Ethan 01-10-2008

• Outline a clear plan with the team before starting work on something. Experiments should efficiently/elegantly test alternative hypotheses - classic article from Science magazine about efficient progress in science
• Spend time in lab, and spend it efficiently: Stick to the plan, but stop and fix things that need fixing (don’t keeping working on half-working equipment).
• Tell somebody whenever your equipment does something you don't expect. This will supplement your previous equipment training, improve everyone's safety, and minimize broken equipment.
• Keep a lab notebook, essentially a research diary, that clearly records everything that you try. Anyone should be able to open your lab notebook and find out what trials/experiments you did yesterday, and what conclusions you drew. This is an important part of working efficiently. Without good records, you will end up doing the same experiment again later. With good records you will see trends that emerge over the course of many experiments (see data management below).
• Communicate your conclusions clearly using the data that supports your conclusion. For example
  • email jpegs of important AFM images to everyone involved
  • add corrections and comments to wiki recipes as you learn more

It is exciting to build devices that no one has built before. Or make new scientific instruments, or develop new processes. But sometimes the temptation leads to unnecessary frustration. Perhaps a tried and true method would suffice? Trade offs between risks and benefits will always be tricky.

To make a logical/realistic decision about the best path forward, consider this rule of thumb:

• If I am following a recipe that has been tested by N people, the probability of failure on the first attempt is ~ 0.5^(N+1).

Example 1: You are pouring a concrete slab. The recipe on the bag of concrete has been tested by hundreds of thousands of people. The probability of failure is basically zero.

Example 2: You read a paper explaining how to grow high density aligned CNTs. The recipe hasn't been tested by any other groups. The probability of failure on the first attempt is ~ 0.5.

Some common sense rules of thumb about organizing a meeting or event by email:

• Choose a time/date far enough in advance that you can receive responses from people (assume that people check their email once a day, 5 days a week)
• Put as much info into the email as possible (rather than sending multiple emails and/or last minute details).
• If you don't get an response, follow up with a second email.
• To organize many people try making a poll to find a common time

Some things are better done face-to-face:

• Asking someone to be on your committee.
• Asking someone to cover your TA office hours.
• …