PKM Systems That Actually Ship
Building knowledge systems that produce output, not just archives.
I’ve migrated my notes five times in the last four years. Evernote to Notion. Notion to Roam. Roam to Logseq. Logseq to Obsidian. Obsidian to… still Obsidian, but with three complete reorganizations along the way. Each migration took days. Each one felt like progress. Looking back, exactly none of them made me meaningfully more productive. They made me feel like I was being productive, which is a different thing entirely.
The personal knowledge management (PKM) world has a problem. It’s drowning in systems, frameworks, and tools, and the people most obsessed with PKM are often the people who produce the least from it. The person with the pristine Zettelkasten and the beautifully interlinked graph view and the custom CSS theme has, in many cases, published less original work than the person using Apple Notes and a word processor.
This post is my honest assessment of the major PKM systems, what they actually do well, where they fail, and what I’ve learned about building a knowledge system that ships artifacts instead of just accumulating index cards.
The systems: an honest map
Before we get into the tools, we need to talk about the systems, the organizational philosophies that determine how notes are structured and used. The tool is just the software that implements the system. Most people pick a tool first and then try to figure out a system. That’s backwards. Pick the system that matches how you work, then pick the cheapest tool that supports it.
Here are the major systems that matter in 2026, assessed by what they’re actually good at:
| System | Core philosophy | Best for | Worst for |
|---|---|---|---|
| Zettelkasten | One idea per note, linked by association | Academics, long-form writers, researchers | People who need project management |
| PARA | Organize by actionability, not topic | Professionals juggling multiple projects | Deep research with long time horizons |
| Building a Second Brain | Capture everything, organize later, distill for output | Generalists, content consumers | People who already know exactly what they need |
| Digital gardens | Think in public, notes evolve over time | Writers, teachers, public thinkers | Anyone who needs private working space |
| Evergreen notes | Concept-oriented, densely linked, written to last | Deep thinkers, interdisciplinary researchers | Quick project-based workflows |
Each of these systems has real value. Each also has a failure mode that its enthusiasts rarely discuss. Let me walk through them.
Zettelkasten: the intellectual’s seduction
I covered Luhmann’s system in my post on note-taking for output, so I’ll be brief on the mechanics. One idea per card. Explicit links. The system as a conversation partner. Luhmann published 70 books using this method. It clearly works.
But here’s what the Zettelkasten community doesn’t talk about enough: Luhmann was a full-time academic who spent his entire career working within a single (very broad) intellectual framework, systems theory. His Zettelkasten was optimized for a specific use case: generating sociology papers from a dense web of interconnected concepts accumulated over decades.
If that’s your use case, great. But most knowledge workers aren’t building a lifelong intellectual edifice. They’re trying to ship a design doc this week, prepare a presentation for next month, and maybe write a blog post when they have time. The Zettelkasten was not designed for this workflow.
The failure modes I’ve seen in practice:
Permanent note perfectionism. People spend 20 minutes crafting a single permanent note, agonizing over whether it’s truly “atomic” and whether the title properly captures a single idea. For a note they’ll probably never reference again. The system’s emphasis on note quality becomes a trap when applied indiscriminately to every piece of information.
Link addiction. The graph view in Obsidian (and similar tools) makes links feel like the point. People create links between every even slightly related note, producing a graph that looks impressive and is practically useless. A link that doesn’t generate unexpected juxtapositions is just metadata overhead.
The long game fallacy. “It’ll pay off in twenty years.” Maybe. But you’re trying to write a blog post today. A system that requires two decades of investment before it starts producing returns is a system designed for a different lifetime than most of us are living.
When Zettelkasten works, it works beautifully. I know researchers who’ve used it to produce genuinely novel interdisciplinary insights. But the gap between “using Zettelkasten” and “using Zettelkasten well” is enormous, and most people who adopt it end up with an over-engineered filing system instead of a thinking partner.
PARA: the pragmatist’s answer
Tiago Forte’s PARA system is the most practical of the bunch. It stands for:
| Category | Definition | Example |
|---|---|---|
| Projects | Short-term efforts with a deadline and a deliverable | “Write Q1 product roadmap by March 1” |
| Areas | Ongoing responsibilities you maintain | “Health,” “Engineering management,” “Personal finance” |
| Resources | Topics you’re interested in | “Machine learning,” “Typography,” “Fermentation” |
| Archive | Completed or inactive items from the above | Past projects, deprecated references |
The insight is organizational: sort by actionability, not by topic. A note about machine learning goes in Projects if you’re building an ML feature right now, in Areas if you manage an ML team, in Resources if you’re just interested, or in Archive if the project shipped. The same note can move between categories as your relationship to it changes.
