@Romain672 I think a 2players game can be more strategic because there will be no other external factors affecting the game, so the condition that an opponent make mistake and another opponent benefit from it wont appear. There is higher control of game. Also if a player has very high elo in the game, putting this player to be 3rd may guarantee elo gain for the other 2players, which I think is not a kind of king making because those 2players are trying their best to get their best elo. Just like you 2 play 2players legend inventors, right 
?
@ibav I think the problem of ELO decay is that some players possess great skills but rarely play will suffer from unfair punishment. (your last point?) If we look into leaderboard in Caylus, a strategic-based game, The first player has 1288 elo, but there are players who go unranked just because they rarely play here, for example Doodads (1050 elo) and ickoicko (1454elo 151victories in 153games, who nearly win all games against every strong player). In such games I guess their ability wont really decay that much, the problem is just because they dont have time to play or they may feel boring because they mostly win against everyone.
I also think there is a problem of elo upward mobility. As the K value has been decreased, it is much more difficult to get a high amount of ELO. Players have to play many many games. The strategy of playing with beginners was good to boost elo, for example, in Caylus, deszczyk got 842 elo (387victories in 428 gaames), he mostly plays with players with <200, which is comparable to 1300 elo and 1400 players in the old elo system, and he maintains his top10 rank. But in his recently games he actually just get an average of 0elo per victory, because he plays with beginners only. That means, in the new elo system, players cannot use the strategies of playing with beginners, but those who did it in the past maintain in high elo. I would say this as 'a past benefit that the system provides now disappears, but players being benefited in the past maintain their high elo'. This can also be seen in NOIR: Killer vs Inspector. In the early implementation of the game the site provides benefit by allowing default killer in the normal mode. This allows players to boost elo (daveraux, 701 elo, 203 victories in 249games). This setting is then canceled. There is still a setting of default inspector in normal mode, which table administrators can leave then rejoin after their opponent comes to ensure they are killer. However, some players doing this are reset their elo, but some still remains (Kaosbombix, 575 elo). The above 2players are 'unranked', not sure if the site ban them to have proper elo or not. But then, on 30 March, the pure advantage of citizen arrest mode arrives, and when it is first implemented, inspector can do unlimited times of citizen arrest, which gives them a pure advantage. I would say it can be 90% winning if the inspector do not make mistake and possess basic skills. The current ranked first player, Yegor, raised from 284 elo to 627 elo until 18 April, which the day of slightly balanced version of citizen arrest is implemented. Yes, he raised his elo from 284 to 627! His current elo is now 645, which is 2month after the implemented version arrives. This shows that elo boost become much more difficult after this new version, making no possibility for players to challenge his first place.
I would like to talk more about difficult elo upward mobility in less popular games. Let's talk about evo. After the new elo system is implemented, old players get a direct 200+ elo bonus to their existing elo. Even if a player play one no elo game, he should get 1500elo in the past, but he is said to have 200 elo under the new system. But now, if a guy have not tried the game, he has 0elo only. In evo, the first player has 558 elo, from the 2nd to the 7th, they have elo from 371 to 208. 371 elo means about 1671 elo in the past, which can be easily achieved by playing 8-12games with beginners who do not know rules at all and win the game easily. But I would like to focus on the elo of the 8th player, garchon. If you search his playing record, including games dont finished properly, his total ELO points won, calculated by the system, is -8, negative 8. We can then look into another player, who plays his first game after the elo system changes, i.e. he starts with 0elo. He is Kaosbombix again. He gets 128 elo (10 victories in 13games), and his total elo points won is 128, 120 higher than garchon. Kaosbombix has not played for 60days so he is ranked now, he actually ranks 11th only, even lower than garchon. In fact, in less popular games, it is quite difficult to find players, it often costs more than an hour. How many players are willing to spend so much time for just a game? They may feel demotivated when they try to play or upgrade their rank, because there is nearly no players playing these games, and they can only accept the fact that old players get benefit because of the ELO system.
