A well-tuned Obsidian vault is a beautiful thing. The app is local-first and durable, and it bends to whatever shape you want.

Add the right community plugins and a metadata scheme you stick to, and you end up with something close to a second brain.

But there's a trap in there, and it got me. Past a certain point, the maintenance eventually eats the work.

You have a barely shaped idea, and instead of writing it down or thinking about it thoroughly, you are picking a folder and assigning tags.

Early, half-formed thinking needs a different environment. So I set out to find one. Two months and a paid subscription later, I was home again.

When note-taking becomes database management

Every note wants a schema now

Bases is a core plugin that changed the rules for local Markdown. Your notes can be queried by property and displayed as Notion-style tables, cards, or interactive lists.

Most of the scaffolding we used to build by hand is now unnecessary. You can run a reading list or a personal task manager out of local text files.

But Bases only sees notes that carry the right metadata, which means a new note must be already structured to be worth anything in a database view.

Rough ideas are fragments, and a schema kills them before they develop.

Milanote lets me be disorganized on purpose

Two months on a pasteboard

My search for a replacement initially led me to Milanote, an environment built on the opposite principle. It's a visual workspace that behaves like a pasteboard.

Milanote gets capture right. Text cards, images, files, and links go wherever you drop them. A pasted URL becomes a rich preview on its own.

Nothing has to settle before it's ready. Cards sit scattered until a connection earns its place, and I can't fully explain why, but moving them helped me think in a way a linear document never has.

You put two cards side by side and see patterns that no backlink would have told you.

Then you group them, draw an arrow, or color code after the cluster looks like something. Nested boards go as deep as you need without the rigidity a file tree imposes.

Unfortunately, my time with it had to come to an end. Milanote needs a live connection to view or edit a board, and there's no true offline mode to fall back on.

Then there's the price. You get 100 combined cards and 10 file uploads on the free tier, and unlimited cards cost $10 a month when billed annually.

The phone experience is the other gap. There's a native Android app, and it does the job, more or less.

You can do quick notes, but nothing about it is worth writing about. In fairness, no canvas app I've used has fully cracked the mobile experience, and it's always best treated as a capture endpoint.

What $120 taught me about reading release notes

Back to Obsidian Canvas

Milanote asked me for $120. I said, let me think about it. Thinking about it led me to a comparison Chrome tab, and in that tab was the sentence "Obsidian has a built-in canvas." I closed the tab.

There's a specific flavor of stupid that hits when the free alternative you're searching for is already on your hard drive. When Canvas launched, I read the release notes, decided it was for mind maps, and moved on.

Little did I know everything I liked about Milanote was there. An infinite board that takes notes, images, audio, PDFs, URLs, and even whole web pages, arranged with cards, arrows, and groups.

Every trade-off I listed for Milanote reverses here. A board is a .canvas file sitting on my disk I can version like anything else. It works with Wi-Fi off. It's free.

Credit where it's due. Milanote's web clipper handles research with a single click, its link previews are better looking, and the product carries a polish a core plugin doesn't. Good, but not $120-a-year good.

Mobile is meh on both. I use Google Keep for capture on the go and move files through Drive when I need to.

Thinking and storing are different jobs

Thinking and storing are different jobs, and they want different tools. A permanent knowledge base needs structure so it stays searchable and relevant a decade from now.

Early-stage thinking needs room to be wrong and out of order. Force a rough idea to satisfy a database schema and the best ones will get abandoned before they develop into anything.

It's worth checking whether the tool you're about to leave has already solved this, so you don't repeat my mistake.