1. osu! forums
  2. osu!
  3. Gameplay & Rankings

Just a quick little introduction here. I've been on these forums for quite a long time and one thing I find that many newer players don't have a proper understanding of is reading. It's a very large topic and no one has really put out anything comprehensive on it. So I decided to write something for all the newer players out there to help them build a process for their reading. It may also help out the more experienced guys as well.

For background about myself, I used to be a high rated player back around 5 years ago (was in the top 500 in ppv2). Since that time, I've decided to go a bit more casual and ended up falling in love with playing EZ mod which I've played for around 4 years on and off. Just a couple of months ago, I decided to switch from playing EZ mod to playing maps edited down to AR0. With that switch, I can officially say I've played almost all ARs with FCs from AR0 all the way to AR10.8.

As such, this guide was built from knowledge I've accumulated from my own experiences. I'm aware that I'm possibly wrong so I'm open to criticism. I stress, this is the method I use to read and think about reading. The guide is not meant to be interpreted as the only way to read, it's just (in my opinion based on logic and experience) the best way to read. As such, it's important to note that everyone reads the game differently and it's up to you to figure out what resonates with you. I may discount some methods of reading however, you would find at least some top player(s) (or former top players) whom read in those discounted manners. So just a quick summary: In how we read, I talk about a model I made that describes the steps we take when we read. In The Signal and the Noise I talk about how noise effects our ability to read. I then talk about specific factors that change reading difficulty in What factors change reading difficulty. In Reading rhythm, I talk all about reading the rhythm of a map (still to be done). I give out some advice on developing a reading process in How to read and then give some advice on what to play in Improving reading. I tried to highlight things I felt were in important in bold.

The main point in this guide is to help give you a better understanding of reading and how it effects the way you play and in helping you develop a specific reading methodology rather than randomly reading and hoping for the best.

From my experience, I find that the process of reading can be divided in 4 steps:
  1. Gathering data
  2. Processing data
  3. Creating a plan
  4. Execution

The first step, gathering data, is the step where you take in the data from the game from the circles on the screen, the approach circles, music, etc. This part is dependent on where you are looking at and what you are focusing your attention on.

Once you have your data, you have to process it to understand what each piece of data means and how to use it. This includes recognising how each circles fits to the music, which circle you need to hit next, etc.

After you process the data, you create a plan of attack. Basically, how you are going to play using the data you gathered. Then you simply need to execute the plan.

All these steps are happening simultaneously while you are playing. You are continuously gathering data, processing it and updating your plan as you gain more information. And obviously, you’re also continuously executing your plans.

These steps are also interconnected pretty heavily. Lacking in 1 area will cause problems along the rest of the chain with step 4 connecting to step 1. Where you’re focusing your attention affects both how you gather data as well as how you execute. To illustrate this, if you’re are playing with too high object density, you’ll tend to overload yourself with data. This data overload is difficult for you process and as such, you can’t create a good plan for it which causes problems during the execution. Or, if you’re playing too high AR, then since you don’t have enough time gathering and processing the data, you can’t create a plan and play it.
Back during my 2nd year of playing, I spent all my time playing HD DT on everything. That’s literally all I played. Now, that’s all well and good if I wanted to only play HD DT (I’ll get onto that later) but when I ventured out to play some equivalent difficulty no mod songs, I hit a wall. Many jumps that I could hit with consistency, I was randomly missing. What I realised in that period was that playing so much HD reduced my ability to play with approach circles. It also led me to a key idea: reading is about filtering for what’s important.

You often hear newer players try to look ahead as much as they can because they worry that if they don’t look ahead, they’ll fall behind. In fact, this is the complete wrong approach. One should be focusing on only the next circle. Okay, I exaggerate a little here but my main point is that you should only pay attention to the minimal amount of information as possible. I consider all information that’ll be relevant in the distant future (and irrelevant at the moment) as noise.

