Taliyah post-release follow up

So I think what you may be feeling here is that the W has a spellcast lockout until the eruption. Here's my thinking.

W is not a vector cast spell. I'm trying to avoid the novel input combined with time pressure effect that makes vector casts near impossible to learn. I want players to have as much time as possible to press the W button a second time, so I'm giving them all of the 0.85s of the W's delay to press the W button a second time. If I allowed you to cast another spell before the eruption (say after you pressed W a second time), I would re-instate the problematic part of vector casts, which is that you have to input the button down and button up as close to each other as possible.

So you may ask why not just allow it as an option?

Here's one of my favourite design mantras:

Optional power is mandatory

A designer can convince herself as much she wants that this mechanic she designed is an option that players can use if they want to, but that's never the real world a competitive game will be once the mechanic is mastered. If you gain power (and being able to cast another spell sooner is power) for doing an optional thing, that thing will become mandatory. If I said "hey you have all the time in the world to input W2, but you can only cast after you've done it", I might as well be standing behind your chair yelling "HURRY HURRY HURRY" into your ear.

Here's another reason:

Lane games are best when both players play

When both players can outplay each other, a lane becomes a true display of skill [tm]. If Taliyah could cast E after beginning her W cast, the correct play would be to never cast E unless you saw that a W was going to land, at which point your enemy can't really do anything anymore. When you have to cast E first, the enemy is put on notice: Taliyah is looking for a W now! Be vigilant and dodge for 4 seconds, in which case you come out ahead because her E is incredibly expensive to cast. I feel this way we get the fairer game and more interactive lane action.

Hello Vanirru, and thank you for your post! I’ve been waiting for someone to mention this! This is exactly the kind of feedback post I was hoping for!

Let’s start at the end of your post: E maxing! I only have one argument here: Faker maxes Q. CHECKMATE ATHEISTS! (From what I can tell he’s played one ranked game on her, maxed Q, went 4/5/9 and got carried by his top lane Fizz. GG)

No, but seriously: I like that E maxing can be situationally good! For instance if you lane against the likes of Yasuo or LeBlanc: champions that are likely to repeatedly try and approach you with a dash. However, if you have to E max in lane you will most likely be weak come the first couple of objective fights. So much of Taliyah’s damage budget is in the Q. At rank 5, hitting all 5 Q missiles deal 420 + 1.2 AP. Every 4 seconds. That’s absurd. So I will continue to buff the lane cases of the Q until Q maxing seems the no-brainer, while leaving E maxing open as a situational counter.

Speaking of the strength of the Q! Before we move on to the meat of this post, I should point out that this Q is STUPID. Holy shit how did they let me ship this? Once you have Rylajs and reliably hit all 5 missiles, this thing just straight up murders people. The only reason I got away with shipping this is something we call gating. Normally when we refer to gating, we’re talking about cooldowns or mana costs--things that act as a gate to your usage of the ability. In Taliyah’s case, the Q is uniquely gated by ground-as-a-resource. Not only was this a thematic home run, it also provided us with a really cool mastery point for future Taliyah mains as well as a soft “play around this” telegraph for the enemy. Success!

But as you correctly point out, that gating is painful and harsh. It has to be to be worth anything. Let’s talk about size and duration of Worked Ground first for a moment, before we dive into a discussion of sharpness.

To be frank up front: we fucked up on the duration of Worked Ground. Playtesters learned way too well to play around Worked Ground in iterations where its duration was much, much shorter, and we kept upping the duration until we saw actual constraints. We ended up with a number that was way too large because it was arrived at in an eco-system where Taliyah’s players had already mastered the ground mini game. With a number as big as 180, we never gave players on live a chance to even learn the intricacies of the game. I’m of course not sure at all 140s is the correct number, either; there’s a chance we need to reduce it more.

Here’s what I strongly believe we shouldn’t do, though: reduce this number to where you can use fresh ground Qs more than once from the same place in one teamfight. Strategically it doesn’t make much sense to ask players to memorize the locations of Worked Ground in between fights, but tactically? I think this is a really meaningful gameplay element. You can go back and forth in a fight, and if you manage to push Taliyah back onto ground she’s already used up, you have the upper hand. That’s exciting and new! If Worked Ground refreshed mid-fight, that’d be gone.

