tourney.ru форумЪ

http://www.battle.net/forums/war3/threa … &tmp=1

I re-read the 45 page request for the 1.14 AMM and looked for specific issues that I could address. However, I wasn't able to find much. Below are the topics I found most prevalent. Feel free to post any specific questions and I will do my best to answer them.

Level does not reflect skill

This is only true for players who have played very small numbers of games. A player's "skill" is actually determined quickly, with a reasonable estimate developed in 10 games or less, and continues to refine as they play more matches. They are accelerated towards the level that corresponds with their real skill quickly, despite being matched against tough opponents. A player with the real skill of a level 45 will start getting matched against stronger players fairly early on. Stronger players will accelerate to their real skill faster despite playing better opponents, and thus losing more games, as they move up the ranks.

Some players who are currently at or near the very top of the ladders have a valid complaint, and we are implementing changes that will address this in the very near future. There is no way to discriminate between two capped level 50's currently. One may be "very good" while the other may be one of the best players on Earth (Azeroth?). All players in the upper tiers of the ladder are strong players though, as is evidence by only a very small fraction of players on the entire ladder reaching those levels.

Arranged/Random team games have long wait times at higher levels

Higher level random team players have a shorter search time with the current system, while higher level arranged teams have a longer one. The system is already generous with how far out it will extend its search range. At the maximum team range it searches, there's a 95% probability that the higher rated team will win. Extending the range outward further would shorten wait times, but cause a degradation in game quality. For random teams, the search range is actually less discriminating than it was with 1.14. Search time high level players experience are the direct result of there being only so many high level teams and players searching for each other.

Team games do not match fairly

Arranged teams actually match quite well with the new system relative to the old one. In the old system if there was a team of a level 40 player and a level 12 player, they would be considered a level 26 team. Now say that team was actually two level 40 skill players, one of who had just reset. If it's finding level 21 arranged team opponents, the match quality is going to be low. With our current system it's going to be searching for teams much closer to level 40.

As a result of discriminating the search ranges less for random team players in the current implementation, there may be a wider variety of skill levels in a random team game. We may re-evaluate the balance between game quality and search times in the future.

The ladder is less competitive now

The ladder is more competitive now than it was before due to better matching. The perception problem stems from the fact that the average level in 1.14 was 10, and it's now 20, an intentional change. In the previous system there were many, many players within a few levels of 10. The best players were spread over a range of 15-50 with very few players at each level. This creates poor quality games for the highly compressed majority of level 5-15 players, who one game may be up against a top gamer, and another against a newbie. With the current system the level is a better reflection of skill, but those levels can not be compared to the levels used in 1.14. Player level and what it means to be level 40 is only comparable to other players on the current system. There is now a much better distribution of players, with still a very small percentage of players actually at 50. As mentioned before, we will soon be making some changes that will allow discrimination in skill between the players who are now level 50.

Dislike being "forced" into a 50% win ratio

If they are winning approximately 50% of their games the ladder is working. While some people might like to bash on players of lower skill and win all their games, the ladder's goal are ultimately (a) to accurately rank players based on their ladder achievements based on their level and (b) provide them a mechanism for finding opponents of similar skill that will provide a good game. However, there is absolutely no mechanism in place that "forces" a 50% ratio. The system won't pair a player with 10 wins and 9 losses against an opponent who is 95% certain to beat them to "even it out". Instead, it will pair with someone of equal skill. Absolutely nothing in the ladder looks at win ratios.

Win percentage is not a measurement of skill, achieving a high level is. Most players could defeat their pet dog 100% of the time and obtain a perfect winning percentage. But unless the dog has an exceptional talent for RTS games, this means relatively little. Another player winning half of their games against top ladder players is better indication of skill. In this ladder system that player will be rewarded with a high level, while the player with a high winning percentage against weaker opponents will not.

сам пока не читал и не знаю осилю ли
кому не лень переведите