Workflows by profession
A research Mac is a thousand PDFs named by publisher randomness, plus data that must outlive every laptop.
Reference managers (Zotero, BibDesk, Papers) solve this properly — they store, rename and index the literature. If you use one fully, its search wins.
The failure mode is the *partial* library: half the PDFs managed, half as 1-s2.0-S0167...-main.pdf in Downloads. Two rules close the gap:
2019 Smith - Mangrove economics.pdfZotero's stored files live under ~/Zotero/storage in hashed folders — findable by content search, not by browsing.
The question is always "which paper said this":
mdfind -onlyin ~/Zotero "phrase from the paper"
mdfind -onlyin ~/Documents/Papers "tidal subsidy"
Spotlight indexes PDF text (where the PDF has a text layer — scanned older papers need OCR first), which makes it a legitimate literature-search tool aimed at your own corpus.
Data outlives laptops or it was never data. Per project:
Project/
data/raw/ ← never edited, ever
data/processed/
code/
paper/
raw/ is append-only; processed data is reproducible from raw + code, which is what makes it safe to lose. Large raw data ages onto external drives — with the offline-search problem that implies, and the same solutions (listings, or an index that keeps offline volumes).
A paper is many files; name the root main.tex in a folder named for the paper, and search for the *folder*. Build artefacts (.aux, .log, .synctex.gz) are noise — .gitignore them, and a search tool that demotes rather than surfaces them earns its keep at every \include.
mdfind -onlyin over your PDF folder searches full text of every paper with a text layer; OCR the scans first.
~/Zotero/storage, one hashed subfolder per item — meant to be reached through Zotero, not browsed.
Everywhere keeps track of every file on your Mac — including the folders Spotlight hides and drives you have unplugged. Free for a day, then $19 once.
Download for Mac