I have a backgroung with a lot of rhythm games, mostly dance games. With a lot of experience on scoring systems, and a small hate for combo based scoring, I thought of ways to include a non-combo based scoring system to osu, similiar to the percentage based system that we use for In The Groove scoring on the international ranking. We had an irc-conversation about possibilities, and came to a conclusion that the accuracy statistic is just what I was looking for, but it would require some modification. In addition to presenting a non-combo scoring system the changes that were discussed would fix some problems that the current accuracy system has, such as awkward accuracy ratings for accurate players if they started off with a really poor accuracy in the beginning of their osu carreer.
Basically this is what we discussed:
* Each difficulty level of each beatmap (easy, normal, etc) would have it's accuracy counted from the best score of the player. In other words: There's not just one accuracy score per song, but as many accuracy scores as there are difficulty levels in it.
* The total accuracy of the player would be the avarage of these beatmap difficulty level accuracies.
* The player would also have the avarage accuracy for each difficulty level separately. The total accuracy for easy would be the avarage of all easy level accuracies, etc.
There are two ways to calculate the total avarages. Either all unplayed songs can be ignored, or all of them can be counted as accuray = 0%. The first method allows players to just play one or two songs of one difficulty level to get a very high avarage accuracy (for example play one easy song with 100% accuracy and get a 100% avarage accuracy for easy level), but does not a player to play all of the beatmaps available. The second method works as a progress meter for playing all songs, and only at the later stages - when players have played all songs - as a real accuracy meter.
I personally think that the first method for avarage calculating would work better, since the main ranking method for osu is still the points system, so playing just a few songs harms your score gathering, and with this method the accuracy rating actually tells something about the accuracy of the player.
Also, since the old accuracy meter was sort of a 'dynamic meter' that rises and falls depending on your current game play and not the best accomplishments that you have for each song, a new meter would most likely be created to work as a replacement for the old accuracy rating. This meter would reflect your current state of game play, rising and falling depending on your plays, just like accuracy rating currently does. This means that no one loses anything, and non-combo oriented players get a non-combo oriented rating system.
Basically this is what we discussed:
* Each difficulty level of each beatmap (easy, normal, etc) would have it's accuracy counted from the best score of the player. In other words: There's not just one accuracy score per song, but as many accuracy scores as there are difficulty levels in it.
* The total accuracy of the player would be the avarage of these beatmap difficulty level accuracies.
* The player would also have the avarage accuracy for each difficulty level separately. The total accuracy for easy would be the avarage of all easy level accuracies, etc.
There are two ways to calculate the total avarages. Either all unplayed songs can be ignored, or all of them can be counted as accuray = 0%. The first method allows players to just play one or two songs of one difficulty level to get a very high avarage accuracy (for example play one easy song with 100% accuracy and get a 100% avarage accuracy for easy level), but does not a player to play all of the beatmaps available. The second method works as a progress meter for playing all songs, and only at the later stages - when players have played all songs - as a real accuracy meter.
I personally think that the first method for avarage calculating would work better, since the main ranking method for osu is still the points system, so playing just a few songs harms your score gathering, and with this method the accuracy rating actually tells something about the accuracy of the player.
Also, since the old accuracy meter was sort of a 'dynamic meter' that rises and falls depending on your current game play and not the best accomplishments that you have for each song, a new meter would most likely be created to work as a replacement for the old accuracy rating. This meter would reflect your current state of game play, rising and falling depending on your plays, just like accuracy rating currently does. This means that no one loses anything, and non-combo oriented players get a non-combo oriented rating system.