Every game on the Cubecraft server has statistics that show various numbers regarding a player's gameplay. One such statistic determines the leaderboard for that game. I was thinking of another statistic called elo, a concept invented by Arpad Elo originally designed for tennis, but it has been adopted by various games such as chess and backgammon. Elo is a number that purely represents a player's skill. A player could play ten thousand of a certain type of game but just not have a natural talent for it and thus still not be quite as good as someone who has only played a hundred games (a common reason for this would be that the more experienced player does not have as much experience with PvP in general, while the other player does and can therefore still win). A player's elo will never increase unless they get better at the game (i.e. this is not a statistic that will increase with games played, such as wins, kills in most games, egg breaks in eggwars, or Snowman Survival medals). My suggestion is to introduce Elo into a player's statistics. This has been suggested before, both by me and by some other people on the Cubecraft forums, but I am suggesting it only as a representative number, rather than a way of ensuring players are matched only with other players of similar skill, as this idea has been commonly regected because it would severely slow matchmaking.
A player's elo would likely begin at a higher number if they already have a high Elo in similar games, but I'll explain an example using an 8-player game of Skywars and a starting elo of 1000. If everyone goes into a game with 1000 elo, the total elo at the start of the game is 8000, so the total elo at the end of the game must also be 8000 (to prevent players from getting a higher elo by just playing more than everyone else). For example, after the game, the winner's new elo is 1020, 2nd is 1005, 3rd is 1002, 4th is still 1000, 5th is 998, 6th is 995, 7th is 992, and the first dead is 988. The total elo of all the players is unchanged, but the winner gains the most elo while some of the players who died first lose some of their own. For bigger differences in elo rating (as, again, it is a representative number because using it as a system to ensure even matchmaking would slow it), players of higher elo will gain less elo for winning and lose more for losing. For a two-player duel with one player at 1500 elo and the other at 800, the player of higher elo would gain only 1 point for winning, but would lose 30 points for losing due to the extreme difference in ratings (in a real elo system, the higher-rated player would actually lose closer to 57 points, but this is just an example). This further ensures that elo is based entirely on skill and not how many games someone has played, as winning the same amount of elo for every game would mean infinite grinding against noobs.
Here are a few ideas of how this could also be more than a representative number among a player's statistics:
A player's elo would likely begin at a higher number if they already have a high Elo in similar games, but I'll explain an example using an 8-player game of Skywars and a starting elo of 1000. If everyone goes into a game with 1000 elo, the total elo at the start of the game is 8000, so the total elo at the end of the game must also be 8000 (to prevent players from getting a higher elo by just playing more than everyone else). For example, after the game, the winner's new elo is 1020, 2nd is 1005, 3rd is 1002, 4th is still 1000, 5th is 998, 6th is 995, 7th is 992, and the first dead is 988. The total elo of all the players is unchanged, but the winner gains the most elo while some of the players who died first lose some of their own. For bigger differences in elo rating (as, again, it is a representative number because using it as a system to ensure even matchmaking would slow it), players of higher elo will gain less elo for winning and lose more for losing. For a two-player duel with one player at 1500 elo and the other at 800, the player of higher elo would gain only 1 point for winning, but would lose 30 points for losing due to the extreme difference in ratings (in a real elo system, the higher-rated player would actually lose closer to 57 points, but this is just an example). This further ensures that elo is based entirely on skill and not how many games someone has played, as winning the same amount of elo for every game would mean infinite grinding against noobs.
Here are a few ideas of how this could also be more than a representative number among a player's statistics:
- Use players' elo ratings to ensure players of high skill are not paired with players of low skill. Again, this would slow matchmaking, but if matchmaking is going sufficiently fast, this can ensure that someone who is brand new to a game like Snowman Survival (where basically all the experience a player has in other areas of Minecraft goes out the window) is not paired with someone like me, who has played over five thousand games of it. This would slow matchmaking and would likely be used only on Bedrock edition as the player base is larger, and would only be used to prevent extreme differences of skill levels in a game.
- Make a leaderboard for the players of highest elo rating. This does not have to replace the leaderboards that already exist in each game. This could be an alternate leaderboard that shows the most skilled players, rather than merely those who have played the most.
- If there is a significant difference in elo ratings in a game, give the players with lower rating an advantage and give those of higher rating a disadvantage. This should be subtle. The main takeaway of this is that it could encourage sandbagging (intentionally losing games to play easier opponents, or, in this case, get an advantage).
