Discussion about this post

User's avatar
Froseph's avatar

Some thoughts:

Improper focus on win rate:

1/ Win rate is dependent on the skill differential in of games played. In most match making games (colonist.io, lichess, Counterstrike, etc) the servers match you in games based on the their estimate of your skill / rank. When a good player create a new account, they rack up much more wins early on as they rise through the ranks. Depending on how many games people play, the early wins may have more or less weighting in their historical win rate.

2/ It's more accurate to look at win rate across similarly skilled players.

3/ Even in games that are "pure skill", i.e. Chess there is still some randomness (luck). Sickness, emotional state, etc can all play a factor in how well someone's playing in a day. Chess uses Elo ratings where people with the same rating have 50% win rate. A +200 point Elo rating difference equaling ~75% win rate.

4/ Given that Catan is typically played as a 4 person game, 4 people of equal skill should have 25% chance of winning. Thus, an elo rating should reflect a 25% win rate against players with the same Elo.

5/ 50+% win rate is only possible if you only play people with much much lower skill than you. Is this fun?

Yet, results oriented thinking is wrong. Use win rate instead.

1/ In any skill based game, there is always some amount of luck. Even chess.

2/ In analyzing a win in a game, you should be aware of where luck may have been involved, or if you bad play happened to be lucky.

3/ The way around this is to consider looking at the factors and and to look at a series of games, i.e. win rate, to help validate strategies and tactics. AI / computer modeling has typically been a factor in pushing this forward-- see Backgammon, Poker, or Chess.

On the benefits of Tournaments to a community:

1/ Tournaments and competitions are a great way to get people to push the game forward. Some people will strive hard to be the best and find new / innovative strategies. Removing tournaments reduces this incentive and may harm the skill level and the competitive scene.

2/ A solution to diminish the effects of variance is to create tournaments with a series of games[1]. Existing examples this include Chess and Baseball. The question is how many games should you play to determine the winner in a tournament and how much variance should be expected by the community?

3/ Is variance harmful to the community / players? Poker tournaments play relatively few hands but they still thrive. The top players all know and understand that variance is a part of the game, and some of the skill is dealing with their emotions so the avoid making bad emotional plays. FWIW in poker grudge matches, 25k+ hands isn't uncommon, and with poker AI bots, they used 80k+ hands to determine the winner.

[1] Alternatively you could take the Contract Bridge approach where everyone plays the same board, but are only competing with teams in the same position. This means multiple winners in each tournament.

Expand full comment
1 more comment...

No posts