Ben Brode was on Reddit today answering some questions about the ladder and how they might be changed in the future. This was in reference to how there’s currently a bug that gives win streaks at two wins instead of three.
I’ve quoted and linked the comments below.
Seeing some comments here about how people are enjoying easier laddering due to this bug, and hoping we leave it unfixed. I thought I might chime in and talk about the ladder a bit, and hopefully get some feedback!
We have been discussing the ladder system a lot recently – we’re not 100% happy with it.
Here are some things we are currently discussing:
- Rank 18 players are higher ranked than 50% of HS players. That number doesn’t make you feel like you are in the top 50%, and that’s a missed opportunity. We try and counter this by telling you all over the place what the mapping is to the rest of the population, but it’d be better if expectations and reality matched here.
- We’ve received feedback that the last-minute jostling for high Legend ranks at the end of a season doesn’t feel all that great.
- We’ve received feedback that the ladder can feel like a grind.
- We are reanalyzing the number of ranks, the number of stars per rank, the number of bonus stars given out at the start of the season, and other parts of the system.
- We are developing simulation systems that let us predict what changes to the ladder would do to the population curve. If we inflate too many stars, the whole population ends up in the Legend bucket and while that might feel great for a single month, the entire system falls apart eventually. People who played waaaay back may remember when “3-star master” was the pinnacle of achievement, and it meant nothing because so many people ended up in that bucket. With better simulation tools, we are planning on trying a lot of crazy things. Iteration is important in design, and getting the tools to iterate quickly is very important.
Something I want to emphasize is that while I think we can improve the ladder, the metric for that improvement isn’t necessarily any one player’s individual rank increasing. Players want the better rewards (and prestige) associated with high ranks, or the Legend card back, so any change we make that increases the chances of those are likely to be perceived as “good”, at least for the short term. But part of what makes the ranked ladder compelling is that exists to rank players. If you want to see how you stack up, ranked is the place to do it. So while some inflation might improve the experience, we need to be careful and make sure we end up with a system that makes people feel rewarded for increases in personal skill or for finding a new deck that breaks the meta.
Someone asked whether or not they’ve considered reducing the fall in rank at the end of the season.
Yes, that’s what I meant by this: We are reanalyzing […] the number of bonus stars given out at the start of the season.
We think it’s a reasonable direction to explore in, but in general it’s hard to predict the results on the population distribution among the ranks. Hoping to do some simulation there and make sure there aren’t other, better options, too.
Someone asked whether it would be that big of a deal to have more people at legend.
Yes, I think so. Imagine just one bucket and we sort by MMR. There aren’t breakpoints with rewards as you increase in skill. There isn’t an obvious way to communicate with others about how good you are. It’s also difficult to know if you are bad or good. (Is Legend 135003 good? What about Legend 27809?)
The way we communicate our skill or progress is important. Ever heard a friend say “dude! I got to rank 5 this month!”? What is that achievement in a world where everyone is in the same bucket? Watching discreet rank buckets go by (and feeling that progression) and feeling the thrill of reaching a new rank that you’ve never hit before… those are pretty important, I think.
There was a question asked about whether or not they’ve considered revamping Casual ladder.
We’ve implemented MANY improvements to casual matchmaking over the last 2 years and new players have a much higher winrate there now. New players do so much better in casual (over ranked) that we have been considering locking ranked the same way we do Arena for new players.
A question about why win streaks end at Rank 5.
The original thinking there was that hitting legend should be through beating the other best players competing for Legend, not getting a lucky win-streak. This is another thing we have been discussing, though.
Someone asked if they’ve considered an open beta of sorts that would allow them to test things in a more realistic environment.
That wouldn’t let us iterate fast enough as we need to update it, get thousands of players to play there, and then wait for data to come in slowly over the course of a month.