To tie this back to the 4 step process, in particular steps 1 and 4, how you focus your attention impacts how you execute your plan. What I mean by this is that if you look too far ahead, it will impact how you aim and vice versa, look too closely at what you are playing and you won’t have enough time to process the information. Thus, you need to find your Goldilocks Zone; not looking too far ahead that you impact your aim and not too ‘on the now’ to reduce your ability to gather and process the data. This is also why I consider the major reason for most misses is due to poor reading, not poor aim.

A good way to think about reading is to think about it as a queue. A queue is a first-in, first-out system. As circles get added to your queue, the circle at the front of the queue would be your main piece of attention and all circles after that noise. The first circle will leave the queue at a certain point in time depending on your Goldilocks Zone and all the other circles will move down the queue. When the queue moves down, you switch your attention to the next circle.

This queue system does not mean you have to keep track of all the circles at a time. Doing so would be almost impossible and counter-productive. More information does not equal better decision making. In fact, holding more data in your head increases the chance you will make an error in your execution. Personally, when I play, my queue is basically empty and when it’s time to switch my attention, I have to actively find the next circle at the same time! Playing in this way, I improve my execution as much as possible and I only need to improve my data gathering process.
The most obvious factor that everyone knows is object density. Object density is the number of active circles on the screen per unit of time. The more circles you have on your screen at a time, the harder it is to gather and process the data. The object density of a map can be modulated via AR, whereby increasing AR reduces difficulty and reducing AR increases difficulty.

Higher ARs also increase reading difficulty. To dispel any confusions, I’ll address a common myth later on that many players seem to believe which is playing lower ARs improves reading. Continuing on though, higher ARs reduces the time you have to gather and process the data. It should be noted that increasing the AR on a given map will overall reduce the difficulty of the reading as object density is the main factor to reading difficulty. However, a higher AR map doesn’t need as high of an object density compared to a lower AR map to be made just as difficult to read.

Another less thought about change in difficulty comes from circle size. Larger circle sizes contribute to screen clutter. Screen clutter is just noise that makes it harder to gather data. On the flip side, smaller circles are harder to read when they are spaced out. If you play on a large monitor, small spaced circles (like in cross screen jumps) and difficult to read as data gathering is made harder.

The final factor that increases reading difficulty is the way the circles are placed on the screen. This is usually what people refer to as ‘patterns’ but I’ll call it circle configurations as it’s more inclusive of all the various ways circles can be placed on the screen (a pattern is then a type of circle configuration). Certain configurations are more difficult to read than others. The most obvious of these are stacks which can cause a mess when it comes to gathering and process the information. Some more subtle configurations can specifically mess with steps 1 and 4. These tend to be configurations that are closer together and incline the player to move their cursor in a more circular manner as the closer circles provide noise to distract the player.

Mods modulate the reading difficulty due to the first 2 factors. DT increases reading difficulty as the object density remains the same but there’s high AR which means the same amount of data needs to be processed in less time. HR overall reduces reading difficulty as the higher AR reduces the object density and screen clutter (though this is very minor) however, the faster processing speeds are required plus cross screen jumps are also made harder to read. EZ increases reading difficulty in 2 ways; lower AR to increase object density and bigger circles which add more screen clutter.
The last mod that hasn’t been mentioned yet, HD, has an interesting relationship to reading. It can both make things harder to read as well as easier. Because HD hides circles, there is less screen clutter so processing data is made easier. Additionally, the hidden approach circles makes execution a lot easier as approach circles are noise during that step. Conversely, when HD is paired up with higher object densities, it’s extremely difficult to read. The difficulty comes from finding the Goldilocks Zone as disappearing circles adds complexity to the planning, execution and data gathering steps. You only have so much attention to give at any point in time. This is what makes playing EZ HD a nightmare (it’s fun once you figure it out).

In general, there’s a positive correlation between the star difficulty rating of a map and its reading difficulty. Thus, as one strives to improve their reading, one tends to only need to increase the difficulty of the maps they play.
To be done
The first step in learning to read properly, is knowing where to look and when. My general advice for this aspect is to look at what’s most important. This means focusing your attention between the circle you need to hit now and the circle that you need to hit next. All other circles are noise. This means looking directly at the circle you need to hit and viewing everything else through your periphery.