In terms of game difficulty, I would like to classify game as pure luck-based, luck-based with low strategy entrance level, luck-based with medium strategy entrance level and strategy based game (which are actually luck-based game with high strategy entrance level, or abstract games dont involve luck at all). I refer strategy entrance level to 'when a player possess basic strategy level and he does not make mistake in the game, the victory bases on luck if his opponent also possess that strategy'. I would say Nile and Senet is the first type of games, most casual game include cant stop is the second type and some 'games for regular players' and Libertalia is belonged to the third. I would like to talk about the first and third type in particular.
In the first type of games, even though there are some strategies involved, you can still have about 40% chance to win even though you do not know the rules at all and do what the system allows you to do. Senet is one of them. In the past, players try to rank high by playing with players with as high elo as possible because they can gain more or loss less elo. Players who get luck in win games with a row result in high elo. The player who ranks first on this game is 507elo (35victories in 48games), which is 1807 elo in the past and he has a net gain of 307 elo. In the past, there some cheators like deszcz (demon now) gets 100% victory rates because he plays with friends, and his friend abandon or concede in the game to make sure deszcz wins. But this is not the case for fedetsrm. None of his game in senet is abandoned, meaning he is really blessed to get such high elo. This achievement is actually impossible to have because the victory rates of most players who play a considerable amount of games will not exceed 65%. The elo of the 2nd in this game is 415 and the 3rd has 379. However, as the K value in the new system decreases, it is impossible for players to challenge fedetsrm's rank and his 50k trophies are guaranteed. For fedetsrm, in order to guarantee his rank, which is obtained by some strategies and a good method to choose opponents, he plays a game every 60days to avoid high elo loss. This is actually reasonable because he has no incentives to suffer a risk in losing his elo. I do not think this is his fault, he is protecting something difficult to get. There are complaints about players who rank first and play a massive amount of no elo games. Again I don't think it is a problem because there is no reason for these players to play elo on and they have their right to play. It is difficult to judge whether a player is just busy to have a game, getting bored of it or they use other accounts to play (they can just deny that is different person, maybe parents or friends), it is too difficult to judge. The real problem is that after the elo system changes, players feel more difficult in challenging leaders.
This is also due to the fact that the numbers of players in a game gradually shrinks after the game is implemented. The period when the game is just implemented is a golden period for players to boost elo because there are many high elo beginners in that period. Some players like Oso use this strategy to boost elo. This creates a group of high elo players in the leader board, plus their 200 elo bonus, their ranks seems to be more difficult to be challenged. But the main problem is that there are not as many high elo players are playing a game after a long period of time. Those who rank high may keep playing, but as their maximum elo loss in a game is just 20, they wont suffer that much. The golden period in the past is the first 30games, they get elo bonus, which is higher than now, and the number of games also exceed the current 20games that beginners can get bonus. For aggressive elo players, they can actually learn the game before it is implemented by keeping tract on the news bga release and look at games that admins and active developers play. They may be testing games, but those games still appear in their game record, though the games are not allowed by others to play. The current elo bonus in the first 20games may bring a player to about 300 or 400 elo, but it is still much far from those leaders. As the current system encourages players to play with similar level players, the giant killing way to boost elo in the past no longer works that much, a player can only get a maximum of 20 elo in each game.
Last I would talk about games with medium strategy entrance level, Libertalia. It is quite a luck-based one if a game is played within strong players, in real time. Frequent players like supermarky, Mumaspejo and and typhoon81 usually has similar chances to win in 2players game, especially the one who get the black color cards (only red and black in 2players). They may reach a high rank, but as these active players play more games against similar level players, their elo will gradually be balanced and they cant reach the top. This again promotes the behavior of playing with beginners in such games. The current first player in this game is deszczyk, with 725 elo (146 victories in 176games). Again, he played many games against beginners with low elo, just like what he do in the past. This allows him to get about 2 to 3 elo in each game in the old system in the past, then play a game every 60days after that. The second in this game, firstgrader uses the strategy of playing turn-based games. In fact, players who play turn-based game generally have lower elo, and this is the situation in this game as most active players play real time. This is because counting cards are required in this game, which is better to be done in real time. Players may use the 'note' function in turn-based to count. As most players do not count in turn-based, I would say they enjoy casual games and they can play freely, players who count in turn-based can greatly benefit. The player with highest elo in this game is Scaraoski with 794 elo. As mentioned, this game is actually quite luck-based, Scaraoski wins 30games out of 33games with a same IP player, kush. Kush is said to be an expert, but his victory rates are especially low against Scaraoski, but his victory rates are normal against other strong players. Again this is something strange.
no offenses to most of the players mentioned above.