Similarly, the size is carefully balanced such that you can get about three fresh Q casts side by side in lane, but never more than one side by side in a jungle passage. This is a cool self-balancing factor: Taliyah is massively more powerful by her nature in choke points (because of her AOE and her area denial; also if you can’t dodge to the side, you will eat all of her Q), but the Worked Ground mechanic balances this from the other side: she’ll be more powerful in the jungle for a short time, after which she has to press forward.

So TLDR: Could see us reducing duration further--probably not much below 90s though (because you’re looking at roughly 50s at 45% CDR then, and anything below that risks becoming available twice in a very long, drawn out fight). Size is probably right once people have learned to optimize the spell and once we’ve found the correct duration.

A MUCH more interesting question is why don’t we make her Worked Ground Qs straight up stronger. You suggest letting it shoot 2 missiles instead of just 1, others have suggested giving it a shorter CD, or even making it straight up hit harder. I don’t think any of these changes are right for her, and the reason why is something we call “sharpness”.

SIDEBAR: Sharpness When we speak of a mechanic being sharp, what we mean by that is that there is a clear difference in states. Usually this is how we hope to achieve interesting and meaningful gameplay. Let’s use Poppy’s Heroic Charge as an example: Poppy charges at an enemy champion, carrying them with her for a short distance. If she hits a wall, the champion is stunned. Now we could imagine a world in which Poppy has trouble hitting Heroic Charges into walls, and maybe a well-meaning designer will say, well, Poppy has trouble accessing this power in her kit, why don’t we make it a little more accessible? Heroic Charge always stuns enemy champions, but the stun duration is doubled when they hit a wall. I would argue this is actually a worse ability. Now instead of trying to angle yourself in such a way that you get an enemy champion between yourself and a wall, you have to do the following math in your head: yes, I would get twice the stun duration from a wall, but maybe I can use the non-wall case three times as often? And isn’t half the stun duration enough anyway for my team to land follow-up CC? At the same time, the enemy is confused: Poppy charges at me and stuns me, but sometimes that stun seems to last longer? By removing the sharpness of “wall vs no wall”, we made the ability more muddy and harder to understand and learn for both sides.

Returning to Taliyah’s Q, we had a bunch of the suggested versions in testing early on. There were both versions where the difference between fresh and worked ground was that worked ground sent out fewer, but still more than one, missiles, and versions where the worked ground cast would send out a single missile that was just a bit stronger than the normal Q missiles; maybe about 1.5 times as strong. What happened was that people started getting confused about worked ground and actually had to track it more accurately as the enemy, because it was no longer obviously clear if a cast was a worked or a fresh ground cast. On Taliyah’s side, the mental arithmetic then became about “do I even need to worry about worked ground? Yes it’s weaker, but it’s also cheaper and still does enough for me!” This would get even more complicated if we started factoring in partial CD refunds.

Worked Ground Qs must be strictly worse than Fresh Ground Qs. They should have a fallback function of “cheap but weak harass or last hitting assist”, and I admit that they’re currently not living up to this. I’m hoping to ship a 50% damage boost to minions change in 6.11 to at least make it a good CS assist tool. Base damage changes are also not out of the question further down the line. Note also that Taliyah rushes Rylaj’s as a core component, so even if you don’t get the sweet, sweet full damage Q, you do get a slow, and who knows, maybe that allows you to land an EW combo.

To sum up: Taliyah’s Q is ridiculously strong (if players, like yourself, are noticing it’s powerful this soon after launch, it’s probably too strong; took players months to realize that Lucian’s ult is absurd). We’re buying that power with gating, but gating must be sharp and meaningful.

Thank you for bringing this up! I hoped someone would!

I'd like to preface this by saying that I will explain in this post why I very consciously decided not to use vector casting on her W. I hope I don't come across as defensive; do note that I didn't just decide this on a whim. I had to defend my decisions against many other designers playing devil's advocate for vector casting, I had to get leadership on board, I had our Insights team run hyper-focused user labs on this question, and there were many late night sessions where my arguments (see below) were beaten up by my peers. The reason Taliyah shipped with this W is not because Daniel Z Klein hates vector casting (although, guilty as charged, I do.) The reason she shipped with this W is because I strongly felt we could make a less punishing, more learnable version of vector casting, other designers agreed, and this W is my attempt to make it.

Also apologies that this is a little lengthy. I am very, very passionate about this video game nonsense and have spent the last 10 months obsessing over this and other similar questions. There's still a good chance I'm wrong, but in the interest of an open and transparent question I wanted to lay out all of my thinking. Grab a beverage and enjoy!