The importance of looking directly at a circle is two-fold; increase your aim precision and filtering out noise. The general way your eye works is light gets filtered towards your retina with where you focus your attention towards being filtered to a structure called the fovea centralis. The information coming from the fovea centralis is the most accurate and clear whereas the information coming from the rest of your retina is blurry. Therefore, if you try to aim where you aren’t looking directly, you will be prone to more error. The second benefit comes from the blurry data you receive from the rest of the playing field hitting the rest of the retina. Because the data isn’t very clear, you will be less distracted by it and therefore can execute more effectively.

There are 2 major drawbacks with this method of play. The first drawback is that blind spots are more impactful. This is more of an issue when playing with small circles that are spaced out. However, this issue is quite rare (I can count how many times I’ve experienced it on 1 hand) and one can easily compensate by just memorising small sections. The other drawback is that moving your eye many times can quickly cause fatigue. This drawback can be overcome with practice.

The when to look is a tricky part that will come with lots of experimentation. Like Goldilocks, you have to taste the too hot and too cold porridge before you find the one that’s just right. The best advice I have to give for this is to look at the next circle just as you’re about to hit the current circle.

In terms of dealing with data from your periphery, I find it’s important to note something if you can but not to make too big of a deal if you can’t. Over time, you will generally just get better at dealing with peripheral data as you play.

This brings me to the topic of patterns (or what I’ve called circle configurations). Our brain is good at recognising patterns and because of this, we like to label all patterns with names like squares, stars, etc. Because we label these patterns, people like to figure out specific ways of playing these patterns. This is in fact the wrong approach. This is because these patterns aren’t mapped out the exact same way all the time. For example, you could practice playing a specific star pattern in a specific way but you’d fall flat if the pattern came in the opposite direction instead. The better approach would be to look at each circle individually. That way, you’re just moving your cursor from A to B to C instead of trying to plan out the exact way to get from A to C while having to hit B as well. If you add in my noise principle, then while moving from A to B, C is just noise. In that way, what you are playing is no different from just a couple of simple straight line jumps. This can be extended towards every type of circle configuration from high spacing to very little spacing.

The importance is just to focus on what you need to focus on and drowning out all the noise. In that way, a map can be distilled into a combination of many single straight line jumps. This is obviously easier said than done as noise will impact your ability to read in this manner. Good reading is therefore the ability to filter out the noise completely until it is relevant.

The last piece of advice on how to read I will leave you with is the aspect of focus. Reading is an extremely effortful activity. Because of this, when we find aspects that are ‘easy’, we tend to lose focus. It’s important to remember that you need to maintain focus as much as possible while reading. Any lapse in focus will cause mistakes like random misses and if the map is somewhat difficult to read, complete lapses in processing.
Reading is an extremely difficult skillset to get better at. It’s extremely difficult to measure, fluctuates a lot and has various independent facets. The phrase “you are what you eat” applies extremely well to reading as a skillset. Simply put, if you want to get good at a certain aspect of reading, you have to practice it. Reading is like a muscle memory activity, you have to practice a lot and practice a large variety.

When looking to improve you’re reading, the first step would be to ask yourself “what are my goals?” If you want to become a HR player, then you only really need to play HR to improve your reading. If you want to become a HD player, then you only really need to play HD to improve your reading. Etc, etc. Conventional wisdom says that you need to play everything, but in reality, you only need to play a variety of whatever you want to get good at whether it be HD, DT, etc.

I caution here because improving reading in one aspect of the game does not necessarily improve your reading in other aspects of the game. For example, I switched from playing EZ mod to playing maps edited down to AR0 and my reading of EZ hasn’t gotten any better (in fact, it’s gone worse from neglect). Don’t get me wrong though, EZ mod definitely helped in my transition to AR0 as well as AR0 helped in some aspects of my EZ mod game. There are definitely overlaps but those overlaps aren’t enough if you want to specialise.