PARA’s strength is its simplicity. Four categories. Clear rules for what goes where. No need for elaborate linking or graph views. You open your Projects folder and see exactly what you’re working on right now. That focus is genuinely valuable.
The failure mode of PARA is also its simplicity. The system doesn’t have a mechanism for generating new ideas. It’s organizational infrastructure, not thinking infrastructure. Notes go in, notes get organized, notes get used when a project calls for them. But there’s no equivalent to the Zettelkasten’s “follow the links and be surprised” mechanism. PARA organizes what you already know. It doesn’t help you discover what you don’t.
For someone who produces primarily project-based work (design docs, presentations, code, decisions), PARA is probably the right system. It won’t generate serendipitous connections. It will make sure you can find what you need when you need it, which is the more pressing problem for most professionals.
Building a Second Brain: the full framework
Forte’s Building a Second Brain (BASB) is the umbrella framework that PARA sits within. Where PARA is the organizational scheme, BASB adds the workflow: Capture, Organize, Distill, Express (CODE).
| Stage | What happens | Key practice |
|---|---|---|
| Capture | Save anything that resonates | Quick capture, don’t judge |
| Organize | Sort into PARA categories | Ask “what project does this serve?” |
| Distill | Progressive summarization | Bold, highlight, summarize |
| Express | Create something from your notes | Write, present, ship |
The Express stage is supposed to be the payoff. Everything before it is setup. But in practice, most BASB practitioners I’ve encountered get stuck in the Capture-Organize loop. They become world-class information collectors with a clean PARA structure and thousands of progressively summarized notes, and they never reach Express.
This isn’t Forte’s fault. He’s explicit about the importance of output. But the system’s mechanics work against it. Capture is easy and satisfying. Organize is easy and satisfying. Distill is moderately hard but still feels productive (you’re “improving” your notes). Express is hard, uncomfortable, and exposes you to judgment. The system creates a smooth on-ramp to the first three stages and a cliff at the fourth.
The honest assessment: BASB is a good system for information management. It’s a mediocre system for information production. The organizational infrastructure is sound. The output mechanism is underspecified.
Digital gardens: thinking in public
The digital garden movement, popularized by Maggie Appleton, Mike Caufield, and others, takes a fundamentally different approach. Instead of building a private knowledge system that eventually produces public output, you build the knowledge system in public.
Notes exist on a spectrum from “seedling” (rough, incomplete) to “evergreen” (polished, well-developed). Visitors navigate non-linearly, following links between ideas rather than scrolling chronologically. The garden evolves over time. Notes get updated, expanded, pruned. Nothing is ever “final.”
The appeal is real. Digital gardens solve the output problem by making the notes themselves the output. You don’t have to write a separate blog post; your notes are the blog. The incremental nature means you never face the intimidating blank page of a full essay. You just tend to your garden, adding a paragraph here, a link there.
The failure modes are subtle:
Public accountability without public quality. Calling something a “seedling” gives you permission to publish half-baked ideas. That’s liberating, but it can also mean your garden is full of fragments that never get developed because there’s no pressure to finish anything. The lack of deadlines and deliverables, the exact things that make gardens appealing, is also what prevents many gardens from producing substantial work.
Navigation is the reader’s problem. A well-written blog post guides the reader through an argument in a specific order. A digital garden says “here are 500 interconnected notes, good luck.” Some gardens handle this well with curated entry points. Many don’t.
The audience is smaller than you think. Blog posts get shared. Digital garden notes rarely do. The SEO characteristics are different (evergreen but hard to discover). The shareability is lower (who shares a single note?). If one of your goals is reaching an audience, the garden format works against you.
Digital gardens are great for a specific kind of person: someone who thinks publicly, updates ideas over time, and doesn’t need to produce polished deliverables. For everyone else, they’re a way to feel productive about note-taking without the discipline of finishing anything.
Evergreen notes: the high-effort option
Andy Matuschak’s evergreen notes are the most intellectually demanding system on this list. Every note must be atomic (one idea), concept-oriented (titled with a claim), densely linked, and written to stand alone. No topic notes. No quick captures that stick around. Every permanent note is a finished thought.
The payoff, when it works, is extraordinary. A dense network of well-formed ideas that can be recombined in unexpected ways. Matuschak’s own public notes demonstrate this: follow any link chain through his notes and you’ll encounter juxtapositions that generate genuinely new insights.
