Why the scoring system should change

[Fio]

Getting this out of the way: this is probably the most fun rhythm game I've ever played and it's why I took the time to complain. People have brought up points about the current scoring system ( https://twitter.com/BeatSaber/status/992782948515622913 ) being insufficient but I don't think all of what I find are the biggest reasons are accurately covered or easily understood in a single post.

1. It doesn't scale with what the community will throw at it.

The current scoring system will not be able to scale with the kinds of 'super expert' tracks the community will undoubtedly make (read: has been making). The scoring system for a game that will be largely community-created should be scalable to the widest spectrum of possibilities (read: simple). Defining the correct full stroke of a cut renders certain speeds and patterns impossible to execute 'properly'. These kinds of super-fast beatfiles will be what comes to represent Beat Saber in media and the current scoring system won't be able to score them with any degree of meaning.

You can argue "Well the community shouldn't create beatfiles that the engine can't score", but I really hope you've realized how ridiculous that statement is before you've finished saying it.

2. It isn't intuitive. At all.

I'm not against doing things differently and I'm not saying Beat Saber needs to be like StepMania, but I am saying that StepMania's scoring works for every kind of stepfile that the community can make, and, importantly, is intuitive. Hitting the right arrow gives me points. Hitting them closer to the beat gives me more points. Hitting multiple in a row gives me combos. Nobody ever told me how to play StepMania and it scores me exactly how it seems like it should.

Nobody had to tell me how to play Beat Saber either, but Beat Saber does not score me how it seems like it should. You are being scored on hitting the correct direction, you are being scored (slightly) on accuracy of your cut. Okay, I am with you still, those make sense. But you're not being scored on timing and you are being scored on... starting with a 90 degree swing and ending with a 30 degree follow-through...? It is unintuitive and needlessly complex, when the rest of the game is a glowing achievement of simplicity. You don't have to reinvent the wheel when the wheel rolls like we expect it to.

3. It downplays one of the most unique aspects of the game.

Returning to a comparison with StepMania: I've played too much StepMania, definitely a favorite, but I have no problem admitting that there is really no personal expression in StepMania. It's a series of instructions and you are expected to execute them as-ordered, on time. To me one of the most interesting aspects of Beat Saber is how the physicality of VR opens up so much room for personal expression. I can watch people playing the same track with the same difficulty I play it at, but who approach their movements significantly differently. There's a surprising range of choices to make about how your body and arms interact to achieve the same goal.

However, that freedom isn't supported by the current scoring engine. The current scoring engine wants you to cut a certain way, it reduces the range of techniques to achieve what shouldn't be required of you to do.

Is the fun lost in cut freedom offset by fun gained in requiring us to make cuts a specific way? If not, you are at a fun-negative value and have a feature that should not exist. Let's be clear: if you like making huge cuts you can still do that with a more generalized scoring system. That's what's nice about freedom. Sometimes I like making really exaggerated cuts, too.


Personally, if the current scoring system stays I'm just going to ignore scores. Telling me how to swing isn't making the game more fun, and scores are not going to be relevant with the sugar-rush community tracks that I'm interested in. I'll still play Beat Saber because it's great fun, but I hope the scoring is made more generic so it can be applied in a more consistent way. Thanks for reading if you got this far.

[xenicat]

Początkowo opublikowane przez McFC:

I dare say that mathematically speaking, it is impossible to get both FC and highest possible score in the same run because the action economy of the technique rewards are antagonistic to the requirements for an FC in numerous cases.

This is literally the dumbest thing I have read in quite a while. It's the opposite of reality no matter how you interpret it but since you brought up math let's actually look at the numbers to see why.

The highest possible score is achieved by getting 110 on every block for a score of (blocks*8-70)*110 (the -70 takes care of that you don't start at x8.) So for example a song with 377 blocks has a highest possible score of 324060. Not getting a full combo means hitting a bomb/obstacle which will drop your multiplier down to x4 for the next 8 blocks which means you miss out on 3520 points or you miss a block which will cost you 880 from that missed block in addition to dropping the multiplier for a total of 4400 points off of the highest possible score. So no full combo means automatically no highest possible score.