Viktor’s Death Ray and Rumble’s Equalizer use a casting paradigm we refer to internally as “vector casting”. You can think of a vector as an arrow with an origin point, a direction, and (unused in this case) a length. The way you use vector casts is either by pressing the ability button and then holding down left mouse button at the origin point and dragging and releasing left mouse button when you have the direction you like, or, if you’re using Quick Cast, by holding down the E button, moving your mouse cursor, and releasing the E button.

There are two subtleties to consider here:

1. This is the only ability type in our game that asks you to not only think about keeping a button pressed (channel while moving abilities such as Varus Q do this too), but also to think about where your mouse is when you start pressing the button. This isn’t good or bad in itself, just unique.

2. The effect, including any pre-cast delays (such as Rumble’s cast time or Viktor's laser's missile travel time) begin to happen only when you have finished all the above inputs; that is to say, only when you release your R button or your left mouse button does Rumble’s ult cast time begin.

The combination of a novel input type you’re not primed for anywhere else in the game (Point 1) above) with a cast sequence that tells you to hurry up with your second input (Point 2) above) makes for the hardest type of ability in the game to learn and master. From personal experience, I do alright with Viktor lasers in lane when I’m just last hitting / hoping for harass, but once the pressure is on and an enemy jungler is breathing down my neck I completely lose my shit and shoot lasers off into obviously bad directions / from bad places / fail to shoot them at all because I get so flustered.

It’s not just me, either. (This is where I pull out some data. Please don't think of this as me hitting you over the head with numbers; do argue with my interpretation of this data if you like!)

Both Viktor and Rumble have very low pick rates considering their roles and apparent power levels (although caveat here being that correlation doesn’t necessarily imply causation, and lower pick rates automatically mean higher win rates due to a self selection effect). They also have relatively steep mastery curves considering that in both cases, the vector cast ability is the only truly mechanically demanding part of their kit, and most importantly, they exhibit a massive amount of player churn. That last one means that a very large percentage of players who try out the champions do NOT stick with them. This compared to other mechanically demanding champions such as Yasuo, who has some of the LOWEST churn in the game. Let me show you some graphics (you'll have to click on these to see the full version)!

http://i.imgur.com/NcddmZe.png

These are the mastery curves for Viktor and Rumble. The yellow into green line is their average win rate over games played previously; at the left end of this curve is your chance to win your very first game on these champions (41.6% and 43.3% on Viktor and Rumble respectively) and at the right hand of the curve is the chance to win your 80th game (I stopped at 80 because, as you can see from the jitter in Rumble’s curve, data was getting rather sketchy. There aren’t enough players who’ve played 80 games on Rumble, which is sort of my point).

The grey curves at the bottom are sample sizes. Leftmost is amount of players who played the champion exactly once, rightmost is amount of players who’ve played the champion 80 times. The very sharp spike to the left is something we generally associate more with a younger champion (Aurelion’s curve is all spike right now). Ahri, who came out one slot before Viktor, has a super flat churn curve, with a gentle, flat hill to the left instead of a spike. But Ahri’s curve is much, much less steep than Viktor’s and Rumble’s (she goes from 48.6% to 54.5%, about 4% less growth over games played). So let’s compare to a champion with a steeper curve (again remember to click to open):

http://imgur.com/DFgaV1P

This, as you can probably tell, is Yasuo’s mastery curve. He grows from 40.7% to 52.3%, roughly 2% more than Viktor and Rumble. But look at the bottom! Is that Ayers Rock? No! It’s Yasuo’s curve of Previous Games Played. Notice the absence of a spike, and even notice how gentle the slope is. People pick up Yasuo and stick with him.

The point here being, my working theory is that vector casting is a very hard to learn input paradigm that causes players to churn from a champion. The two points we discussed above (hard to learn input type that nothing else in the game teaches you + YOU MUST HURRY) sufficiently explain to me why this would be. While Yasuo is one of the hardest to learn champions in the game, all of his inputs are bog standard: place your mouse over the dude and press E to dash through him. Place your mouse over where you think the dude will be in 0.1s and press Q to point-blank skillshot him. Place your mouse where the bad guys are shooting from and press W to make some waist-high cover/a windwall.

