about this though.

I agree that the smurfing issue isn’t a big enough issue to completely derail someones climb up the ladder. But that should not be an excuse to dismiss the problem entirely. There shouldn’t be this issue to begin with.

There 100% needs to be a system in place to at the very least make it more difficult for players to smurf. Anyone who is against this is likely one of the smurfing trolls to begin with. So their opinion is jaded at best.

If you want a 2nd account for anonymity and its within 1 or 2 leagues of your real account thats fine, no real issues. I can at least see the merit in it. Maybe you want to practice off meta stuff or something.

But if you are the person who is auto losing games in order to troll low league players then you should be ashamed of yourself. Grow some balls and play at your own skill level.

2 Likes
84

That’s more or less what I thought. To take an example, it won’t make much of a difference to a black belt karate-ka if he faces a yellow belt or a green one. And a gold player will feel quite strained against opponents from just one league above.

However, I think that a master player would know better than the lowleagues about the progression in the leagues close to masters. But it have to focus on those to notice : and while it would indeed be easier for a master to do so upon casting a replay or a stream ; when playing himself, in the fog of war, it might be way harder to differenciate a gold 3 from a silver 2 since both won’t have enough micro/reactivity, and so will lose in the first minutes anyway.

So far it seems that those who opposed this indeed were smurfs themselves, regularly or not. :mag:

I don’t see how having a matchmaking doing a better job at match-making could be seen negatively from regular users. The only possible hinder of having a freelose-detection system would be to Bronze-to-Masters tutorial makers. But considering that a new account requires only 10 days to get (more vespene gas or additional pylons put aside ^^), and that those account could be led to bronze MMR in rather few games, they could get around this without too much trouble. :thinking:

Getting a second account is extremely easy, you just have to play on another server. As for having it one league or two below yours, as long as it’s not deliberate (for example people may really be 2 leagues under theirs in offrace situations), that’s not a issue. The issues come with the freeloses, that’s the heart of the problem.

1 Like
85

Watching the replay post game yes its easier to tell. But in the actual game itself it is near impossible for me to make an accurate determination of if this is a silver/gold, or a gold/plat player. Nothing these players can do will even remotely tax my abilities so they are all going to feel on average the same to play against.

A good player will naturally make someone in plat look like a gold leaguer. There is just nothing the plat player can do to me when i am significantly better at every aspect of the game. So they will feel like a gold player. But if the plat plays vs a gold those small differences in skill will come through allowing the plat to win the majority of his games while smurfing. Someone of higher league in the low metal leagues just does not have the same perspective that an actual player of that skill has.

Yeah thats what I ment. Like as long as you aren’t deliberately keeping it there I don’t see a problem with it. Like just for example. For me to keep an account below diamond 2 (hell even diamond 1) I would have to actively try. No matter what I do I can usually crush anyone below diamond 2, even if I am doing just ridiculous builds. So nobody should have an account more than 2 leagues below their real account in my opinion.

I could for sure keep an account in Diamond 1 legitimately, if I am doing builds out side of my norm however. Diamond 2 I feel like I would have to be doing some very weird stuff. But I’ll grant people the wiggle room and say 2 leagues below is fine even though I feel like 1 league below is probably the more realistic and fair compromise.

1 Like
86

So if you’re master 3, and that you feel having a diamond 1 or 2 account for occasional builds experiments ; you’re not meaning two leagues, but two tiers below your actual one. The precision is important !

Because, if on your main race, two leagues below (for example gold 1 if you’re diam 1) would seem unlikely if not deliberate.

For my part, in my current 1.1 procedure for my smurfs reports, I’ve considered that two leagues (4-6 tiers) or more below your past level, with your main race, can’t be anything but deliberate. For example if I spot a Terran in silver 2, while he has been several times platinum in the past, I do straight categorize him as a smurf. And I think current criteria are good, as it often happens that when someone is a smurf, he often meets several of those criteria simultaneously. :mag:

This isn’t true for offraces though. I’ve now screened a lot of profiles, and it seems that for Protoss or Zerg mains, Terran is a bit more awkward to handle. And so it’s not infrequent that Protoss or Zerg players have a terran offrace two leagues below their main. So as long as its offrace, I’ve not set any league floor limit. :slightly_smiling_face:

87

Yes sorry, my fault. I meant 2 tiers. 2 full leagues is absurd. XD I would never lose a game to a plat 3 player. I could lose to a D1 though, D2 maybe.

