My research organization has gotten a little too byzantine over the years. Navigating to an active research project is now six levels deep on my computer. If I want to get to my book manuscript, e.g., I have to do the following: Dropbox/AcademicWork/Projects/Inprogress/mssBook Then, I still have to navigate within that folder between my archival material, grant reporting, the book proposal, and the manuscript. It’s a mess.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
AcademicWork/Projects/Inprogress

- IRJRF_CCES_2018
	- Accepted_Changes
	- Analysis
	- Data
	- Paper
	- Presentation
	- RnR_Notes
	- RnR_Letter
- Locke_Immg
- Mss_book
	- Archival_Materials
	- Archive_Staged
	- GrantReporting
	- Proposal
	- ResearchMaterials
	- RA_Hiring
	- mss
		- chapter01
		- chapter02

In the example above, I included the folder structure for another paper I am working which is in the final stages. Project directories are structured right. Lincoln Mullen’s tutorial recommends keeping all the associated files together. He is not alone. Folders like data, paper, and figures keep things neat and tidy. They especially make things reproducible because they create walls between things like raw data which should never be modified from the sausage of writing a paper.

The rest needs to be revised. How I got here is somewhat understandable. I got lazy. When I started graduate school, I worked primarily in Scrivener, which is still an amazing app, and a pot-pourri of iOS apps on my iPhone (and later iPad). Much of the complexity creep was abated even more (read: ignored) by moving the Inprogress folder in my Mac’s Finder sidebar. That made getting to an active project one level deep on my Mac and as simple as opening the app I needed to manipulate files on an iOS device. And once I got past my comp exams and started the disseration, I didn’t want to change horses midstream.

While writing my dissertation, I noticed an increasing amount of friction. I moved to a plain-text workflow during graduate school. Writing a makefile, e.g., was and is more challenging when the relative path between things is more complicated. Here, too, the complexity can be mitigated (read: ignored) through a clever use of prefixes and shortcuts within makefiles.

1
2
3
4
5

PREFIX = /Users/lmp/Dropbox/AcademicWork/Pandoc_Templates

BIB = /Users/lmp/Dropbox/AcademicWork/Bibs/refs.bib
BIB2 = /Users/lmp/Dropbox/AcademicWork/Bibs/primary.bib

I haven’t found a good set of best practices beyond Mullen’s tutorial and a few others. It was very helpful to re-read them after a few years. What I need, however, is something beyond mere guidelines and some working examples. In a few weeks I will update what I settled with, but for now, I just want to get a good statement of the problem.