On August 11 2015 05:17 Yorbon wrote: As long as you cannot be reasonably sure you can repeat a score, that score does not represent your skill level.
I have huge issues with this statement actually. Very few players would have any kind of representation of their skill level on their profile if that were true.
I have some freak scores up there. Mostly it's just a lucky run on acc - hard to repeat, but the FC itself not that difficult
On August 11 2015 05:17 Yorbon wrote: As long as you cannot be reasonably sure you can repeat a score, that score does not represent your skill level.
I have huge issues with this statement actually. Very few players would have any kind of representation of their skill level on their profile if that were true.
This is not entirely correct. I do agree that the individual scores on someone's profile do not represent the skill level of that person, assuming what i said. However, the whole of someone's scores could say something about his skill level, because those scores collectively are unobtainable for someone with a (significantly) lower skill level within a reasonable amount of time. One can estimate a minimum skill level that way. This still gives a very weak estimate, however, as it still is dependent on time spent achieving those scores as well as skill. Someone who has spent 2 years at the same skill level, will be expected to have higher pp than someone at that skill level who's recently improved a lot. You might also say that increased farming activity (per amount of time) drives amount of pp upwards. So regardless of whether a score represents your skill level, your skill level is not represented well by your profile at all. (Please note that I said 'not well at all' and not 'not at all' just in case). I would argue that the higher the rank achieved, the more accurate the amount of pp is, because the attainability of those scores through sheer repetition decrease more rapidly (than at the lower levels) the higher the amount of pp.
Worth noting is also that this 'upward bias' is not necessarily a consequence of the current pp system or something. It is a bias coming from the fact that osu takes your maximum score, which correlates with highest pp score (note: correlate; no 1 on 1 relation). These kind of biases cannot be prevented by any kind of (reasonable) system, I think. So while I still accept the current system, it has it's flaws. One might see the scores on someone's profile as a first approximation of someone's skill.
Going back to when I saw your profile online just now. I saw you were rank 500, so assumed a (probable) minimum level. After that I just looked at scores you obtained and saw you are probably better than me (you at least were at that moment). Not based on rank, profile scores, or scores obtained after repeating, but based on scores on maps you had no way of knowing were coming.
I hope this is structured enough to follow. And sorry for all things in brackets; i hate written media.
tldr: Profile does not represent skill anyway, except at the highest level.
Sorry for being blunt, but the "skill" you're picturing there, that your referring to, does not exist. At least in the way you make it out to be (unified overall skill) - Me being better in maps that I have no idea were coming could be something ENTIRELY different than the skill needed to fc this or that DT map. I disagree with anyone stating the existence of a strong correlation between random map performance and sort-of-grindy-kindof map performance needed to gain PP.
There are at least 5 areas in osu! which require completely different skills. All the PP system does is take your peak performance and make it into a number, completely disregarding long term consistency, a possible glaring weakness in either aim, accuracy or speed, and the finer points of good sightreading.
So all PP can possibly do is show your peak "skill" over some time and it does that well: A few outliers in terms of single performances show a high peak performance
tl;dr there is no pure skill in PP anyway, because pure skill doesn't exist, and i think PP does a very good job showing your peak performance capabilities.
I have tried out just now few maps now. I can't survive 4.5 stars and I'm not in the best form for 4 stars. Looks like I'll need more to be back to where I used to be.
On August 11 2015 05:17 Yorbon wrote: As long as you cannot be reasonably sure you can repeat a score, that score does not represent your skill level.
I have huge issues with this statement actually. Very few players would have any kind of representation of their skill level on their profile if that were true.
This is not entirely correct. I do agree that the individual scores on someone's profile do not represent the skill level of that person, assuming what i said. However, the whole of someone's scores could say something about his skill level, because those scores collectively are unobtainable for someone with a (significantly) lower skill level within a reasonable amount of time. One can estimate a minimum skill level that way. This still gives a very weak estimate, however, as it still is dependent on time spent achieving those scores as well as skill. Someone who has spent 2 years at the same skill level, will be expected to have higher pp than someone at that skill level who's recently improved a lot. You might also say that increased farming activity (per amount of time) drives amount of pp upwards. So regardless of whether a score represents your skill level, your skill level is not represented well by your profile at all. (Please note that I said 'not well at all' and not 'not at all' just in case). I would argue that the higher the rank achieved, the more accurate the amount of pp is, because the attainability of those scores through sheer repetition decrease more rapidly (than at the lower levels) the higher the amount of pp.
Worth noting is also that this 'upward bias' is not necessarily a consequence of the current pp system or something. It is a bias coming from the fact that osu takes your maximum score, which correlates with highest pp score (note: correlate; no 1 on 1 relation). These kind of biases cannot be prevented by any kind of (reasonable) system, I think. So while I still accept the current system, it has it's flaws. One might see the scores on someone's profile as a first approximation of someone's skill.
