Boards›Gameplay

Partial Defeat / Partial Victory

RegRobust·11/7/2014, 3:25:17 PM·1 votes·518 views

The IP won or lost on a game should be modified by how well the team has done in the game.

In the past, Riot has said they would not include things like KDA's or cs, and they are right to do so. Their (correct) reasoning is that modifying IP losses and gains through things like KDA's would distort the incentives of players. Players would try to rack up a good KDA instead of doing things that are good for the team. For example, a support might not sacrifice herself for the ADC even if doing so increased her teams chances of wining. LOL is a team game, and IP gains and losses should reflect team performance.

Indeed, this is the crux of the matter. You can measure team success through things other than just destruction of the nexus. For example, whether an inhib was destroyed or the turret count. Let's spell this one out.

  1. Total Victory - your team destroys the enemy nexus and none of your team's inhibitors were destroyed 2.** Partial Victory** - your team destroys the enemy nexus but at least one of your inhibitors was destroyed
  2. Partial Defeat - your team's nexus was destroyed, but you also destroyed one of their inhibitors
  3. Total Defeat - your team's nexus was destroyed and you did not destroy any of their inhibitors.

What's the Point?

There are two benefits to modifying ELO gains and losses based on other events in the game. First, doing so improves the speed at which a player's ELO converges on his latent skill. The current binary measure of success is a messy measure of how well your team did; a simple win/loss outcome throws away a lot of information in the game that is informative of a team's performance. Consequently, it takes a lot of games for your ELO to converge on your latent skill level. Incorporating intermediate outcomes like inhibitor kills would improve the informational content of the game's outcome.

Second, doing so would improve the motivation of players on both the winning and the losing team. In many games I have played, teammates have given up when winning the game becomes a long shot. But it is easier to kill an inhibitor than to win. So losing teams might decide that the chances of winning are slim but the chances of mitigating IP losses with an inhibitor kill are worth exerting more effort. On the other hand, many winning teams drag out games because they get overconfident and feel assured of victory. The possibility of gaining fewer points from a partial victory would incentivize them finish games faster.

An objection to this is that it would distort incentives, and this is partially true. A team that is losing might decided to have 4 champs defend the other team's 5-man siege while sending their fifth to split push for an inhibitor. This might decrease the chances the losing team makes a comeback, while at the same time maximizing the expected IP gains. Purists might consider this a problem. However, I think this would actually make the game more interesting. A team that has snowballed hard and is virtually assured of victory must remain diligent to ensure they do not throw away their total victory.

Thoughts?

1 Comments

ValyrianBlade11/7/2014, 3:44:26 PM1 votes

I like it, and would include nexus turrets as well (for something between total and partial). I've lost a game where my team took 2 barons to 0, both nexus turrets and each inhibitor twice and one of them three times. We lost because the other team had a much better teamfight comp and turtled really well to defend. Eventually they scored an ace at their nexus and 4 ran down to clear our base while the other just stopped our minions from winning the game. This loss felt much different from a loss where I took the bot Tier 1 turret at 20 min and we got nothing else all game.

I think this would really help because such close Games indicate that the matchmaking was appropriate. We lost, but the game definitely could have gone either way. While some individual players may have too high or too low mmr, the teams were even and that's valuable information. In contrast, a lopsided contest would inform that at least 1 player on one of the teams is at the wrong mmr. After a few lopsided Games in a row for that player, the correction could be made a lot faster.

I'd want to count inhibitors though. 1 inhibitor shouldn't make a large impact (maybe 1 or 2 points) as you can still either split push one or win 1 teamfight in a lopsided affair. E.g. Trade inhibitor for Baron then lose on the next push. If you take 3+ inhibitors and lose though, it was likely a really close match.

I've also wondered why the length of the match isn't considered properly. I'm rewarded better (ip and xp wise) for a 27min win than a 45min loss (fine) but I get more for a 55min win than for a 27min win. I get that you should be rewarded for your time, but really if I'm winning the game fast we dominated whereas if I win slow it was closer. On the losing side it's appropriate, but I just find it disgusting when I lose a 55min match and get so little rewards relative to had I won it when it was so close, whereas surrendering at 20min in a lopsided contest I lose very little in rewards compared to had I won in 20 min. The only consolation is the long game is typically more fun :) I just think longer Games souks lessen the gap between the winning and losing team, although this would discourage surrendering so I like the inhibitor idea (as you'd still surrender if you don't think you can even get an inhibitor).