Ok, let's not be a grammar nazi here and assume you meant highest score that's practically possible which means there would be some kind of situation where missing a block on purpose in favor of a perfect swing at another block could net you a larger score overall thus making no FC better than FC. Alright as I have mentioned earlier missing a block drops your multiplier to x4 for the next eight blocks until it goes up to x8 again. So the situation we will look at is you are at a x8 multiplier and go for a perfect swing then miss the next one as a result and afterwards hit the eight blocks for 10 points each and we will compare it to hitting all these ten blocks for 10 points. Afterwards both strategies will be back at x8 where both are equal again so these ten blocks are the only ones we have to consider to calculate what the impact of each strategy on the score is. We'll get:

Intentional miss: 1*8*110 + 0 + 8*4*10 = 1200
FC: 10*8*10 = 800

Looks like a good deal doesn't it? Sure but in reality getting only 10 points per block consistently is pretty damn impossible the lowest I could go consistently was around 25 without the game counting it as a bad cut and that is including only chopping off the corner to not get the 10 points for precision. So if we replace the 10 points per block with 25 and do the same calculation we get:

Intentional miss: 1*8*110 + 0 + 8*4*25 = 1680
FC: 10*8*25 = 2000

It's clearly in favor of FC now. If you want the exact number where FC becomes better than intentionally missing a block it's starting at 19 points. If you apply this to getting a perfect hit with each saber and then missing one each in return your multiplier will drop to x2 but it's slightly better, now FC only becomes better starting at 21 points per block but it is still unrealistic to hit such low scores consistently (prove me wrong but I really tried and when I was hitting so light and off center that I got under 20 more than half of the blocks didn't even count as a hit anymore.)

Since we've just looked at the lower end but you mentionend being competitive in your post let's also look at how punishing a miss (intentional or not) is at higher levels. I don't consider myself the best but I've played for over 60 hours at this point and I've got a couple top 20 scores on the original expert maps so I think my scores and averages will do as an example. For those who don't know a good estimate (to get the exact average you would need to know the score of each individual block) for your average score per block if you had a full combo is:

Score/(blocks*8-70)

If you had misses it's more complicated. With isolated misses (meaning you were at x8 and got back to x8 before the next miss/end of song) you subtract another 32 for each one from the denominator.

So here are three of my highest ranked scores all of them were full combos:
Turn me on, expert, 341 blocks, 267214 -> 101 average
Balearic Pumping, expert, 459 blocks, 350954 -> 97 average
$100 Bills (one saber), expert, 377, 280631 -> 95 average

Taking just these three into account thats around 98 average overall. Putting that into the equations above gives:

Intentional miss: 1*8*110 + 0 + 8*4*98 = 4016
FC: 10*8*98 = 7840

That means in full try hard mode an isolated miss costs me 3824 points which is an auto restart if I'm for a PB. Even if you could make up for it by raising your average which is really hard since there isn't much room to grow you will know that you could have had almost 4000 points more by just not making that mistake and that would take all the joy out of getting a new PB for me.

So we've established that a full combo is better in every realistic situation to get a higher score and your statement is mathematically speaking straight up stupid.

But there is more.

Początkowo opublikowane przez McFC:

I'm admittedly competitive, and this game really pisses me off with its scoring. Being "better" doesn't net you more points. Instead, you gotta be randomly flashy in ways unknown.
[...]
Personally, I don't want to play worse to get a better score. That doesn't make sense to me.
[...]
This forces you to choose between two goals: Do you want to beat every song fully? Or do you want the highest score amongst your group?
[...]
I haven't tested this, but I'm fairly sure that I could probably just flail my arms around in a semi intelligent manner, focusing just enough to not fail the level, and still get a solid competitive score. I've witnessed others do it.

This is just you fundamentally disagreeing on what it means to be good at the game. Sure part of it is hitting all blocks but it is and in my opinion should be even more important to actually swing hard at these blocks (I'll get to wrist flicking later.) Thats the whole point of even using VR for this game, enabling and encouraging you to actually move. If you want to take that away and reduce it to tapping on the highlighted side as precisely in the middle as you can you could just make a phone port out of it. It's pretty much Fruit Ninja with a lightsaber skin at that point. I would say that swinging hard is more important to the devs as well judging by how the grade system C to SS is based on score (with the scoring system they implemented) and not longest combo.

