Red Post Collection: Upcoming changes to Refer-a-Friend, "Suggested Players" discussion, Semifinals MVP Bundle,&more!

Tonight's red post collection features an announcement regarding upcoming changes to the Refer-a-Friend system, Lyte discussing the "Suggested Players" feature in game lobbies,  CertainlyT chatting about hyperscaling and base stat offsets, a look at the new semifinals  MVP bundle, and more!
Continue reading for more information!

• This week in eSports -2014 WC Semifinals Results, Amazing Leaves TSM, and more!
• Champion and Skin Sale 10/14 - 10/17
• Free Champion Rotation 2014 Season Week 39
• Dunkmaster Darius now available

• 10/13 PBE Update: Eleven New Champion Splash Arts, Preseason Changes reverted, and more

Upcoming changes to Refer-a-Friend

Riot Metasystem has posted up an announcement regarding upcoming changes to the Refer-a-Friend system that are aimed at combating botting and fraud, include changes to rewards and a much lower threshold to obtain Grey Warwick and Medieval Twitch.

"The TL;DR for those of you who just want to get back into game: we’re making some changes to Refer-a-Friend in the upcoming weeks to combat abuses like botting and fraud.

We started RAF to reward players for their part in growing League. Refer-a-Friend has quite the history at this point, with multiple incarnations dating back to the original launch of the game, and it’s been a rewarding experience for lots of players.

However, despite the best intentions of almost everyone, a small group of people have severely abused the system, creating negative play experiences for others, particularly new players.

Any rewards program like RAF runs the risk of some participants trying to game the system, and ordinarily we might tolerate low levels of fringe misuse if it meant we were still doing right by the vast majority of players. However, RAF abuse was increasingly degrading the average players' experience, forcing us to contemplate changes.

We’ll start with the symptoms of RAF abuse - things like account selling and botting (i.e. the use of scripts, programs, or other hacks to automate playing a game of League of Legends). The two are closely related in the case of RAF, as botting referrals enables account sellers to create accounts loaded with currency and skins.

In the past, RAF botting mostly took place in private custom games, where players’ experiences weren’t affected. However, as we said, we've recently seen bot usage swell in volume and spill into places like new player games and into game modes like Dominion. We'll be taking action against an array of offenders with the goal of stemming this trend. Additionally, to most effectively combat this type of botting we have to widen our toolkit to include more than just reactive measures like bans.

Ultimately, the best way to attack RAF abuse is to address the underlying symptoms that make it attractive to abuse in the first place. Specifically, the key cause of most misuse is actually the nature of the program itself.

Basically, abusers could make a ton of RP and real money by conducting large-scale botting for the purpose of selling accounts. Worse still, the average player faces a dilemma where attractive rewards like Grey Warwick and Medieval Twitch are prohibitively difficult to obtain honestly.

Let’s talk numbers. We found that while only about 1.4% of RAF participants attempt to abuse the system, if we isolate accounts with 10 or more referrals the percentage jumps to 70%, and then skyrockets to 99% when you count only those with 100 or more referrals. Consider that Grey Warwick and Medieval Twitch currently require 25 and 50 referrals, respectively, and the issue is clear. We understand that the current RAF reward structure isn’t ideal for the average player, which is why the reactive efforts we mentioned earlier are focused specifically on dedicated and repeated offenders - no one's going to be punished for referring a smurf or two.

We're working on updates to the Refer-a-Friend program to simultaneously remove incentives for abuse and better serve most League players. One of the key issues here is that it’s difficult for a single program to fully meet the needs of both the average player who might refer a couple friends and a hypothetical YouTube personality who might refer thousands.

We found that the majority of League players refer between one and five friends, and we’ve designed the RAF update to reward you for bringing those friends into the game. With the new update, for every friend you refer who reaches level 10, you’ll earn 1000 IP. You can refer a total of five friends, and with three successful referrals, you’ll unlock the Grey Warwick skin. Recruit five friends and you’ll unlock Medieval Twitch.

