Feature: add Pings to Telemetry

[schneefux]

I'd like to request adding in game pings to Telemetry, because it gives a better view on player's personalities and takes analysis of in game behaviour to a whole new level.

[mrpranavgrover]

I'm a bit concerned with this, as the data can be easily misinterpreted. For example, a typical example of toxicity is rapid question mark pinging. However, question mark spamming is also often done to afk-ers, which isn't really that toxic as no one is on the receiving end of the frustration, and the player's frustration is understandable and should not really be punished. Another possible example of a thrower would be someone taunting, standing out in the open, and dying. However, this could also be easily mistaken by someone using a taunt after securing an ace and dying to a turret shot, or someone taunting the enemy team as the Kraken is about to destroy the Vain and dying as the game ends. Interpreting this data to find someone who is truly toxic in game would be extremely difficult. However, it would be more conceivable to interpret the current data of builds, as someone who solely buys scout traps and contraption the entire game to litter the base with traps is unapologetically trolling. I certainly like the idea of having some way to identify players with poor behavior, but using pings to view this could be misinterpreted and thus abused.

[PierreAndreis]

I dont think this info is useful in any way

[schneefux]

It is just data, data itself cannot be toxic or evil. It's the developer's choice to label the source of data as "toxic" or "bad" behavior, and I agree that it has to be done very carefully, or even better it should not be judged at all.

I have a use for the data, that is why I opened this issue.

[kvahuja]

@mrpranavgrover - sure it can be misinterpreted just like any data can be. The point we (shutter and I represent vainsocial.com) is that we believe we can model data and use it and drive behaviors. It's about the modeling of data that will bring it power. If someone uses it incorrectly on their site that's should not mean data should not be exposed. We fully understand what you say here and we will make sure when we model it we take care of it. Your inputs definitely valuable.

@PierreAndreis - that's fine if you don't want to use it. that doesn't mean we can't ask for it :) however i can debate your fact. Pings exist in game for a reason, they are supposed to be used and their is an effectiveness scale. Let's take an example (and here @genexp is our case for asking this) - I am captain and see my Carry Die in lane, I see SAW and Ringo pushing hard on turret, I ping 3x for my jungler to reach the lane. Now 2 things can happen - Jungler ignores or he responds. In each case we may or may not loose the turret. But if a jungler doesn't respond and we loose a turret that's a behavior which is clearly identifiable. Now take it a step forward. If the junger was weak, or was farming enemie's jungle or fighting enemy junglers we can proably see why. So just as Pings are useful in game, we are confident they are useful in Telemtry too. given we are focused on Social element it's even more important for us.

vgpro only focuses on Pro players and i see why it may not be relevant. we are targeting a very different user group hence the ask.

[mrpranavgrover]

@kvahuja @schneefux I definitely find what you're doing is interesting, and it seems very cool. My initial impression was that you would want to use the pings to identify toxic players, which was an assumption I should not have jumped to so quickly. I'm just concerned that a site may use this data to flag a player as toxic, which could easily be incorrect information. Also, everyone uses pings somewhat differently. I expect it would very difficult to automatically produce consistently meaningful data, which can lead to some amount of misinterpretation and, consequently, misinformation. However, if you feel as if you are capable of overcoming that challenge, I wouldn't mind seeing another feature being added if it is beneficial to the devs/community.

[BrianGLHF]

Since no one uses pings in the same way if they even use them at all, it seems that what @kvahuja and @schneefux are looking to accomplished could be done without using specifically identifiable pings, but just the telemetry that pings have happened and the location. Keeping [ping type] fuzzy [but maintaining the location specifics] would prevent the data from being used in ways that would be considered witch hunting while at the same time still giving some insight on how players use or reacts to them.

Also don't forget that adding this functionality for pings, we would also need to know wither a player was muted or not. To keep this data from being used in a witch hunt, mute issued information would also have to be fuzzy and not specific.

[idmonfish]

I think specific pings you give more rich data - ie could tell you how directive a Captain is and even if their shot calling is good (ie pings target on gold miner at sensible times (enemy have all just ported home for example.

In terms of it being used to falsely flag accounts as toxic - well thats a matter of getting the algorithm right isn't it?

[kvahuja]

exactly @idmonfish - great example you give there. so it's all in interpretation model being good or bad. we feel once we have the data available opportunities to see how we use open up

[schneefux]

I want to debate the term "toxic" here, I would differ between bad toxicity and average toxicity.

Bad toxicity is when a player stops playing the rules of the game, i. e. the player starts to buy traps and places them all over the base. Bad toxicity is often accompanied by a massive amount of pings of various sorts, but not always. This is behavior that should have players flagged and banned.

Then there is average toxicity. Average toxicity is just any excessive or provoking sort of communication while the player still follows the rule of the game. In effect, that means even though the player ragepinged me for dying, we can still play as a team and win together.

Bad toxicity can already be measured in parts using build information and kill/death participation.

Average toxicity cannot be measured at the moment, however it is an important part to the game. There is data about positioning available right now. Imagine the situation of a gank, people stand still for a while and nothing happens. Using ping information, we can build an algorithm to identify and judge the player's intentions here. There are different situations, where the toxicity comes into play: Humans tend to stress out under certain circumstances which will be expressed using pings, it can either end there and the player relaxes or it goes further which can lead to making mistakes or throwing the game. This is what I want to analyze at VainSocial, and this is why I think that the data has very high value to analysts.

[idmonfish]

Fully agree - and one of the things that causes toxicity is that people are very sensitive to percieving toxicity and vastly overreacting to that perception by ironically becoming increasingly toxic. This problem worsens at higher levels with players percieving inefficient play as "trolling"

So for example long ago I once started B on my Catherine and my jungler question mark ping spammed me for it. After we had won handily largely due to clutch roam plays by myself I added him to ask him what his problem was and he told me sheepishly he thought that me picking B first meant I would max it when A was the ability that Cath usually maxed first in those days.

[schneefux]

  {
    "time": "2017-04-22T15:16:58+0000",
    "type": "UsePing",
    "payload": {
      "Team": "Right",
      "Actor": "*Phinn*",
      "Ping": "PURPLE_QUESTIONMARK",
      "Position": [
        79.66,
        1.5,
        -0.91
      ],
      "TargetActor": "*SAW*",
      "TargetPosition": [
        0.0,
        0.0,
        0.0
      ]
    }
  },