Also if that second part takes only half a brain why can't you just integrate it into your gameplay? Your friends are good at the second part and you are "good" at the first. I've put it in quotes because I imagine you struggling really really hard to get a full combo because there is no way you would write this otherwise:

Początkowo opublikowane przez McFC:

The real achievement is getting an FC. The score is just so bonkers that it is meaningless. The youtube videos that wow people all have FC in their title, but FC doesn't get the credit it deserves in the current scoring scheme. It should come with bad ass animations and light shows at the end. Your achievement enshrined into the leaderboard in a rewarding way.

I get it the first time you figure out that last tricky part and finally get that full combo feels great but get over it. Nnow that you finally did figure the song out it will happen more often until it becomes just normal (I'm just talking about "regular" maps not ones that are just insane with 10 blocks per second getting full combos there is quite the achievement.) At this point at the latest you should focus on getting bigger and bigger swings in which will require you to get faster and smarter about the paths your sabers take and your score will reflect that and you will pass your friends.

About the people putting FC in the title of their videos, it's nothing but clickbait to draw more attention. If the maps would go to a million points I guarantee you they would be "OMG OVER A MILLION POINTS IN BEAT SABER PERFECT RUN." If you want some videos that actually wow people it's mixed reality ones with entertaining gameplay. Do you like to watch videos with a head bobbing POV recording where you can see the sabers only half the time and at the end screen look at the full combo text and think that's some good ♥♥♥♥ right there let's share it with my friends? I certainly don't. But for example Ragesaq with his Darth Maul setup is so far away from full combo you probably would be disgusted rakes in millions of views with his videos because those are actually fun to watch.

And should he actually get a full combo with that kind of handicap it definitely deserves everyones respect (even how good he already is with it deserves it) and you can finally stop whining about how hard it is to do it with two individual sabers and needing additional mutipliers and fireworks and whatnot for it.

EDIT: That madman actually did it: https://www.youtube.com/watch?v=XEK2UpPwjqs

Początkowo opublikowane przez McFC:

All of my FC's are outclassed by non FCs. Not just by a little bit, but 40,000+. That's disheartening when the goal of the game is to hit every note.

Playing with people is a crap shoot. My GF can flail her arms around, getting combos maybe into the 30's on a good song, then gets 20+k more than me on the final score despite me having an FC.
[...]
I play with people who have high score records yet never got an FC. They don't even know what it means. Besides, my FC records get buried beneath a list of non FC high scores eventually. It's a meaningless achievement atm and I hate that personally.
[...]
Endgame in this is [...]

You talk about fully beating a level and endgame but what do you even know about that? You're getting bodied by your "unworthy" friends to the tune of 40000 points difference or by your girlfriend making 20+ mistakes (if 30 is the max combo she has to make a lot) which is a ton of points she misses out on and still gets 20000 more than you? Do you know what will happen when once they get even more comfortable with the maps and reach a full combo? They will absolutely destroy your scores and the devs can't keep adjusting the scoring to keep you above them.

Początkowo opublikowane przez Tacobara Sushito:

Thanks for keeping things so civil guys! I was worried if the thread would just attract thin complaints but you're all being super constructive and thorough and that's very cool.

I'm sorry if that reply sounded a little too hostile but that very first quote really annoyed me being so factually wrong (if I'm wrong and he is right somehow please tell me I'd love to know how I can increase my score by missing blocks) and throwing in all these $5 words to sound smarter triggered me a little. But I have an opinion on the topic as well. I don't expect anyone to read that wall of text above so I might repeat myself on some stuff.

The only rhythm game I played before this was Crypt of the Necrodancer which is pretty different from other rhythm games since it doesn't tell you what to press just that you have to do one input per beat and you plan your own movement to deal with the enemies. So I don't have any expectations about how a rhythm game should be scored I can just say what feels right to me for this game. I've played for over 60 hours by now and I think I've got a good enough impression of the scoring. Overall it is pretty good but there are a few things that in my opinion would improve it.

