Greedy Goblin

Friday, August 18, 2017

Unintentional newbie bashing in PUBG

Newbie bashing is a problem in every PvP game. It means that skilled players purposefully play against new or hopelessly bad players and slaughter them. This is also known as twinking or sealclubbing. This is very hurtful at the game as new players are facing those who are much better than them and have the only purpose of killing them. They usually die without having a clue what they did wrong.

The usual solution to newbie bashing is rating system: good players elevate on the ladder and are matched other good players while receive the rating as progression. A League of Legends platinum player won't be facing Bronze 5 baddies. This can be gamed by those who want to massacre newbies by running alts. I suspect that most streamers and celebrities in PUBG are guilty of purposeful newbie bashing. They intentionally thrash their ratings when not streaming, so when they do, they are facing 1200 rated newbies who helplessly line up for those glorious 10-kills streaks. If they would always play like they show it, they would quickly rise to the very top of the leaderboards and they are not there.

But PUBG faces a bigger problem: large scale unintentional newbie bashing. To understand what's the problem, we must understand the rating system. It's not easy because it's not officially published. However we can extrapolate from the point rewards. There is little reason to give lots of points to an activity that isn't considered hard, therefore gives rating. We can assume that the point you get at the end of the game is the same that is used to calculate your rating. I wrote down the points I got at the end of the games and plotted them vs my final position:
The average position point is 96. A killing blow gives 20 points, damaging full HP gives another 20. If we assume that 95 people die to other people and they lose 150% HP (damaged, healed, killed), the average kill points are 45. Which means that position has twice the weight of kills. It also means that if half of the team gets rating increase and the bottom half loses rating, you get to reach 130 points to go up. Which is #19 position without a kill or #24 and one kill or #70 and two kills.

Now the problem: if someone follows the dumb "git gud" mantra, he tries to get into action as soon as possible to "learn to PvP", which means that he dies often after either scoring a few kills or not. Since his position is low, he must score consistently 2+ kills to gain rating. Until he does, he sits with the newbies. If lots of people do it, scoring 2 kills consistently is impossible. I mean if newbies jump out randomly and "gid gud" players land on the hotspots only, they are facing equal "gid gud" players which means 1:1 K/D, which means 1 kill per game. Sure the survivors of the hotspot then go and massacre newbies, but most of the "git gud" ones are forever stuck in low rating. This means that at low rating genuine new (or hopelessly bad) players are facing mechanically good players who will not leave low rating until they get significantly above the other "git gud" ones. So newbies are massacred by people who do not intend to massacre newbies, simply follow the bad advice of "git gud".

This of course has a positive effect. The "git gud" players either die early and get back into action or - if they survive the initial hotspot massacre - they roflstomp the newbies with "omg 10 kills + win" memories. But the negatives obviously outweight it: newbies have very hard time progressing, practically the only way for them is to hide and reach 1500 rating where the average mechanical skill is lower to get kills. Also, many good players forever gets stuck in low rating, denying themselves the chance to face challenging opponents.

What would be the solution?
The numbers would be maximum possible kills. It means that if you are low rated and killed 2 people, your weapons would be taken and a message appear "you can't kill more people in this rating, please improve your rating by surviving as long as possible". This would take away the goal of intentional newbie bashing, no one could do massacre. I guess most streamers would be pretty upset if they would have to fight 1600+ rated people instead of clueless newbies. But the main goal would be forcing unintentional newbie bashers to focus on learning surviving instead of killing until they reach higher rating.


Anonymous said...

> What would be the solution?

Wouldn't it be easier to just publish the formula and include some in-game tools such as progress bars? Let the "git gud" players see that they're stuck in Bronze *because* of their blaze-of-glory deathmatch behavior. Show them a locked achievement which says "survive at least ten minutes in three successive matches." Let them figure it out (or let them watch the thousands of Youtube videos which quickly appear to explain the optimal strategy).

People are going to be upset if you take away their guns.

maxim said...

The one for whom newbie bashing is an important source of memories is not the game designer.
The game designer himself has a conflict of interest. Do you make game more newbie-friendly by driving away a good chunk of current streamers and fans? Once you do, what will be left in the game to enjoy? The survival aspect will be hurt if there are not enough predators to hunt you.

Anonymous said...

Are you thinking of buying a second copy of Pubg and play the "git gud" way? For science!

dobablo said...

A rating system provides a poor matchmaking guide so you suggest implementing game-play systems to promote behavior that will help the matchmaker? Wouldn't tweaking the matchmaker be better and simpler?
For example, change the matchmaker so it uses the higher of the win and kill ratings. That provides two methods to advance up the tiers. Killers with >1 K/D ratio will rank up (getting them out of the newbie tier), while survivors that can avoid dying in the killer phase also rank up.

Freddo said...

Protect the newbies by grouping them in a match, adding in a decent amount of other players. For players with strong kill steaks maintain a hidden counter and group those players to different matches.

But if the game is selling well and growing, don't expect them to devote too much developer resources to such niceties.

Gevlon said...

@Maxim: it's a buy to play game. Old player retention is not as important as new player recruitment.

@Anon: no way in hell. Simply because after I "git gud", it will affect my main account play

@dobablo: no matchmaking tweak works against those who want to game the matchmaker. You have to take their end goal.

@Freddo: the problem is lucky streaks. Where do you classify the player who jumps down to School 5 times, dies instantly 4 times and on the fifth he lands on top of an AKM and gets 7 kills?

retsep said...

Removing ability to shoot is strange. Getting tired (and therefore not dealing full damage) wouldn't be so much. Therefore I am proposing to reduce efficiency of shots after recent battle. After prolonged shooting, it decreases even further, for example, 0.9, 0.7, 0.4. Higher rating would allow to deal more damage before decrease. Not shooting slowly restores ability to deal full damage.

Anonymous said...

You have #1 on your graph. Did you finally win a game?

Gevlon said...

@retsep: that would teach players to dance around shooting cooldowns, something the "real" high rated game doesn't have. Bad idea.

@anon: no, I've found a screenshot

Anonymous said...

Seems you're having a hard time breaking the top 100. Bad luck or are you actually levelling out how far you can go with pure survival?

Gevlon said...

@Anon: I soon publish my results. The problem is that according to the detailed stats, most top 10 players do the same strategy as I do, plus they kill 1-2 people. I could outgrow the M&S and even the skilled, but not my own mirrors.

NoGuff said...

I've never seen a game where camping was so widely acceptable as a playstyle. I'm seeing youtube videos where players are getting high rankings just by hiding in the bushes until the last few circles. Do the stats somehow account for this?