So when I made Taliyah W, I let these two considerations guide me:

1. This obviously has to be a two part input. Let’s not hurry the player along, so they have time to learn.

2. If possible, avoid novel input types. Can we piece together “point of origin plus direction” from existing inputs?

Turns out we can! W1 (the first input you make in Taliyah’s W) is a vanilla AOE spell input (you place a ground circle), W2 is a skillshot from a proxy target--a little less vanilla (Lulu Q and Taric E do it, as does Zed Q from his shadows) but still just a simple button press. We now have two distinct button presses on a spell that already has to have a delay baked in (to give the enemy a chance to dodge this very powerful effect). This gave me the option to say, hey, Seismic Shove will happen 0.85s after you press W1. You have ALL OF THAT TIME to input W2, and how soon you do (or how much time you take) does not affect gameplay at all. Take your time. This makes Taliyah’s W more learnable. You can actually take your time, and you can even move your champion around during the delay if you need to.

Additionally, if for whatever reason you fail to input W2, you get a partial success case: you get a knockup at the location you placed W1. Often that’s not optimal, but it’s clearly better than nothing.

So to wrap up, the reasons I like Taliyah’s W over vector casting are:

1. You can take your time to learn the ability because we don’t rush you

2. The ability consists of inputs that are already familiar to most players

3. There is a graceful failure case where you get a knockup rather than nothing

All this to say: I hear you, and I'm sure there are players out there who would do better with Taliyah W as a vector cast, but the vast majority of players would churn off her very quickly considering how mechanically demanding and frequent her Q is as well.

Again, I may be wrong, but these are my reasons.

Hey bee's knees! I answered the W question re: vector casts in this post:

http://boards.na.leagueoflegends.com/en/c/developer-corner/5IpizsoF-taliyah-post-release-follow-up?comment=00230000

Please note there is a huge difference between listening to feedback and defaulting to agreeing with the feedback. I am absolutely hearing what you guys are saying, and to a degree I was expecting this reaction. But because of all the reasons I lay out in the above post, I still respectfully disagree. Your assessment that average players can learn vector cast in 3 games is also contradicted by the data: as a matter of fact, almost all players decide after 3 games on a vector cast champion that they do not want to learn this and would rather play another champion instead. (Again, see the data in my post linked to above).

I went into the vector cast issue here: http://boards.na.leagueoflegends.com/en/c/developer-corner/5IpizsoF-taliyah-post-release-follow-up?comment=00230000

Could you read this and report back to me if my reasoning makes sense to you?

More range on the passive is absolutely an option! We played with a version of pathfinding for Taliyah that preferred walls, but it turned out this was very hard to do for very little actual gain.

I think it's absolutely fair that it's too easy to play against Taliyah right now. I would be very worried if this wasn't the case. Her counterplay is universal: dodge these missiles, stay out of this zone, dodge this circle. Her play is unique. Taliyahs are at a huge disadvantage right now, but as you learn the champion I'm positively convinced that you'll find even this version right here is way too powerful. Check back in a year and I can look like a prophet ;)

[{quoted}](name=Fondling Gems,realm=NA,application-id=A7LBtoKc,discussion-id=5IpizsoF,comment-id=0006,timestamp=2016-05-24T21:24:23.142+0000)

Her passive feels kinda tacky. Due to her ult it makes roaming without it feel lackluster, but her ult range means that free movespeed doesn't do much for her. Sure it adds up if I am running to bot lane, but I am not running bot lane, I am only walking 1/2 way then ulting.

On top of that it takes so long to start after combat it is generally useless for escaping or chasing after teamfights.

So I'm going to contradict your first point: it is precisely because both this passive and her ult are on the same kit that they work. Having the passive on the kit means she actually has a realistic chance to get within R range in time; I couldn't possibly give the R more range at rank 1, and the passive is what makes this work.

The passive also strongly signals "roam", and the only way I could give it enough MS to be an effective roaming tool is by sharply gating it in combat. There is no way I could give a disruptor (think zone control mage) even a quarter of this MS in combat). Whenever she's even trying to do a dangerous thing (even fishing for a Q hit), she has to be dropped out of it for a meaningful time (I stole 5s from mobi boots; felt like the appropriate amount).

Look at it this way: I could do the things you suggest, but maybe she'd only get a fifth of the MS then. You'd barely feel it. Constraints buy power.

That's great feedback! I'll forward it to our particle artist (who will probably tear his hair out; he spent WEEKS finding the right balance between "I can see this thing" and "this thing doesn't make our map look super ugly"). Does the highlight when you step into the circle help you at all?