1. The minimum angles to get the 100 points for the swing don't need to be divided between before and after the hit at least I don't see why they designed it like that. All it does is limiting your upswings unless you overextend your wrist or swing from behind your back to reach that 90° before angle.

2. You all know these quick and long alternating up-down combos. For the most part in reasonably designed maps you can score pretty well using your arms alone but in these sections you can get a lot out of adding a lot of floppy wrist movements. Which can be really encouraging for very competitive people to add a lot of strain to your wrists that it's potentially damaging. And even if you do it it still produces a pretty random result. I played a lot on the $100 Bills expert one saber map while testing some unrelated stuff and in that beginning part until the first obstacle appears (around 40 seconds) I had scores ranging from 71000 to 82000 despite not missing anything and playing virtually the same and that's just unfair since it comes down to luck if you get it just right.

I thought about this for a while and I can't find anything better, not that I could implement and test it anyway. Replacing the angle with speed or distance your controller moved is in my mind putting short people or rather people with short arms at a disadvantage. Because of that lack in reach they have to work harder and use bigger angles to get the same speed and distance. This is also somewhat exploitable since the way you hold the controller can give you some extra reach. Maybe I'm overthinking this and it and the differences would be too small to notice but as I said I can't test any of it.

But the other day I was playing the song "Lost Woods (Ephixa Remix)" and It has patterns in it that actually feel very satisfying to do just out of your wrist and I thought this doesn't need to be removed as a whole. It's enough if it was just not better than what you can do with just arm movement. You can still have the choice to do it but the game at least doesn't reward hurting your wrist anymore.

Lowering the total angle that is required for the 100 swing points is probably enough to achieve that and I really like the idea of adjusting it dynamically based on how many blocks are coming at you at the moment. So in these super fast parts it goes down to something like 60° or even less and if the blocks are more spaced out it stays at 150° like it currently is. This would also help a lot to make the scoring work with pretty much all kinds of maps and be more interessting than just counting your good cuts.

3. While doing those calculations above I realized that missing isn't all that bad I mean it's still a restart for me but looking at the leaderboards losing roughly 4000 points isn't that much in terms of ranks so for a start I'd have mistakes drop your multiplier completely instead of just one step back to x4, x2 and then x1. This will double it to almost 8000 which makes makes it more impactful. The top 10 are still quite spaced out so it won't do much in there but it would be the difference between rank 10 and somethere in the 30's which seems ok to me. Adding a stright up x2 to your score for a full combo is just way too much.

TL;DR
start reading after the last quote.

Still TL;DR
Three changes:
1. no distinction between angles before and after the hit just a total angle
2. reduce the required angle to make additional wrist flicking unnecessary maybe even dynamically
3. mistakes drop you to x1

[Torus]

I honestly like the scoring system not focusing to much on timing an encouraging big arm movements. I feel badass this way with big reaching motions. If it would be more on timing, I would use the controller more like precise laser pointers only touching the arrows.

[ZbuffeR]

Torus, I came here to say exactly this, +1 !

I like the current scoring favoring big cuts instead of precise timing.

[564.3]

Anyway, the main focus of the game will keep being movement.
They talked about that subject in many interviews, and already spent a lot of time trying many algorithms. They tried timing and it felt quite random, then accuracy and people became very static/robotic.

The latest interview talking about that was ResearchVR (https://researchvr.podigee.io/75-researchvr69 at 8:30). Another interesting point is that the slashing sound is played on the defined beat no matter what we do: it's dynamically changing as needed.

[nul]

Początkowo opublikowane przez syne:

Still TL;DR
Three changes:
1. no distinction between angles before and after the hit just a total angle
2. reduce the required angle to make additional wrist flicking unnecessary maybe even dynamically
3. mistakes drop you to x1

I did read this whole post, but just quoting the part that matters without most of the rambling.

These changes are completely agreeable. I've noticed an odd trend recently in games where multipliers aren't completely resetting on mistakes and it just feels like cheating.
Someone who can't keep up with their chosen difficulty shouldn't be rewarded. I personally am no expert at the game. And yet I feel like I was being given scores way too high for how I performed. Seeing an 8x multiplier despite whiffing several times throughout the song.
I also don't think the multiplier should go 1 > 2 > 4 > 8.
1 > 2 > 3 > 4 > 6 would be rewarding enough for good players, and with that 3x there to fill in, it's not overrewarding someone for doing 'okay' at slower parts.