There are no additional rewards to earn, and everyone who has participated in Refer-a-Friend in the past will be transferred to the new program with the same number of referrals as they currently have. For example, if you’ve already referred four friends, you’ll earn 4000 IP and Grey Warwick.

As you can see with the plans for the updated Refer-a-Friend, we’re still looking into separate ways of rewarding the highest-volume referrers that don't compromise the experiences of everyone else.

Keep in mind that these changes are focused specifically on combating RAF misuse and its impact on the average player. Botting, account selling, and other forms of fraud have causes beyond RAF, and we continue to cast a wide net when it comes to investigating and fighting them.

We wanted to give to you a heads up on upcoming changes to RAF as well as give you some insight into the motivations behind what we’re doing. Please leave your comments or feedback below!"

"Suggested Player" Discussion

In a reddit thread discussing the "Suggested Player" feature that offers you recommendations on who to invite to your game lobby, Lyte stepped in to comment on a few issues and concerns players were voicing:

"This is a work in progress feature so please be patient with us as we iron out the kinks. 

1) We're analyzing the data to see if players need (and want) an option to decline these types of invites from friends of friends. We're noticing that for some players, they have huge social networks so they are getting a LOT of friend invites. We need to find ways to improve the quality of each invite. 

2) We're improving the algorithms to ensure that suggested friends of friends fit a certain MMR range so you have better matches (and lower numbers of random invites). 

3) The system already suggests players you premade with, and players you have Honored. We're tweaking the numbers to make sure each category has the right "weights," so it's possible your friends of friends are weighted too heavily right now. 

4) We'll look into the system showing players you already play with often. Right now, it suggests premades you've played with recently, which isn't exactly the same and we realize that!"

In response to the player suggestion of being able to "pin" certain friends to the top of the recommended list, Lyte replied:

"A few teams have been tossing the idea around of a "Favorites" option. This would require some work on the Friends List, but definitely possible."

When asked about friends who you've recently played with not showing up correctly, he noted:

"This should be fixed in the latest patch. There was an error where premades you just played with wouldn't show up that we had to patch up."

He continued:

"This sounds like an edge case we haven't covered. Let me try to explain how the system should work, and you can let me know if it doesn't and we need to fix something. 

1) You want to play with a friend, and see him on the Recommended List. 2) You invite him from the Recommended List and he accepts. 3) You play a game together. 4) You start a new game, and the friend should show at the top of the Recommended List. 

After #4, does the friend leave the 2nd lobby you've created after they've joined?"

As for the invites showing you more information about the person you recently played with, Lyte noted:

"Right now, it does show Riot Lyte (friend of XYZ) would like to invite you to a game. However, it's possible that the custom invite messages are being missed because players are so used to seeing a default message there. 

The idea of providing more context/details to help you trigger who a person is is interesting, and an idea we're exploring."

When asked what the purpose of the suggested friend feature is, Lyte commented:

"I think the purpose of the feature varies greatly from player to player. 

For some players, they aren't too keen on getting suggestions from seemingly random players, even if they have a friend in common. For other players, they want to play with players they have some connection with. 

What if the connection was someone from the same university as you? Same city? Would these serve better purposes to you? 

We're actually finding that these games with players that you are somewhat familiar with result in better game experiences. What if we offered some incentives like IP bonuses to play with more premades, would you then take these invites more often?"

He continued, expanding on the geographical recommendation comment from his last post:

"As you can tell with a feature like this, ideas like geographic location being considered are all possible; however, we have to take into account privacy (and respect privacy!) and all the other factors like whether players actually want to know who is nearby, at the same university, or workplace, or LAN/PC Cafe. 

Although many players might love the feature, some will hate it, and we have to carefully balance the experiences."