?@ibav I think the problem of ELO decay is that some players possess great skills but rarely play will suffer from unfair punishment. (your last point?) If we look into leaderboard in Caylus, a strategic-based game, The first player has 1288 elo, but there are players who go unranked just because they rarely play here, for example Doodads (1050 elo) and ickoicko (1454elo 151victories in 153games, who nearly win all games against every strong player). In such games I guess their ability wont really decay that much, the problem is just because they dont have time to play or they may feel boring because they mostly win against everyone.
I also think there is a problem of elo upward mobility. As the K value has been decreased, it is much more difficult to get a high amount of ELO. Players have to play many many games. The strategy of playing with beginners was good to boost elo, for example, in Caylus, deszczyk got 842 elo (387victories in 428 gaames), he mostly plays with players with <200, which is comparable to 1300 elo and 1400 players in the old elo system, and he maintains his top10 rank. But in his recently games he actually just get an average of 0elo per victory, because he plays with beginners only. That means, in the new elo system, players cannot use the strategies of playing with beginners, but those who did it in the past maintain in high elo. I would say this as 'a past benefit that the system provides now disappears, but players being benefited in the past maintain their high elo'. This can also be seen in NOIR: Killer vs Inspector. In the early implementation of the game the site provides benefit by allowing default killer in the normal mode. This allows players to boost elo (daveraux, 701 elo, 203 victories in 249games). This setting is then canceled. There is still a setting of default inspector in normal mode, which table administrators can leave then rejoin after their opponent comes to ensure they are killer. However, some players doing this are reset their elo, but some still remains (Kaosbombix, 575 elo). The above 2players are 'unranked', not sure if the site ban them to have proper elo or not. But then, on 30 March, the pure advantage of citizen arrest mode arrives, and when it is first implemented, inspector can do unlimited times of citizen arrest, which gives them a pure advantage. I would say it can be 90% winning if the inspector do not make mistake and possess basic skills. The current ranked first player, Yegor, raised from 284 elo to 627 elo until 18 April, which the day of slightly balanced version of citizen arrest is implemented. Yes, he raised his elo from 284 to 627! His current elo is now 645, which is 2month after the implemented version arrives. This shows that elo boost become much more difficult after this new version, making no possibility for players to challenge his first place.
I would like to talk more about difficult elo upward mobility in less popular games. Let's talk about evo. After the new elo system is implemented, old players get a direct 200+ elo bonus to their existing elo. Even if a player play one no elo game, he should get 1500elo in the past, but he is said to have 200 elo under the new system. But now, if a guy have not tried the game, he has 0elo only. In evo, the first player has 558 elo, from the 2nd to the 7th, they have elo from 371 to 208. 371 elo means about 1671 elo in the past, which can be easily achieved by playing 8-12games with beginners who do not know rules at all and win the game easily. But I would like to focus on the elo of the 8th player, garchon. If you search his playing record, including games dont finished properly, his total ELO points won, calculated by the system, is -8, negative 8. We can then look into another player, who plays his first game after the elo system changes, i.e. he starts with 0elo. He is Kaosbombix again. He gets 128 elo (10 victories in 13games), and his total elo points won is 128, 120 higher than garchon. Kaosbombix has not played for 60days so he is ranked now, he actually ranks 11th only, even lower than garchon. In fact, in less popular games, it is quite difficult to find players, it often costs more than an hour. How many players are willing to spend so much time for just a game? They may feel demotivated when they try to play or upgrade their rank, because there is nearly no players playing these games, and they can only accept the fact that old players get benefit because of the ELO system.
In terms of game difficulty, I would like to classify game as pure luck-based, luck-based with low strategy entrance level, luck-based with medium strategy entrance level and strategy based game (which are actually luck-based game with high strategy entrance level, or abstract games dont involve luck at all). I refer strategy entrance level to 'when a player possess basic strategy level and he does not make mistake in the game, the victory bases on luck if his opponent also possess that strategy'. I would say Nile and Senet is the first type of games, most casual game include cant stop is the second type and some 'games for regular players' and Libertalia is belonged to the third. I would like to talk about the first and third type in particular.
