7 Notetaking Apps Evaluated on One Question
Most researchers pick notetaking apps for search and storage. Here's how Obsidian, Roam, Notion, Zotero, Evernote, Bear, and Logseq actually perform for synthesis.
What if the app you've been using to manage your research is the reason your notes never become arguments?
Most researchers choose notetaking tools the way they choose filing cabinets: capacity, search speed, organizational flexibility. These are reasonable criteria for an archive. They are the wrong criteria for a thinking system. The question that actually predicts whether a notetaking app will make you a more productive researcher isn't "how well does it store?" — it's "how well does it help you build?" There is a difference between a literature review and a contribution, and the tool you use to connect your reading to your writing is where that difference gets made or lost.
This post applies one evaluative standard to seven of the most widely used research notetaking apps: does it help you think, or does it just help you store? The framework comes from Sönke Ahrens, whose How to Take Smart Notes (2017) remains the most rigorous treatment of notetaking as an intellectual practice rather than a productivity habit. Ahrens draws on Niklas Luhmann — the sociologist who used a Zettelkasten of 90,000 linked index cards to produce 58 books and over 550 scholarly articles across four decades — to argue that the slip-box is not "a tool for storing ideas, but for developing them" (Luhmann, 1981). That distinction is the entire review. Microsoft OneNote, dominant in institutional Windows environments, shares the storage-first architecture of Evernote and is excluded for the same reason: no structural support for synthesis.
Ahrens identifies three note types that matter for research:
- Fleeting notes — quick captures, meant to be processed and discarded
- Literature notes — what a source argues, in your own words, with citation
- Permanent notes — your own ideas developed in response to literature, written as if for a reader, linked to related permanent notes
The first two types are what most apps support. The third — the one that compounds into original argument — is where most apps fail, and where the real evaluation begins. "A note that is not connected to other notes will be forgotten," Ahrens writes. "Only if a note is connected to other notes will it be retrievable at all." The implication is architectural: retrieval requires connection, and connection requires a system designed around it.
There's a cognitive reason this matters beyond productivity. The generation effect — documented across 86 studies by Bertsch et al. (2007) with an average effect size of d = 0.40 — shows that information a learner produces through reformulation is retained significantly better than information passively read or highlighted. The friction of restating an argument in your own words is not inefficiency. It is encoding. Apps that make highlighting easier than reformulating are optimizing for the wrong thing. Apps that structurally require you to write your literature notes in your own words — and then connect them to permanent notes that build toward an argument — are doing the cognitive work of research, not just the logistical work.
Obsidian scores highest on the synthesis standard. Its core architecture is the bidirectional link: every connection you draw between notes is visible from both ends, and the graph view makes the structure of your thinking legible at a glance. For researchers willing to invest the setup cost, it's the closest available approximation of Luhmann's slip-box.
- ✅ Full Zettelkasten workflow (fleeting → literature → permanent)
- ✅ Bidirectional links, graph view, no imposed folder structure
- ✅ Large plugin ecosystem; free for personal use
- ⚠️ Steep setup curve — new users often spend weeks configuring before writing a note
- ⚠️ Zotero integration requires a third-party plugin with uneven results
Roam Research pioneered the bidirectional link before Obsidian made it mainstream. Its outliner structure maps naturally onto how arguments develop — thesis, sub-points, evidence, all collapsible and re-orderable. Block-level referencing lets you embed any note block in multiple contexts, powerful for researchers who need the same evidence to appear across different arguments.
- ✅ Block-level referencing — the most granular linking model in this list
- ✅ Daily notes enforce a capture-first, organize-later workflow
- ⚠️ $15/month — the highest subscription cost here
- ⚠️ Development pace has slowed noticeably since 2022
Logseq occupies similar territory to Roam — outliner-based, bidirectional links, graph view — but is open-source, free, and stores notes as plain markdown files on your local machine. For researchers with privacy requirements (clinical data, pre-publication material, institutional cloud restrictions), the local-first model is a meaningful advantage.
