People (other than Grand Masters) play unpredictably and make more "novel" errors than bots. Bots, unlike a person, will never miss an error you make.

A person's rating is different depending on who they play as each of us is more likely to win against a different playing style. Hence it is sometimes claimed you need to play 30 different people within 200 of your level so as to get a highly predictive rating.

So a bot that on average scores zero against a correctly rated 1600 person will often lose against some lower rated people and win against some higher rated people.

People (other than Grand Masters) play unpredictably and make more "novel" errors than bots. Bots, unlike a person, will never miss an error you make.

A person's rating is different depending on who they play as each of us is more likely to win against a different playing style. Hence it is sometimes claimed you need to play 30 different people within 200 of your level so as to get a highly predictive rating.

So a bot that on average scores zero against a correctly rated 1600 person will often lose against some lower rated people and win against some higher rated people.

We also have the interesting problem that a person who have played against a bot a few times is likely to do better then a person who is playing against the bot for the 1st time, even when the too people have the same correct rating. So should a bot be rated based on the results of only the 1st game each person have with it, or based on the results against the players who have the most games with it?

People (other then Grand Masters) play unpredictably and make more "novel" errors then bots, bots unlike a person will never miss an error you make.

A person's rating is different depending on who they play as each of us is more likely to win against a different playing style. Hence it is sometimes claimed you need to play 30 different people within 200 of your level so as to get a highly predictive rating.

So a bot that on average scores zero against a correctly rated 1600 person will often lose against some lower rated people and win agaist some higher rated people.

People (other than Grand Masters) play unpredictably and make more "novel" errors than bots. Bots, unlike a person, will never miss an error you make.

A person's rating is different depending on who they play as each of us is more likely to win against a different playing style. Hence it is sometimes claimed you need to play 30 different people within 200 of your level so as to get a highly predictive rating.

So a bot that on average scores zero against a correctly rated 1600 person will often lose against some lower rated people and win against some higher rated people.