Yeah agreed. What I mention mostly goes for the players main race. Offracing is a bit different. The races are quite different from each other. So being a league or 2 below your main race isn’t to weird. Though I feel like you will eventually get within at least 2 or 3 tiers of your main race if you played more than 100 games.

I for example am M3 toss, D2 Zerg, D3 Terran. I’ve got close to 8k games toss, 400ish games zerg, and like 200-300ish games terran. I recently switched to playing all the races more often. My terran lags a bit but I mostly chalk it up to the lack of experience with them. I have no doubt with more games Ill be equal skill with zerg and terran. I doubt I’ll ever catch up to toss though unless I just stop playing toss.

As for this I feel like it would depend on the time gap between when they were plat to their now silver level. I could see someone taking a year break and coming back much much worse than they were. But they will likely regain the skill relatively quickly. When I was D3 I took about a 5 month break. When I came back I was low plat level for sure. But after like 2 weeks I was back to D3.

1 Like
88

At any rate though, keep up the work bro. Keep voicing the concerns about this. Blizz needs to do something about it, and screw everyone who thinks this isn’t an issue, they are likely part of the problem to begin with.

1 Like
89

I think blizzard should solve the problem by changing your MMR proportionately to the length of the game. That way Freelosing wouldn’t affect your MMR at all, or at the least would be much more time consuming and difficult to do.

The problem with this is that people would just instantly surrender vs players or match-ups they don’t feel favored in. Also, it opens the opportunity to purposefully draw out games you’ve already won to farm a greater amount of MMR. Most solutions offered to ‘solve’ Smurfing just create bigger issues. The only viable solution is to just keep playing and never assume/care that your opponent might be a smurf. Eventually you’re going to hit an opponent that is better than you.

90

There is an easy way to fix the problem. Increase the que after a game that lasts less than say 3 mins. Then keep increasing the que time the more games consecutive games after that, that last less than 3 mins. Example: after 1- 1 min que time, 2- 2 min que time, 3- 3 min que time, ect ect ect. You can apply any value to it you want, just an example.

This would not impact cheese games much because most cheeses arent over in less than 3 mins. For the ones that would be, it is a shame but I doubt they will fight it more than once in a row so its a necessary casualty.

It will force the smurfers to be more committed to the smurf. Either they gotta wait out the 3 mins every game they want to auto lose or wait out the ever increasing que times. Yes obviously this is not going to stop everyone. There will be those that suffer through this. But it will deter many of them I am sure.

Its either this or blizz needs to take reports more seriously. Actually investigate reported players and if they see a correlation give the accounts temp bans. But this will never happen so option 1 is the only way.

Blizz even has a similar system in place already that they could easily apply to que times. Currently if your match lasts less than 3 mins you get no xp from that match. So they already have a system in place that has the same basic principle. All they need to do is apply it to que times.

91

Cheese games can be decided in under 3 minutes. Losing is punishment enough for people who didn’t scout and didn’t notice a cannon rush for instance, I don’t think they deserve a longer queue for it. You will never convince me that there exists such a thing as a ‘necessary casualty’ relating to this conversation.

You guys are suggesting developer work that will both not solve the problem but also create additional problems.

1 Like
92

Even cannon rushes hardly ever are decided in 3 mins unless the person getting rushed just doesn’t want to deal with it at all and leaves the second they see a forge.

The only cheese I can think of that can actually end the game this quick is 12 pool in PvZ if the toss doesn’t wall. But even then if the toss try to hold it at all it will likely last longer than 3 mins. But even if you arent for the 3 min then make it 2 or 1. Something, literally anything would be better than nothing.