PonPonWeiWei

To be honest a good chunk of it is also interpretation. It is difficult to see something like "has energy" and meaningfully conclude what that actually means. Internal terminology or leaked keywords can often be leveraged in different than expected ways.

It's always good to keep a skeptical eye on anything coming from datamining, not because it is right or wrong but difficult to interpret and often out of explicit context. (And sometimes we do just change our designs!)

I just tested this and I couldn't reproduce. Can you send me more information like what other gods were present/what specifically hit you?

It does seem more responsible to being double tapped and cancelled, but I want to make sure there isn't something more going on.

Mostly a matter of the frequency of a absolute new player to Conquest playing Ranked Conquest is uncommon. Additionally any requirement can be bypassed given time. Someone having an account in S4 playing Conquest and then jumping in late S7 has a lot of learning to still do, but they can queue.

You may want instead a "X recently played" but this is a decent undertaking for something that we know statistically doesn't occur often. Could be discussed more though.

If you want a Matchmaking system that does the two following things :

1. Distributes players based on their skill to provide fair matches.
2. Attempts to maximize the speed at which is does this.

Then yes, this is a problem that cannot be fixed. We can tweak 1 to maybe start players based on some other skill (Their casual conquest performance for instance) but that adds noise into the system that will potentially provide more accurate initial placement but noise gets distributed throughout the curve lowering quality overall. We can tweak 2 to slow people down, but this means people will be at their unintended MMR range for longer. Rather than being pretty confident after 10-20 games and being very accurate at 35-40, you delay pretty confident to 35-40 and very accurate to 60+.

When a player enters the pool we don't assume they are average, instead their MMR really represents our confidence. We have 0% confidence we know their skill so we plac...

Bear charge can now execute, as being charged by a bear is bad for your health.

<3

Machine Gun Ra. You would utilize the bonus Power from the heal and the slow from the 2, slap an Executioner on him and you were good to go.

Hel also could Basic Attack cancel on stance switch, leading to it being a common strategy to just double hit enemies in lane.

Gem of Isolation slow on Bastet Bleed / Thor Mjolnir. This was never good but was funny.

All Basic Attacks on every melee god is a frontal cone. If we have a longer range it will increase the size to the left and right that you can hit.

Awilix has a similar "hitbox" type interaction. Through SMITEs history we found that giving this leeway on Basics makes them feel more responsive and correct most of the time, versus the alternative.

(Fun note, I actually broke this on a god during development to where it would work more like what is "expected" and wow, it was awful.)

AJ went more into detail here : ...

That's a trickier one we have discussed. I can bring it up for discussion again. She has warning FX associated with it that may prompt users to burn X tool to try to respond. On interrupt she would get to use it again now on a target who now lost that tool. Especially for something coming over a wall, imagine using Aegis as you know you are about to die to Da Ji (you wait til the last possible second) and it doesn't come, only to suddenly happen again 1s later. There is often a feedback issue due to the abilities function which prevents the person being affected from knowing why or why not an ability went off.

This argument can be expanded out to other abilities where we give the cooldown back but they often lack the hard warning FX and common situations that Da Ji creates. Its a situation where Hard Rules vs. Use Cases create some murky waters.

This is actually getting fixed next patch. She currently has a .2ish second warmup before she actually gains any of her benefits, but when interrupted it goes on cooldown.

We are changing it so that she gets her benefits during that .2s warmup. Should be more consistent in providing protections to Baba and prevent cases like this where it goes on cooldown without providing a benefit.

Š̶̬͉̼͍͛́̒̌̏̽̍̽̈́̔͜͝o̸͖͖̝͓̟̭͊͌͋̅͋̔͘ͅo̵̧̡̙̝͚̬̙͛̄̾n̵̤̻̙͙̠̣̫̿͋̂̑͆̅̑̍͗͊̓̒̕͘͠,̶̛̪͈̖̙̩̟̦̃͜ ̷̡̧̨̤̲̗͔̯̣̮̪̍̌͑̈́̈́̑̃̍̓̓͐̕̕͜͝ą̵̢̰͇̯͔̲̙͇̲͍̆͜l̶̨̫̘͓̩̝͖͔͈̯̲̭̳͚̎̂̅͘ͅl̵̛͙̩̲̪͉̱̳̮̗͔̙͕͑̎̿ ̷̗̻̎̏͊̓̇̀̐̏͐͝w̵̧̗͉͍̱̣͍̦̣̠͋̓̊̈̿̀͂̅̈́̌̀̏̈́ͅḯ̴͕̭̬͔l̵̹̘̖̩̰͕̹̟̠̪̗̖̠͌́ͅl̸̢̢̮̪̲̭̪̘̹͗̑̒̂̎͝͠ ̷̛̮͊̌̄͑͛̍̑̈͆͛̚͠b̵̼̦̘̠̝̗̂͌̌̌͘ę̶̼͆̉͑̇ ̶̹̠̫̠̝̾̋̌̆̅̐͛͒̐̏̈̚Ḵ̸̢̡̣̖̲̺͔͙̱̼́̌̀̾͑͆̓͝p̸̨̢̢̢͎̭̗̙͋o̴͙͙̺̼̭̘̝̭̘̙̔̌͊̒̉p̵͕̎̍͛̄͐̊͘͠͝

For Baba's Cabin it is intended to be a blocking collider for player movement. It creates some odd scenarios with hiding inside of it otherwise, especially non-player entities. This is true while stationary and while moving.

It is not intended to block abilities/attacks, those should pass right through.

I bundled it into high level competitive use case as ranked ban rate and SPL often have parallels but should have been more specific. Ban Rate is a major stat we look at.

Because it doesn't always work that way. Things that players value and statistical win rate don't always align. For example, Hades has been basically top 10 for years at this point. Not a modest amount of time, years. If we nerfed him I don't think people would be fully on board with it.

We tend to evaluate based on win rate (with supplemental stats), popularity, and high level competitive use case. Mattene is basically correct. Nike was generally perceived as underwhelming, seeing a low use at high level and struggling to see play elsehwere even though her stats were solid. Buffs tend to have anywhere from 2 months to 2 weeks of buildup time before we announce them. Perceptions can shift during that time and when they shift as we get very close to a patch it can become tricky.

This often comes down to a debate between Win Rate being a "valid" stat or not. Some feel Win Rate should be the only metric, while others see it as a relatively meaningless stat. The truth i...

No problem :)