Well put -- I agree but would add two minor modifications for data papers. Start writing the Methods as soon as you start collecting data (or as you start to compile data) -- perhaps even before you have results in hand. This fast start not only reduces the risk that you forget to record some aspect of methodology (likely, especially for projects that may require months to complete), but also reduces the inertia that many beginning scientists feel when trying to jump start a paper.
Absolutely. In fact, something that I have seen happen all the time is we try to get all the methods cleaned up and documented at the very end of paper writing and then I notice some detail in how things were done and decide that it was done wrong and all the work needs to be redone with a different protocol. If we could avoid this trap that would be great. Writing up the methods early will help, as long as they're written carefully and in detail.
Well put -- I agree but would add two minor modifications for data papers. Start writing the Methods as soon as you start collecting data (or as you start to compile data) -- perhaps even before you have results in hand. This fast start not only reduces the risk that you forget to record some aspect of methodology (likely, especially for projects that may require months to complete), but also reduces the inertia that many beginning scientists feel when trying to jump start a paper.
Absolutely. In fact, something that I have seen happen all the time is we try to get all the methods cleaned up and documented at the very end of paper writing and then I notice some detail in how things were done and decide that it was done wrong and all the work needs to be redone with a different protocol. If we could avoid this trap that would be great. Writing up the methods early will help, as long as they're written carefully and in detail.