In the first type of games, even though there are some strategies involved, you can still have about 40% chance to win even though you do not know the rules at all and do what the system allows you to do. Senet is one of them. In the past, players try to rank high by playing with players with as high elo as possible because they can gain more or loss less elo. Players who get luck in win games with a row result in high elo. The player who ranks first on this game is 507elo (35victories in 48games), which is 1807 elo in the past and he has a net gain of 307 elo. In the past, there some cheators like deszcz (demon now) gets 100% victory rates because he plays with friends, and his friend abandon or concede in the game to make sure deszcz wins. But this is not the case for fedetsrm. None of his game in senet is abandoned, meaning he is really blessed to get such high elo. This achievement is actually impossible to have because the victory rates of most players who play a considerable amount of games will not exceed 65%. The elo of the 2nd in this game is 415 and the 3rd has 379. However, as the K value in the new system decreases, it is impossible for players to challenge fedetsrm's rank and his 50k trophies are guaranteed. For fedetsrm, in order to guarantee his rank, which is obtained by some strategies and a good method to choose opponents, he plays a game every 60days to avoid high elo loss. This is actually reasonable because he has no incentives to suffer a risk in losing his elo. I do not think this is his fault, he is protecting something difficult to get. There are complaints about players who rank first and play a massive amount of no elo games. Again I don't think it is a problem because there is no reason for these players to play elo on and they have their right to play. It is difficult to judge whether a player is just busy to have a game, getting bored of it or they use other accounts to play (they can just deny that is different person, maybe parents or friends), it is too difficult to judge. The real problem is that after the elo system changes, players feel more difficult in challenging leaders.
This is also due to the fact that the numbers of players in a game gradually shrinks after the game is implemented. The period when the game is just implemented is a golden period for players to boost elo because there are many high elo beginners in that period. Some players like Oso use this strategy to boost elo. This creates a group of high elo players in the leader board, plus their 200 elo bonus, their ranks seems to be more difficult to be challenged. But the main problem is that there are not as many high elo players are playing a game after a long period of time. Those who rank high may keep playing, but as their maximum elo loss in a game is just 20, they wont suffer that much. The golden period in the past is the first 30games, they get elo bonus, which is higher than now, and the number of games also exceed the current 20games that beginners can get bonus. For aggressive elo players, they can actually learn the game before it is implemented by keeping tract on the news bga release and look at games that admins and active developers play. They may be testing games, but those games still appear in their game record, though the games are not allowed by others to play. The current elo bonus in the first 20games may bring a player to about 300 or 400 elo, but it is still much far from those leaders. As the current system encourages players to play with similar level players, the giant killing way to boost elo in the past no longer works that much, a player can only get a maximum of 20 elo in each game.
Last I would talk about games with medium strategy entrance level, Libertalia. It is quite a luck-based one if a game is played within strong players, in real time. Frequent players like supermarky, Mumaspejo and and typhoon81 usually has similar chances to win in 2players game, especially the one who get the black color cards (only red and black in 2players). They may reach a high rank, but as these active players play more games against similar level players, their elo will gradually be balanced and they cant reach the top. This again promotes the behavior of playing with beginners in such games. The current first player in this game is deszczyk, with 725 elo (146 victories in 176games). Again, he played many games against beginners with low elo, just like what he do in the past. This allows him to get about 2 to 3 elo in each game in the old system in the past, then play a game every 60days after that. The second in this game, firstgrader uses the strategy of playing turn-based games. In fact, players who play turn-based game generally have lower elo, and this is the situation in this game as most active players play real time. This is because counting cards are required in this game, which is better to be done in real time. Players may use the 'note' function in turn-based to count. As most players do not count in turn-based, I would say they enjoy casual games and they can play freely, players who count in turn-based can greatly benefit. The player with highest elo in this game is Scaraoski with 794 elo. As mentioned, this game is actually quite luck-based, Scaraoski wins 30games out of 33games with a same IP player, kush. Kush is said to be an expert, but his victory rates are especially low against Scaraoski, but his victory rates are normal against other strong players. Again this is something strange.
no offenses to most of the players mentioned above.
Thing is, it's not as simple as "elite players shouldn't be required to actually play."