On the other hand, these overlaps can help greatly if you just want to be an all-round player. To give some examples in my experience, learning to read streams on EZ mod taught me how to read streams on the ‘normal’ ARs as I gained a specific process on how to read streams which I didn’t really have prior. It is also my opinion (from experience) that playing streams on lower ARs is the most effective way to learn how to read streams. Also, playing EZ HD made playing HD on the ‘normal’ ARs to be so much easier as snapping on EZ HD is extremely difficult due to the step 1 and 4 effect and learning to snap on EZ HD transferred to HD on the ‘normal’ ARs.

From what I can tell, there are a couple different types of reading training you can do. The first is playing things just outside your comfort zone. Like all other types of training, you need to experience stress to grow. The other type is playing things you find somewhat easy to FC. I liken this type of training to playing scales on a piano; it strengthens fundamental abilities. Playing these type of maps improves your consistency.

This brings me to the myth I spoke about earlier, playing lower AR improves your reading. Generally, when people give this advice, they usually recommend playing maps that are AR8. The problem with this method is that there aren’t that many difficult AR8 maps and AR8 maps come in all various shapes and sizes. If you take this advice at face value, you might be tempted to think that just playing any map at AR8 would improve your reading when in fact, it won’t do any good for you. A better alternative would be to edit maps down in AR. Find maps that you can FC without much problem and then edit them down 1 AR.

Just to be clear, I am not stating that playing AR8 is bad for training reading. It's all about context.
Isn’t this just a long and convoluted way to say pick your skin and practise more at every ar range.
Not at all. If you actually read it, you would've known that I actually disagree with playing every AR range. Nor do I mention anything about skinning. This guide simply expresses (in my opinion) best practices when reading or some ideas on how to become more methodical in reading.
Really well thought out post. I can't believe it hasn't gotten any attention. I really find the topic of reading interesting for the reasons you say ("It’s extremely difficult to measure, fluctuates a lot and has various independent facets") and I've collected the opinions of various people on the forums and youtube.

My favorite part about this guide is that it verifies everything I've come to try and implement into my own gameplay. I heard rohulk talk about his way of reading the notes as they come up, similar to your idea of the "Goldilocks zone" or "queue". This idea makes a lot of sense to me, and when I implement it I feel like I'm playing "properly". I like to think of it as a conveyor belt myself. I've noticed that I feel the most in the zone when I'm reading in that way, but it also takes a lot of focus. Sometimes I want to look at the whole screen because I don't trust myself to be able to hit the pattern without trying to gather more data. However, this is usually counterproductive, because, as you pointed out, I'm trying to gather too much data and end up messing up in the processing or execution step as a result.

I also 100% agree that reading is a lot about focus. When I want to FC consistently, I feel like more than warming up my aim or streams, I just need to focus as much as possible, and that whether or not I FC is often a matter of if I kept my eyes looking at the right stuff. For example, when I start to panic because I'm FCing, I almost always start looking ahead because I "don't want to miss something" and end up moving too fast in execution. In other instances, I will just make my focus very broad. And if the map gets too easy I can lose proper focus and miss on something because I misread it and felt like I could just let my muscle memory do it.

I admit that I'm one of the people that thought AR8 improved reading. Your argument is that most AR8 maps tend to be too easy. However, I think for lower ranked players who are already playing some of the easier maps it can help. Wouldn't you agree?

I want to go into some questions and potential criticisms I have, but it's not to try and undermine what you say, but because I'm honestly curious and want to pick your brain for what you think.

The Model:

I see why you made the model the way you did, and as they say all models are wrong but some are useful. And this is a very useful model. However, I feel like there is some overlap between the steps.

For example, if you overload yourself on information and you mess up, who is to say whether this mess up occured because of your processing or your plan or your execution? Wouldn't an execution mess up be a result of a previous step? Would an execution mess up by itself just be a mechanical error and thus not be related to reading? Overall, the model makes plenty of sense, but I don't fully understand this part.