Going back to when I saw your profile online just now. I saw you were rank 500, so assumed a (probable) minimum level. After that I just looked at scores you obtained and saw you are probably better than me (you at least were at that moment). Not based on rank, profile scores, or scores obtained after repeating, but based on scores on maps you had no way of knowing were coming.
I hope this is structured enough to follow. And sorry for all things in brackets; i hate written media.
tldr: Profile does not represent skill anyway, except at the highest level.
Sorry for being blunt, but the "skill" you're picturing there, that your referring to, does not exist. At least in the way you make it out to be (unified overall skill) - Me being better in maps that I have no idea were coming could be something ENTIRELY different than the skill needed to fc this or that DT map. I disagree with anyone stating the existence of a strong correlation between random map performance and sort-of-grindy-kindof map performance needed to gain PP.
There are at least 5 areas in osu! which require completely different skills. All the PP system does is take your peak performance and make it into a number, completely disregarding long term consistency, a possible glaring weakness in either aim, accuracy or speed, and the finer points of good sightreading.
So all PP can possibly do is show your peak "skill" over some time and it does that well: A few outliers in terms of single performances show a high peak performance
tl;dr there is no pure skill in PP anyway, because pure skill doesn't exist, and i think PP does a very good job showing your peak performance capabilities.
I don't mind anyone being blunt, we'll just agree to disagree (at some parts at least), I guess.
One thing I would like to say though, as clarification, is that I never intended to picture skill as a one-dimensional feature. When you say there are multiple 'areas' requiring different skills I could most probably agree with you, dependent on how you would view them, and any assessment of skill should (imo) be evaluated in each of those areas, however one would choose them.
Hi, I'm looking for some nice 4* - 4,5* maps to play. I'm at the point where I can get S ranks on 3,5* maps and I want to try some harder ones, but I don't know which ones are good. I usually go for songs I like but western songs tend to have pretty shitty beatmaps. Cheers
edit: just saw the post on the last page, will check those out
And with regards to the recent discussion about playing the same map too long, that's pretty much all I do Pick a map, try to complete it, then try to get a perfect combo on it, then move on to another map.
I think going by sources is the easiest way to get the majority of "good" maps in osu. Bemani (iidx, jubeat, reflec beat, popn, sdvx), BMS, and Touhou somehow managed to attract a bunch of the really good mappers. I don't know if they have many in the 3.5*-4.5* range but some of my favourites are Kloyd, Guy, Asphyxia, Cherry Blossom, sionKotori, wring, Liiraye, RLC, Priti, Flower, alacat, eLy, Skystar, and AngelHoney. Try a bunch of those if you haven't. When you play a map and like it, go to the mappers profile and download all their other maps as well, search the artist, search the source etc.
On August 11 2015 17:33 deakachu wrote: When you play a map and like it, go to the mappers profile and download all their other maps as well, search the artist, search the source etc.
This, a thousand times. I've never understood why more people don't pay attention to who mapped something, because it's by far the easiest way to discover new maps.
I kind of see both points of view as far as the PP = skill argument goes. I think that rank is a fairly good summation of all skills, but doesn't really tell you anything about a player's strengths or weaknesses unless you really analyze their top scores. For example, I'm sitting at about rank 30k right now. I think my raw aim potential (good at hitting hard patterns and passing hard maps where missed notes are acceptable), reading, and low bpm (>190) streaming are above my rank, my accuracy is probably the same as my rank, and that my aim consistency (bad at going the entire map without missing, even on maps "below" my level) and speed are below my rank (This assessment could be totally off by the way, just my best guess). All in all, I think that my rank is a pretty good representation of my skills as a whole.
PP is a godawful representation of skill given that skill is a multidimensional variable and PP is a single value. You're always going to lose a lot when you condense multiple values into a single one. But there's still correlation between PP and skill, because while a player may have more skill than their PP suggests, it's much harder to have less skill than your PP suggests. Now, to be clear I'm talking the difference between say a player with like less than 1k PP compared to a player with 2k+ PP. Chances are pretty good that the 2k+ PP player is noticeably better, unless the 1k PP player simply hasn't played shit.
Nice, deakachu! I agree your timing seemed off sometimes, but in terms of aim and consistency during the song, this is quite nice
The play also reminds me that I've been thinking about removing the numbers as part of the hitcircles. Those numbers may be distracting. Could you maybe link me your skin, just to try out?
Edit: I thought it was the extra difficulty when posting this post :O Sorry! I've never actually played this difficulty afaik :$
Here is my skin, it's just Azer's skin with Voyen's numbers I think, and I changed some menu images. I use map hitsounds so don't know what they will be like if you don't.
I think anything that I have problems reading is too fast for me to be able to process the numbers, so they are more of a distraction than anything. It is probably just preference though.