CertainlyT on Hyperscaling

CertainlyT also replied to a forum thread regarding champions having weaknesses in the same areas as their strengths (a la Sion having low base HP but scaling HP on his W), explaining what stat hyperscaling with base offsets allows them to do with a champion's balance:

"Quote:

• You give Cassiopeia huge sources of Ability Power gains. You make her Ratio's abysmally small.

• You give Garen Armor and Magic Resistance gains. You make his base Armor and Magic Resistance's small.

• You give Sion increasing health gains. You make his base health abysmally small.

• When are we going to see Veigar pay for his ability power gains? (You'd better not touch him).

This is an absolute terrible balancing method. What is the point of giving these champions strength's in these area's if you're just going to make them weaker in the SAME EXACT AREA for it. It feels bad. It looks bad. Because it's bad. 

I would really like some sort of ..... clarification on why this is done this way. It's frustrating to see. It makes me feel like I have very little choice in the items I pick up because of how riot is performing balance on these types of champions."

Let's call these sort of additive or multiplicative statistical bonuses "stat hyperscaling." Let's call reduced base stats on champions with stat hyperscaling "base offsets."

I see two variants of the argument presented in GhostStalker's post: 1) Some stat scaling bonuses are undertuned, so that the champion doesn't feel like they end up feeling benefited from their bonus; and 2) Champions with stat hyperscaling should not have lower base stats than their counterparts who lack those bonuses.

I'll not address #1 in depth, but have two short thoughts: First, you might be right. It's entirely possible we've undershot the mark in some places. Second, it is possible that you aren't getting the full benefit of some of these stat hyperscalings. As I'll talk about below, hyperscaling is designed to feel good when availed to its fullest.

As to point #2, this technique (stat hyperscaling with base offsets) allows us to do one of three things.

1) Create game-by-game variance in a champion's strength. We aren't interested in any champion always feeling superlative (tankiest, burstiest, etc.). Rather, we want players to have games where, due to an excellent performance, their champion achieves that feeling of being unkillable, or of annihilating an opponent in a single spell rotation. In the case of Sion, his HP/Level of 104 (currently 70) was retained for much of his rework's development. Feedback from nearly every playtest was that he felt unkillably tanky. After reducing his HP per level but keeping his W passive HP per kill high, Sion reached a point where he sometimes became a super tank and sometimes failed to do so, based upon the Sion players performance and the enemy team's ability to keep him from farming. 

This variance frequently results in player frustration when they fail with a champion and then fault base offsets for depriving them of their superlative quality. In reality, the offsets are not at fault, but rather the champion might be undertuned (so they did not have a fair shot at scaling), the hyperscalar may be undertuned (this is point #1), or the player may be not yet be optimizing their stat hyperscaling (i.e., they are bad, which is normal on newer champions or major reworks).

2) Create distinct gameplay through alternate incentives. Generally, we are happy with LoL's emphasis on Gold and EXP as a means of scaling stats. But in some cases, we like to emphasize one over the other. Garen, for example, emphasizes build choice over character level. He can customize his defensive strengths to suit the game at hand. Thresh emphasizes active resource acquisition (picking up souls) over passive resource acquisition (being in EXP range) because I wanted to both encourage the player to stay closer to the minion wave and to make the same sort of moment-to-moment decisions that come from last hitting. If Thresh tended to up with the exact same amount of armor as a base stat driven champion, it wouldn't bother me in the slightest, since I am happy with the distinct gameplay behind his armor growth.

3) Create intra-game variance, typically over time. This is most obvious on marksmen, who, as a class, tend to have substantial base offsets in exchange for hyperscaling. Similarly, with the Cassiopeia rework, RiotStashu sought to create a sustained damage mage with a marksman-like damage curve by moderating her early ratios in exchange for higher late game ratios (combined with less quantifiable usability changes that resulted in higher DPS). As such, Cassiopeia's power spike is later than a standard mage, and later than her previous self, which is undoubtedly jarring to some players."