The Ladder Smurf Myth

thats what she said

This works under the assumption you do not talk about the original Blizzard B.net ladder system, which was always kind of shit, thanks to bot users, hackers and the abscence of the publisher for the majority of the history.


Any competitive ladder (server) always advertised itself as competitive environment. Competition implies that the participants will play to win, no matter what. This involves a lot of cheese (not an issue) and a player base aiming to improve. I can't come up with one example that said otherwise, e.g. "Join World Gaming Tour for some fun games in your spare time" or "Join the Pro Gaming Tour to find some games whenever you want to". The motto is more like "play with the best, die like the rest".

This being said, before the beta any given ladder system worked well enough to give eager players opponents, which were about right. There's always a margin of error, regardless of the system used. It really doesn't matter if it's Elo, Glicko or something like WGT/PGT schemes. All those are estimators and all of these estimators have their ups and downs. Just one out of many examples:

Any given estimator has a start value. Depending on the punishment terms you go up or down very quickly or not. In Elo you might lose ten games as beginner, before you find "good" opponents, which won't smash you. The start value however is shit for those, who'se actual skill is close to it - e.g. in Elo with a start value of 1000 points. You'll always get a ratio of at least 45%++ people who just started out, but who're actually higher than 1000 points. On ICCup, as D rank, you might run into players who start late. That's shit, but you can't compensate for that, it comes witht he system.
      Almost all of the more dynamic systems (Elo, Glicko) set their start points in a way that they hit beginners and casuals the hardest. It might, just might, benefit the established player class, but not really. If you have a closer look, you most likely will still need more than 30 games to get a good rating - meaning that you'll be placed where you belong. You simply need some observations for your estimator to work. That's statistic basic stuff. It won't change the fact that people with less interest to play some strangers on a casual basis have to endure this one time estimation progress.

Now you might argue that this can be circumvented with never-ending seasons, e.g. no stat reset. Yeah, no, it can't. Any ladder system will suffer from point inflation, this really is a thing for Elo and you can observe it with the current ICCup scheme towards the end of the season. It's another byproduct of the mathematics of the system. Glicko2 might compensate for that (similar to B.net 2.0's whatstheirface point pool) - but this is just another form of ladder reset / pushes you into a circle that estimates you yet again against unsound opponents. If you don't get what I mean: If you can't play for three months, your rating will lose its meaning and you have to win more points again afterwards.

Also, for ladders there's this thing I tried to explain: Beginners and casuals drop out sooner, thanks to the start value conundrum explained before. Since they will face a ton of unequal opponents they start to lose motivation sooner. We had a server (2008-2009) for beginners on bw.de with 3k registered accounts, from which about 600 tried to play more than 50 games. In the end only 50 people stuck around, the rest wasn't satisfied, mostly calling the server "too hard". This was despite the fact that any signed up player had a rating of less than 2500 D+ points on ICCup. The average player was a stable D. And it wasn't even a competitive ladder, more like a place to find other beginners, until you wanted to go on ICCup.



... so if you say that you miss that "we as community" never fixed the problems for casuals, my answer is: casuals you can never really satisfy. They're easier and automatically satisfied if the player base is good enough. Because then, and only then, they can go into public games without ratings. Nobody cares too much in these games.



And now in the case you think about B.net 2.0's system or that of any MOBA:

These only work thanks to a lot more players in the system. Imagine a SCII system of all the divisions with less than 4000 players per base. It would get meaningless. The first ones to experience the idiocy of the system would - as they always do - the top players. Only few opponents, always the same faces. The next to experience boredom would be the bottom in the copper league, or whatever is down there. Same phenomeneon. You need variance, and you simply never had enough of it for BW after 2010.