O2 official use nearly the same algorithm as current star difficulty (density based).Xcrypt wrote:
Interesting. I know that it really depends on what the player spends his time on training, but we also use single diff levels for o2 and BMS right? Same problem there, but I do think the general difficulty is reflected by the level. Also for a ranking system, the difficulty shouldn't be relative to the player, but relative to all players (or one could say objective).xxbidiao wrote:
In one word, to accurately calculate the difficulty of the map using only the patterns with one universal numbers is not fair at all, and to sum up multiple numbers doesn't make sense.
TBH our temporary solution include human work (a lot of them), and use player's past performance data other than patterns itself to determine the difficulty.
And BMS doesn't have an universal difficulty measurement at all!
And yeah, what exactly I was said, is what you have commented: The subjective trap. Which led to the termination of my project.
Answer these simple question, and you will find what subjective trap is.
1. 4K or 7K, which one is harder? (How do you weight 4k/7k?)
2. Maps which is full of hits or maps which is full of holds, which one is harder? (How do you weight holds?)
3. Stairs or random hits, which is harder?
4. What is the definition of "Hard", anyway?
These question actually fires out a war when my project was announced for feedback in Chinese community.
Just take the third as example (which is mentioned in Tear's post).
When osu!mania was on beta, woc ask me this question. I answered random hits, just like what Tear said without any consideration. And this, along with many other feedbacks, are why stairs are so common in ACs "to prevent the map becoming too hard". But after thousands of PC I found myself wrong - I am far better at random hits than stairs. When I discuss this with many players, I found not everyone is like me - some of them is better at stairs, some of them is not. So how do you determine which one is harder? By polling? With a determined factor on this pattern, there would always be some players that benefits most and some players cry out loud, making it unreliable.
And yeah, too many things are just like stairs and random hits. To "balancing" them is pointless.
This is why I say things should be Simple.
We are playing osu!mania for fun, yeah? So why don't we make the calculation fun?
Instead of making tens of patterns having different weight making things hard to understand and isn't reliable at all, why don't we try to point out a way to calculate difficulty easily, like osu!tp, which can be described in 1~2 pages of paper? Actually fairness is not a must, even with TP every one of us have something like "weight" map which is relative easier to just oneself.