Second-Brain Speed Test

Brandon-Vault, read through the board-of-advisers CEO seat, where slow is a cost, not a feeling.

Measured 2026-07-08 · ~/Desktop/Brandon-Vault · every number below is a live disk read, not an estimate.

The finding in one line: the mind is fast, the body is heavy. Your knowledge layer, meaning notes, links, and live queries, sits well inside Obsidian's comfort zone. The vault feels heavy because 73% of its weight is reference PDFs stored inside it, plus a quarter-gig semantic cache. You are paying warehouse rent on a library you rarely open.

B−

Fast mind, heavy body. Core is green. Two fixable weights, a 1.6 GB in-vault PDF library and a 247 MB Smart Connections cache, are taxing every cold start, sync, and backup. Neither is knowledge. Both are optional.

2.2 GB

Total vault size

3,413

Working notes

9,334

Files indexed on open

127

Live query notes

5,461

Wikilinks in graph

Where the weight is

Books / PDFs

1.6 GB

Projects + Inbox

254 MB

Smart Conn. cache

247 MB

.obsidian + plugins

34 MB

Everything else

~65 MB

The 1.6 GB in Books/ is 77 PDFs (top three: 120 MB, 118 MB, 114 MB) serving just 15 markdown notes. Storage, not thinking.

The three seats scored

System	Reading	Grade	CEO note
Knowledge core	3,413 notes · 5,461 links	Green	Obsidian handles 10k+ notes. You are nowhere near the wall.
Live queries (Dataview)	127 query notes	Green	Only 4 scan `Books/`, and that folder is 15 notes. Cheap.
Plugin memory	13 plugins · 247 MB cache	Amber	Smart Connections + local-gpt load vectors into RAM on open. Justified only if used weekly.
Binary warehouse	1.6 GB · 77 PDFs	Red	73% of the vault, 15 notes. Drags every sync, backup, and search index.
Repo hygiene	no .git · 0 iCloud stubs	Green	Not doubled by git, not stalled by offloaded placeholders. Clean.

Top 3 fixes, ranked by payoff

1. Evict the PDF library

effort ~20 min · reclaims 1.6 GB · biggest ROI

Move the 77 reference PDFs out of the vault into a sibling _library/ that Obsidian excludes, or off to a T7 drive. Vault drops 2.2 GB → ~0.4 GB working. Cold start, sync, backup, and global search all get faster because Obsidian no longer stats and holds 1.6 GB it never parses.

2. Audit Smart Connections

effort ~2 min · reclaims 247 MB + startup seconds

The embedding cache is .smart-env at 247 MB, loaded on every open. If you are not running semantic search weekly, disable the plugin: the vault opens instantly and you keep /ask + rag-query (local ollama) for retrieval anyway. If you do use it, keep it, since it is earning the tax.

3. Cap the heavy dashboards

effort ~30 min · smoother editing

Two notes carry 10 live queries each (Three Layers.md, Christian Books Progress.md); the cockpit carries 6. A dashboard with 10 blocks repaints on every keystroke inside it. Split them, or add LIMIT and narrower FROM clauses. Low urgency today, but this is the pattern that bites at scale.

Bottom line for the CEO: the second brain is not slow to think, it is slow to lift. One 20-minute move (evict the PDFs) recovers ~73% of the weight and touches zero knowledge. Do that first; the rest is optional tuning.
Generated by the vault benchmark run · source: live du/find/grep over ~/Desktop/Brandon-Vault · 2026-07-08.

Is the big notebook fast?

Brandon has a giant digital notebook with 3,413 pages and thousands of links between them. We timed it.

Good news: the thinking part is quick. All those pages and links are easy for the app.

The slow part: he also stuffed 77 big PDF books inside the same notebook. They take up almost three-quarters of all the space but they are just 15 note-pages. It is like keeping 77 heavy textbooks in your backpack when you only read one.

The fix (about 20 minutes):

Take the 77 heavy books out of the notebook and put them in a box next to it.
The notebook shrinks from a 2.2 GB backpack to about 0.4 GB, five times lighter.
It opens faster, saves faster, and finds things faster.

Grade: B minus, a smart brain in a heavy backpack. Empty the backpack and it jumps to an A.