Moving in a straight line and focusing on one circle:

While I agree that this makes sense most of the time, I believe there are some patterns that most players can partially hit through muscle memory. I know because with certain things, like back and forths for example, your hand knows how to go back and forth between two spots without even looking. Of course this gets trickier the more complex the pattern becomes and the more likely you are to mess it up (like overlapping star pattern that's really close together and comes back in on itself). Of course it is probably almost always better to look, but I get the feeling that you learn an adaptable muscle memory for certain patterns. Even though no two patterns are exactly alike (with some exceptions), I feel like this muscle memory aids a lot in helping you hit things, and in some cases can even be leaned on as soon as you recognize the pattern.

My exhibit A for this is Cookiezi, who has said in his Happystick interview that at this point he just looks at the center of his screen and plays through periphery. Granted, he admits in the same breath that it is more consistent to actually look at the circles, but it's more fatiguing and makes him too tired, which goes along with what you said. Anyway, Cookie probably looked more at circles in the past and built up his reading that way, but the fact that he can play at such a high level now makes it seem like it's a totally viable way to play.

Pattern Recognition/Circle Configurations:

One more thing with Cookiezi from this interview that reinforces my previous point is that Toy and Happy speculate that Cookiezi looks at osu! as a series of patterns in his head, and once he knows how to play a pattern he can (almost) always do it. Cookiezi also talks in this way. To paraphrase, he said something like "If you can't hit it you just aren't good enough, you can't expect to hit harder versions of a pattern if you can't hit easier ones".

Also, it's commonly accepted wisdom that playing many maps and exposing yourself to patterns makes it easier to play things you have never seen before. I personally think that pattern recognition is a big part of reading as well and it doesn't all come down to hitting what's in front of you. I think in the periphery you recognize patterns and prime yourself to hit them in your "queue". This also ties back into muscle memory, because although your hand hasn't done the same motions, it's certainly done similar ones and they will be easier the next time you do them.

However, this particular part doesn't really run counter to your model, but I think it could be integrated with it. Perhaps this could all be included in execution.


Cookiezi Interview: https://youtu.be/4-0rZ8YwGnE

Rohulk Interview: https://youtu.be/qDTdc4Onnw8
I think everyone who wants to learn how to read low AR should start off with ar 8 don't go off rushing your reading and play maps like oddloop, tewi-ma, (insert other heccin low ar maps)
Great explanation of reading and how it works. I feel that one thing you were missing here is the mention of rhythmic accuracy as an aspect of reading.

Reading can be considered the comprehension and ability to act on what you can comprehend (in other words, process). That being said, how you understand the rhythm should also be a part of reading. This deserves a mention albeit how subconscious it is -- because you don't have to put much conscious effort into understanding the beat, you just know how to play to the beat after playing for some time and can do it better as you play more. But although it is so much more subconscious than visual reading, it can still be improved. After all, the measure of how much more than others that you improve is based on how you train your subconscious.

With skill in osu! being a largely mechanical or subconscious thing, that means that there are ways to improve how fast your subconscious learns. Pertaining to rhythmic accuracy, you naturally learn how to play to the beat the more you successfully play to the beat. So in order to improve in this area you would have to successfully play to the beat more. This follows the idea that consistency is the best way to improve (any skill as well as consistency in itself). Consistency is composed of every skill as much as every skill is composed of consistency, and thus consistency is the ultimate differentiator--the foundation and peak of your skill.

Although consistency is the best way to refine and differentiate yourself from others of the same skill level, you still need to be challenged in order to make reasonable gains in individual skills. This means increasing your skill level entirely. Consistency will only bring minimal gains in every skill, which includes reading (excluding reading extremes such as AR0 and AR11), and thus cannot solely improve your skill level. This is why consistency is a skill in itself as much as it is a component that makes up every other skill.

Ideally, you want to find a good midpoint between consistency and challenge to improve optimally.

Just adding onto what you said as well as introducing a theory.
Thanks for the feedback guys, I took a short hiatus so I didn't notice I got some responses. I'm going to respond to all of it here so bear with me.

I'm glad you liked the guide :)

To your question on the overlaps within the model, you are absolutely correct. I tried to separate the steps to make it simpler but many of the processes within each steps overlap with other steps so it was difficult... In terms of mess ups between the steps, that would be based on a case-by-case basis. It could be mess ups in 1 step or in all. It's up to the individual to decide in what ways they screwed up in specific instances.