As for the actual score you get on the block, I'm completely baffled by the requirement for a 'perfect' swing. I'm all for encouraging more movement, that's the point of buying a VR game to begin with. But before poking my head into this thread, I attributed 'more movement' to faster pace.
I like the approaches to change scoring to make wrist flicking unnecessary rather than making it outright not work. I think the whole dependancy on 'wrist flicking' is pretty silly too. There's games like osu! and stepmania or whatever if you don't want to be swinging your arms around.

Maybe instead of making it 'dynamic', something similar to the 'white notes' of GH/RB could be implemented. 'Soft' blocks that don't require such a strong cut for higher scoring, that usually follow after 'hard' blocks.


Edit:
And to add onto that first and second point...

When you have a V^ or a ><, this means you have to swing at a total of an 180 degree angle repeatedly, then 150 on the last block. Because if you don't do a 90 degree 'follow up', then you're not doing a 90 degree 'pre cut' on the next block.
Sure when they're a bit spaced out it's fine. But almost every song has a point where it's not spaced out.

[xenicat]

An alternative block would work but I see some issues with it.

It's additional work for every map that will be created and creating them is already tedious so I think a lot of people won't bother. Right now very few maps are refined enough to be really fun to play and a lot less will be optimized with scoring/your wrists in mind. All current maps would also have to be redone for it to have any effect.

I would prefer a change to the scoring algorithm which is a one time effort (per iteration) for the devs, will automatically work for all current and future maps and thus be easy to test and compare if it feels better.

Początkowo opublikowane przez nul:

Seeing an 8x multiplier despite whiffing several times throughout the song.
I also don't think the multiplier should go 1 > 2 > 4 > 8.
1 > 2 > 3 > 4 > 6 would be rewarding enough for good players, and with that 3x there to fill in, it's not overrewarding someone for doing 'okay' at slower parts.

I like the multipliers as they are right now and not just because I like powers of two :D Every tier is a significant step up from the previous one increasing your score by 100% every time. Having 1 > 2 > 3 > 4 > 6 is in my opinion too close together and underrewarding for good players.

Seeing x8 all the time right now is due to the fact that it only takes eight blocks to get it back and if you would drop to x1 after a mistake it's still just 14 blocks until you're at x8 again. Instead changing the multipliers itself you could increase the amount of blocks it takes to get to the next tier. Currently it's two, four and eight blocks to go up to x2, x4 and x8 respectively. I'd try eight blocks for each or just doubling them all. Then you would need 24 or 28 blocks to reach x8 which is longer than most slow parts so it really is earned to reach the highest multiplier.

But I think mistakes shouldn't become too punishing either. It's a good thing that it's possible to close the point gap with more effort in your cuts.

[Nagorak]

Początkowo opublikowane przez 564.3:

Anyway, the main focus of the game will keep being movement.
They talked about that subject in many interviews, and already spent a lot of time trying many algorithms. They tried timing and it felt quite random, then accuracy and people became very static/robotic.

The latest interview talking about that was ResearchVR (https://researchvr.podigee.io/75-researchvr69 at 8:30). Another interesting point is that the slashing sound is played on the defined beat no matter what we do: it's dynamically changing as needed.

Maybe focusing on timing felt random because there's so little feedback on whether you're actually hitting on beat? In DDR/ITG you immediately see what your timing was like after each step. If you're off beat on the highest difficulties you'll get a chain of "goods" (or worse) and you'll know to pick it up to get back on beat. When playing you don't really even realize how much you're paying attention to that, until you play on a bad pad and an errant "miss" throws you totally off as you subconsciously try to adjust to it.

The game needs a lot more feedback in general. Right now everything can seem basically fine, and then you miss a couple and suddenly fail "out of the blue". Your life bar needs to be more visible, and the screen needs to change somehow when you're on the verge of failing, so you know that you need to really focus and try to get your act together or you're going to fail.

