Riot announced yesterday that they're revamping their overall ranked ladder system for season 3. They spell out why they want to make a change and explain how the new system will work. They haven't given enough information into the nitty-gritty details to really work out what's going to happen but after thinking about the new system a bit today I fear there's going to be some pretty big problems with the way it's set up.
To start the new system feels on the surface to be like the league system in StarCraft II. You get put in a league with a precious metal name indicating about how good you are. Win a bunch and get promoted up. Lose a bunch and get demoted down. There's a little ladder in your individual tier which lets you compare yourself to other people similar in skill to you. On the surface this all makes sense, and especially it looks like it satisfies the primary stated goal for the change which is to give people short term goals that seem meaningful and attainable.
The difference with SC2 comes in the details. SC2 leagues are purposefully sized to include a precise percentage of the population. People get promoted/demoted as they cross the barriers between being in the bottom 20% of people to being in the 20-40% range. I haven't played SC2 in quite some time so I don't know if it's changed or not, but I think it also tried to match you with people in your own division when possible and certainly paired you within your own league the vast, vast majority of the time. LoL is going to work differently, and in ways that I don't think Riot considered.
In particular, Elo rating isn't going away and will continue to be the sole factor used in matchmaking. Someone with an abnormally low Elo for their league will never get to play against other people in their league. If leagues can be easily seen on the loading screen this is going to be a huge problem, I think. Having a gold league player always playing with silver league teammates is going to cause strife since the gold league player will think he's better than his teammates by virtue of being in a higher league when really he's around the same skill as them. This won't be a problem if promotions/demotions are properly handled, but it really sounds like they won't be.
From the sounds of things you'll earn 15 league points with a win, and lose the same with a loss. Fall too far below 0 and you get demoted. (Except you can't get demoted down a full league type, only down a division in that league. Get gold once and stay there forever!) Get up to 100 and you earn the right to play in a promotion series. Win 2 of your next 3 games and get promoted. Lose at least two of them and get dumped back in your league with 70 points to try again.
Here's the problem... It's a personal zero sum system (wins are worth the same as losses) so you'll only ever get a shot at promotion if you go +7 in games won over a period of time. Then when you get that chance some people will get promoted and others will tumble back down and need to win 2 more games to get the chance again. Note that you could be going -1 in the promotion matches and need to go +2 to try again, so some people will get promoted after going +8 and others could actually go +50 and stay in the same league.
Also note that while your Elo rating is below your equilibrium point you rate to win more games than you lose but once you reach that point you should be hitting a 50% win rate. As such, people who have deflated Elo ratings expect to get promoted quickly while those who are at equilibrium expect to stay stagnant pretty much forever unless they get appreciably better. Coupled with the above point someone who gets lucky enough to always get promoted on their first try will end up far above someone who gets unlucky and has to keep trying over and over again. They may both have started at the same Elo, have won the same percentage of games, have ended up at the same Elo, and end up far apart just based on the timings of their wins. Some 3 game windows are way more important than others in terms of getting promoted but not in terms of Elo which will cause skewed results.
Another problem occurs because the system is actually quite open to abuse. Let's pretend I really want to get into the Challenger league. Also, let's pretend I'm currently at the bottom tier of the gold league. I am going to intentionally lose 60 games in a row. Maybe I'll take this time to learn a bunch of new champions or to try stupid tactics. Play badly enough to lose but not go AFK or be an obvious troll. Or maybe I'll just let someone else who's worse than I am play my account for a while. This will end up massively deflating my Elo rating while keeping my league position the same. I will be at 0 points in the bottom gold division. Then I'll start playing for real. I won't win every game down in 'Elo Hell' but I will rate to win significantly more than I lose. In fact, I should pull off +60 wins vs losses by the time I get back to my real Elo. After the first 7 of those 60 wins I'll get a chance for promotion. Win 2 of those games (and since I'm 53 losses below my real rating this is pretty likely) and get up to the second gold division. Repeat 4 more times for all 4 gold divisions. I'm likely to only use up 45 of my 60 win buffer in the process and will get promoted to the bottom division of the platinum league. Where I can never be demoted again. My Elo rating is still lower than it should be, and lower than it was to earn access to the gold league, but I've made it to platinum.
Repeat. Turf another 60 losses (with the commensurate loss in Elo and no impact to my league standing) and start trying again. I will eventually make it to the challenger league. Heck, I could even repeat at that point! I could top the ladder with a pathetically low Elo just by properly timing when I lose.
But since your listed league doesn't mean anything except bragging rights, why bother with all those losses? Well, for one thing, bragging rights on the internet are a big deal to some people. For another, you get a shot at turning pro if you do well enough on the 5v5 ladder. And while I'm not good enough to be on a 5v5 team that has a chance of winning the tournament at the end of the ladder this path to abuse isn't just open to bad players like myself. Any team with a shot at doing well in that tournament would be a fool to play the 5v5 ladder game legitimately. They should absolutely all sandbag their team Elos upon hitting challenger league and then make a strong push against worse teams right before the qualification period ends. Because that gets them in the door to play in the big event where anything could happen.
And if you don't think pro teams wouldn't take sketchy lines to benefit themselves you haven't paid enough attention to the conduct of some of the current pro teams and the cheating they've done. A team recently got disbanded by their sponsor when it came out that they abused a bug in spectator mode to get a friend watching their tournament match live who then fed them ward placement information during the game.
Is there a way to fix the problems while keeping the core idea around? Well, you could matchmake by league primarily instead of by Elo. If I have to keep playing diamond players once I cheese my way into diamond league it won't matter that I went -60 to start there. I won't get the +60 off the backend because I won't be playing terrible people in the 900 rating bracket. You could give up on letting people stay in their current league and actually demote me back to silver once I lose 60 games in a row at gold. You could make those losses count against me, and have me sitting at -900 points in the gold league and have to earn 1000 points to get my first chance at promotion.
I also think games that matter more to the league system (ie: promotion matches) should make a bigger impact on your Elo rating. So if you lose 2 of your 3 placement matches you actually end up down 2 losses worth of Elo. This way someone with bad win timing doesn't end up in an artifically low league as a result. They get a couple 'easier' games to get back into their placement matches and catch back up to their doppelganger with better timing.