- ✅ Free and open-source; plain markdown storage
- ✅ Local-first — no cloud dependency
- ✅ Built-in PDF annotation in recent versions
- ⚠️ Mobile app lags significantly behind the desktop experience
- ⚠️ Plugin ecosystem thinner than Obsidian's
Notion is the most widely used knowledge management tool in academic settings, and the most overfit to the wrong use case. It is architecturally hostile to associative linking. Its "link to page" feature is a folder system with extra steps: pages live in hierarchies, and the mental model is taxonomic rather than networked. Andy Matuschak's framework for evergreen notes — concept-oriented, atomic, densely linked, accumulative — is nearly impossible to implement in Notion at scale. The pattern is predictable: an impressive-looking database that never generates unexpected connections.
- ✅ Exceptional for structured storage, databases, collaborative wikis
- ✅ Excellent for project management and team coordination
- ❌ No bidirectional links or graph view
- ❌ Taxonomic mental model — pages live in folders, not networks
- ❌ Not a thinking system. Do not mistake it for one.
Zotero is not a notetaking app in the synthesis sense, but it belongs in this list because many researchers treat it as one. It is the best reference manager available — citation metadata, PDF annotation, group libraries, browser import — and it is free. Zotero 6's integrated PDF reader improved the note workflow substantially. But its architecture is designed for bibliography, not argument-building. The workflow gap between where researchers source material and where they build arguments is the most consistent friction point in research practice.
- ✅ Best-in-class reference management; free
- ✅ Zotero 6 PDF reader with linked notes
- ✅ Group libraries and browser import
- ❌ No bidirectional links; designed for bibliography, not synthesis
- ❌ Should feed a synthesis system, not replace one
Evernote built the modern research notetaking category and then failed to evolve past it. Its search is genuinely excellent. But its mental model is the filing cabinet made digital: you put things in, you retrieve them by search or browse. Its 2023 acquisition by Bending Spoons brought pricing increases and feature cuts that accelerated the exit of longtime users.
- ✅ Excellent full-text search
- ✅ Cross-platform capture and web clipper
- ❌ No bidirectional links; no permanent-note layer
- ❌ 2023 acquisition by Bending Spoons brought pricing increases and feature removals
- ❌ For new researchers: no argument for starting here
Bear is the outlier: a writing app that researchers reach for because it's the most pleasant to use — and stay with longer than they should. Bear 2.0 added bidirectional linking, but the model is shallow — page-level only, no block referencing, no graph view.
- ✅ Best-in-class writing and capture experience
- ✅ Clean markdown, good tag system
- ⚠️ Bidirectional links added in Bear 2.0, but page-level only
- ❌ No graph view, no block referencing
- ❌ An excellent fleeting-note environment. Use it to write; use something else to think.
The practical workflow that emerges from this evaluation:
- Use Zotero to manage sources and PDF annotations
- Use the Zotero Better BibTeX plugin + Obsidian Citations plugin to pipe literature into Obsidian
- In Obsidian or Logseq, reformulate each source into a literature note in your own words, then link it to relevant permanent notes
- Treat Notion and Bear as output tools — project management and prose drafting — not thinking tools
There is a deeper issue that no app fully solves. Piolat, Olive, and Kellogg (2005) found that notetaking consumes roughly 0.78 of available cognitive resources — nearly the full capacity of working memory. This means that at the moment of capture, there is almost no cognitive bandwidth left for synthesis. The capture-now, synthesize-later model is not a workaround; it is the correct architecture for research cognition. But it only works if the "later" step is structurally enforced. Most apps make capture easy and leave synthesis to willpower. The Zettelkasten-influenced apps — Obsidian, Roam, Logseq — make synthesis structural by requiring every note to connect to another before it earns its place in the system.
Autogram approaches this same problem from a different angle: rather than requiring you to build the connection layer manually, it surfaces relevant notes at the moment you need them — when you're writing, when you're reviewing, when a new idea lands in the inbox. The architecture is different from a Zettelkasten, but the goal is identical: not a better archive, but a thinking system that works for the future self who has to produce something. Early access is open — join the waitlist.
The honest conclusion from evaluating seven apps against one question is this: most notetaking tools are optimized for the anxiety of losing information, not the ambition of building arguments. They solve the capture problem and leave the synthesis problem untouched. Ahrens' summary is the most useful single sentence for evaluating any notetaking system: "Most people use their notes as a filing system for the past rather than a thinking system for the future." Read that as a question before you choose your next tool. Then answer it honestly.