Can you share specific things that need to happen slowly and why that slowness helps protect study participants?

Some things mentioned in the article clearly increase risk to participants: manually re-assembling translated documents, using custom software and paper forms, and diluting important documentation with boilerplate. Other things seem to purely increase cost: starting meetings with smalltalk, overcharging for server hosting, and cronyism.