* No committee can effectively judge the value of a paper at its release
  * The value of a paper is defined by its impact over time
* Source code and data should live with a paper
  * It should be easy to make it so
* Pseudonyms can limit bias?
  * gender, race, etc bias
* Feedback should be intelligent and constructive
  * "Letters" more than comments
* Notice your incentives
* All incentives encourage releasing good work, not more work