I know that probably sounds a bit off topic for the scoring, but it all sort of comes back to the same issue which is we need to be able to see how we're doing in real time, to then understand what we need to try to change or do better at. Having the scoring more timing based should be viable, but not when you can't even tell if you're on beat or not.

[564.3]

Początkowo opublikowane przez Nagorak:

Maybe focusing on timing felt random because there's so little feedback on whether you're actually hitting on beat? In DDR/ITG you immediately see what your timing was like after each step.

Yeah, that's what they tried in dev builds because that's what other rhythm games do.
But they couldn't figure out how to decide the timing is good in a collision between a sword moving freely in 6DOF and a rounded 3D cube, instead of the very simple and clear binary button pressing.
Nothing seemed to work reliably, so they dropped it for a general "it's hit", that already required some leeway and tweaking to feel right.

[Balbonious]

The current scoring system also sucks for anyone using WMR headsets. Inside out tracking works beautifully for this game... until you ask the user to keep doing massive swings and sending the controller out of field of view regularly. Most of the time it works fine, but failing a song because you lose tracking is the absolute worst thing that can happen in this game.

[Kyle 1413]

Couple of points
1. Timing wouldn't be intuitive for scoring in this game imo. Timing is already impliciently necessary to hit notes properly, and would be difficult to be precise with.

2. Scoring system scales just fine on well made maps, even of extreme difficulty. Top score on Blue Zenith, one of the harder songs in the game, is less than a percent away from SS.

3. The scoring system gives you plenty of freedom once you are familiar with it. I've seen SkyKiwi, one of the best players in the game, maintain over 95% accuracy while dancing like a loon playing Limbo

Comparing Beat Saber's scoring system to other rhythm games doesn't work, because Beat Saber is fundamentally different at it's core. Most other rhythm games are on a 2d plane, so timing and accuracy makes sense for scoring. Beat Saber however uses a 3d space, and you use your physical body to swing. You have lightsabers. It rewards you for using them like lightsabers.

[DuWhen]

Necroing here, but also the hits are way too lenient. Load up 100 Bills on easy, then hit each note in the correct direction, but a single block over, and you'll see that the game still counts it at a hit. The only downside in doing this is that you lose the 10 potential points for an even slice, but your combo continues. This means that you can FC songs like Flower Dance on Expert+ way easier than you really should be able to, because instead of tracing an accurate circle through the blocks going up and down, you can just swing in the middle of them, even if there's a gap in between the blocks.

[Ionizer]

New player here, great game but the scoring needs some love.

The major issue I have with the scoring is I want to know how many notes I hit.

I don't really care how many 90 degree perfect swings I got, I want to know how many notes I missed.

Especially for custom songs like Take On Me. I have no idea if I am doing better or not because it doesn't keep track of accuracy.

It keeps track of my best "combo" which kinda does that but not really. I don't care how many hits in a row I got, I want to know how well I am doing over all.

Please add that. Just one more number, give me an accuracy count.

Thanks.

Great game, as I said at the beginning. :)

[Tenor Sounds]

Początkowo opublikowane przez Ionizer:

Especially for custom songs like Take On Me. I have no idea if I am doing better or not because it doesn't keep track of accuracy.

They really should save it per song so you can check it later, but for what it's worth I'm pretty sure you are able to see [notes hit/total score] after you finish a song. Not perfect, but you can note it down and keep track of your improvement manually.

Don't forget about improving your score/rank too, it'll help build up strength and coordination for any future songs you do.

[Coriarty]

Man you nailed that one on the head. Counter-intuitive is the best way to describe it. Spent a while in game slashing notes w/ the advanced HUD trying to figure out how the scores were generated. Ended up having to look it up. Very disappointed to say the least.

[Tenor Sounds]

Początkowo opublikowane przez Coriarty:

Man you nailed that one on the head. Counter-intuitive is the best way to describe it. Spent a while in game slashing notes w/ the advanced HUD trying to figure out how the scores were generated. Ended up having to look it up. Very disappointed to say the least.

Yeah, the tutorial really needs to go into detail about the scoring system. Right now you have to look it up and that's tedious and bad design.