Learning new things in Obsidian (2024): In defense of remembering

Forgetting may feel natural, but the path to real learning depends on building durable traces of what was learned—so ideas can be tested, connected, and turned into work. The core claim here is a four-part defense of remembering: complexity requires permanence, learning is deliberate practice rather than passive absorption, knowledge work eventually demands concrete results, and forgetting quietly locks people into “cultists of the new” by amplifying availability bias.

The argument starts with complexity. To grow from simple understanding into more complex thinking, past knowledge has to remain accessible. That permanence comes from remembering and writing things down, which lets new ideas stack on top of older ones instead of starting from scratch each time. The process described is intentionally selective: it doesn’t try to capture everything consumed, but it favors sources that make saving highlights easy. Articles and presentations get captured through the Readwise Highlighter browser extension, which can automatically save highlighted text and bring it into a Readwise workflow. Because Readwise supports notes and comments, the captured material can be written using Obsidian syntax so links and structure carry forward later.

Books are handled with a similar pipeline. Fiction is often read without notes, but nonfiction is usually annotated. When deciding what to read next, Shortform is used for one-page summaries; those summaries can also send highlights into Readwise. Reading then happens across Kindle and Kobo, depending on language needs and DRM constraints. Both ecosystems sync well with Readwise, keeping highlights centralized.

Podcasts and YouTube are folded into the same system through tools that convert audio into text. Snipd is used for podcasts because it generates AI summaries and transcripts, which then become highlightable content that flows into Readwise. For YouTube, a plugin sends videos to Readwise using available captions, turning them into text that can be annotated.

Readwise functions as a hub: it collects highlights from books, articles, podcasts, and videos, then routes them into downstream tools like Obsidian and Notion. Reader is described as a staging area for newsletters, PDFs, articles, RSS feeds, and emails, with YouTube captions also transformed into highlightable material there. Not everything makes it through every step, but the system is designed so that important items do.

The second defense point reframes learning itself. Learning isn’t treated as memorizing someone else’s claims; it’s treated as deliberate practice—holding one memory up against other memories, interrogating meaning when ideas sit side by side, and restructuring how concepts connect in the mind. For items that don’t immediately justify full processing, Napkin adds an AI tagging layer on top of Readwise highlights, using a daily mix and progressively improving tags to surface related ideas.

Once highlights reach Obsidian, they become raw material for structured thinking. The workflow includes a “Hegelian dialectic” note style—thesis, antithesis, synthesis—to steelman an author’s premise by testing it against counterarguments and then recombining insights. The third point argues that forgetting blocks production: if someone builds apps, writes code, deploys systems, or creates content, vague impressions won’t substitute for remembered steps and specific facts. The final point warns about cognitive distortion: when memories fade, availability bias takes over, making opinions and thoughts disproportionately shaped by the most recent inputs rather than the full historical library. In that sense, remembering—paired with writing—becomes the antidote to passive “read and forget” cycles and supports the focused attention associated with deep work.