Avi and Bugsby discuss Bugsby's data science process and brainstorm a design for a tool to improve it. There is a great deal of programming jargon and philosophy.

References

Command-line interface (e.g. bash)

Bugsby's workflow

• Gather data
• Process data with commands/scripts
• Feed processed data to model building
• Produce output which may get piped to something else

Python

Visual programming

XKCD - is time spent automating a task worth it 

Semantic web

GUI ("gooey")

"repl" (read-evaluate-print loop) command line

"typed"/"typing"

Caching

Pipes

Macros

Using Dynalist - store instructions in the form of a work log, be able to "tag" sections of it as "programs"

Unit testing

Users/groups/permissions

Transaction log (in the sense of database or file system)

Software liberal