To your second question on pattern recognition, I think we can all agree that playing as if you are only hitting between individual circles is a lot easier than playing a 'star' or a 'square'. To a new player, when they come across these patterns, they tend to try to play the pattern rather than playing it as it is. A major problem with playing a pattern as a pattern is that it encourages laziness in your reading. Instead of looking at circles individually, you look at the whole and this in turn effects your execution (just look at how many 'squares' are being playing in a circle motion). Your brain likes to take shortcuts wherever possible as it conserves energy.

With your Cookiezi quote, I'd like to think of it more as "If you can't play something then you aren't good enough, you can't expect to play a hard map if you can't even play the easier ones". I don't like to think of it in a pattern based context as just getting better at the game in general would allow you to play better overall. As an example, if I have problems reading a map, I know I'm not good enough to play it. I don't need to pick out all the configurations I had problems reading in that map and practice them specifically to eventually be able to read it. If I just challenge myself overall, eventually I'll be able to just play it. The problem is not in the inability to recognise the pattern effectively but more in other issues of my reading.

But hey, whatever floats your boat right? If you believe that pattern recognition is important to your reading, you can add it in to your process. As I stated initially in the OP, this is the way I read and like to think about reading based on my own experiences. I just simply believe pattern recognition isn't a good first step for newer players.

This leads me to your point on periphery reading. I will just say that reading only through periphery is actually a viable strategy. Many top players do it/have done it and I friend of mine who was rated as one of the best EZ mod players back a couple of years ago told me he too reads through periphery only. I too believe that periphery reading is important to some extent. I don't really clearly mention it in my OP but periphery only reading has the major drawback of more error in your aim/execution (because of eye structure) which is why I discourage it.
You are absolutely right about the rhythmic reading being left out. I did hint to it a little in "How we read" when I talk about data gathering. It's actually a topic I find pretty curious as the rhythm of the map is majorly dependent on how the mapper mapped the song thus even a lower difficulty map can be more rhythmically difficult to read. I was initially going to add a section on it but I decided against it as the easiest way around this is just to watch auto or a replay of it to get a feel for the rhythm of the map. I will add it onto my to-do list though as I feel there are things I can mention about it.

And to everyone questioning my "AR8 isn't that great for improving reading" comment, I'll elaborate a little. The advice is simply a massive bandage on the actual problem, it doesn't really give any clear advice on what to do. Firstly, being able to read a large range of ARs doesn't really provide big benefits. The way you read different ARs is vastly different. You have different Goldilocks Zones and different ways you store the data in your queue. I liken it to the humorous joke people make about bodybuilders being unable to lift a pebble. There is definitely some cross-over from being able to play the high density low ARs to higher ARs but it's not as important as one might think. As such, you're better off learning to read the AR you find important so you get more practice refining your Goldilocks Zones and queue process for those ARs.

Secondly, reading difficulty comes more from the object density rather than the AR itself. AR8 doesn't necessarily guarantee higher object density versus AR9 (or dare I say AR10). If anything, if you really wanted to reduce the AR and practice reading solely, you're better off dropping the AR (and star difficulty) a lot lower than 8. But, as I mentioned earlier, it's a pretty fruitless activity.

And finally, I want to emphasise that learning and improving upon your reading process is the most important thing above all else. Playing lower ARs on lower difficulties is the best way at getting to know your reading as it gives you more time to really think about how you are reading rather than just blindly reacting on higher ARs. That is not to say that it's not possible to practice reading on higher ARs. If anything, it's better to learn to read on ARs you find most important (which for most people, doesn't involve low ARs) and just put more effort into figuring out the how and what you are doing.

In regards to szkiller's comment, I disagree with that. Learning to play low ARs isn't difficult at all. In fact, everyone can play AR0 to some extent. The difficulty of playing lower ARs is trying to play it on higher star difficulties. If you really want to learn a low AR, just drop the star difficulty until you can actually play it and go from there.
Please sign in to reply.