Alan is saying that in baseball there is a distinction between "failures" and "errors".
In the non-baseball use of the terms, a "failure" would be interpreted as a bad outcome, but in baseball as in research, "failure" is inherent to the game and should not be seen as an embarrassment or a negative outcome. A very good baseball player gets on base only 33% of the time. A very good researcher may fail dozens of times for any eventual success.
Conversely an "error" is when you fail to do something that is technically feasible but you fail to achieve it because of a personal mistake. In baseball an "error" would be an outfielder who misses catching a fly ball. Fly balls are caught 98-99% of the time by professional players, and it's expected that they can catch one anywhere in their area of coverage. A researcher should be able to achieve any outcome which is "technically feasible", such as creating a new software language, creating a new operating system, or creating a new hardware integrated circuit if necessary.
To sum up his statements; Failures in research should be expected, embraced, and motivating. Errors are the real problem, and point to an inability to achieve outcomes which are feasible. And if you are achieving ALL your research goals without any failures then your goals are not big and challenging enough.
In the non-baseball use of the terms, a "failure" would be interpreted as a bad outcome, but in baseball as in research, "failure" is inherent to the game and should not be seen as an embarrassment or a negative outcome. A very good baseball player gets on base only 33% of the time. A very good researcher may fail dozens of times for any eventual success.
Conversely an "error" is when you fail to do something that is technically feasible but you fail to achieve it because of a personal mistake. In baseball an "error" would be an outfielder who misses catching a fly ball. Fly balls are caught 98-99% of the time by professional players, and it's expected that they can catch one anywhere in their area of coverage. A researcher should be able to achieve any outcome which is "technically feasible", such as creating a new software language, creating a new operating system, or creating a new hardware integrated circuit if necessary.
To sum up his statements; Failures in research should be expected, embraced, and motivating. Errors are the real problem, and point to an inability to achieve outcomes which are feasible. And if you are achieving ALL your research goals without any failures then your goals are not big and challenging enough.