The cost is equally extraordinary. Writing a proper evergreen note takes 10-30 minutes. For a system with hundreds of notes, that’s hundreds of hours of investment. The processing overhead is the highest of any system. You can’t just dump notes in and sort later. Every note must meet the quality bar or it doesn’t belong.
This system is for people who are willing to invest that time because their work demands deep, interconnected thinking over years. Researchers, philosophers, interdisciplinary thinkers. For a software engineer who needs to ship a feature by Friday, it’s massive overkill.
The tool landscape in 2026
The PKM tool market has matured significantly since the Roam Research hype cycle of 2020. Here’s my honest assessment of where things stand:
| Tool | Philosophy | Strength | Weakness | Price | Local/Cloud |
|---|---|---|---|---|---|
| Obsidian | Markdown files you own | Plugin ecosystem, customization, longevity | Mobile experience, collaboration | Free (sync $8/mo) | Local |
| Notion | All-in-one workspace | Databases, team features, polish | Performance, vendor lock-in, search | Free tier, $10/mo pro | Cloud |
| Logseq | Block-based, open source | Daily journals, queries, outlining | Slower development, fewer plugins | Free | Local |
| Capacities | Object-based knowledge | Intuitive object types, clean design | Newer, smaller ecosystem | Free tier, $12/mo pro | Cloud |
| Tana | Supertag-powered knowledge graph | Powerful typing system, AI features | Steep learning curve, cloud only | Free tier, $16/mo pro | Cloud |
| Heptabase | Visual-spatial thinking | Whiteboards + notes, visual connections | Niche use case, newer | $12/mo | Cloud |
| Apple Notes | Just write things down | Speed, sync, simplicity, free | No linking, no plugins, no export | Free | Cloud (iCloud) |
The biggest shift since 2024 is AI integration. Tana, Notion, and Capacities have all shipped AI features that promise to surface connections, summarize notes, and answer questions about your knowledge base. Obsidian has plugins for this (notably Copilot and Smart Connections). My experience with all of them: the AI features are useful for search and summarization, but they don’t replace the thinking step. They can find related notes faster than you can browse, but they can’t tell you what the relationship means for your work.
The other trend worth noting: Obsidian’s dominance is solidifying. Its plugin ecosystem passed 2,000 plugins in 2025, the community is the largest in PKM, and the local-first architecture means your notes survive even if the company doesn’t. 2026 is shaping up as a big year for Obsidian’s plugin infrastructure, with planned improvements to plugin discovery and approval. For most people, Obsidian is the correct default choice. Not because it’s the best at anything, but because it’s good enough at everything and your data is never held hostage.
When PKM becomes procrastination
This is the section that will make some readers uncomfortable. Good.
There is a pathology in the PKM community that I’ve experienced firsthand and seen in dozens of others. It goes like this:
- You discover PKM and feel excited about the possibilities
- You choose a tool and a system
- You spend weeks setting it up, customizing it, reading about best practices
- You capture notes voraciously
- Something feels off, so you read more about PKM
- You discover a new system or tool that seems better
- You migrate
- Return to step 3
This cycle can repeat for years. I know people who’ve been “building their second brain” for four years and have yet to publish a single blog post, write a substantial document, or ship any creative work using their notes. Their PKM system is immaculate. Their output folder is empty.
The pathology is that PKM setup, maintenance, and optimization feel like productive knowledge work. You’re organizing information. You’re linking ideas. You’re building infrastructure for thinking. It looks like work. It feels like work. Your brain rewards it like work.
But it’s not work. It’s meta-work. It’s building the factory instead of building the product. And at some point, the factory needs to produce something, or it’s just an expensive hobby.
Here’s how I diagnose whether PKM has become procrastination:
| Signal | What it means |
|---|---|
| You spend more time organizing notes than writing from them | System is the product, not a tool |
| You’ve migrated tools more than twice in two years | You’re optimizing setup, not output |
| You can describe your system in detail but can’t name 3 outputs from the last quarter | Output gap is real |
| You read more about PKM than about your actual domains of expertise | Meta-work has replaced work |
| Your notes link to other notes but not to projects with deadlines | No connection to output |
| You’ve built custom themes, templates, or plugins for your PKM tool | Building the factory, not the product |
If three or more of these apply, you need to stop improving your system and start producing from it. Today. Not after one more reorganization. Not after migrating to the tool that will finally make it click. Today.
What actually matters (and what doesn’t)
After five years of experimenting with PKM, here’s what I’ve learned matters and what doesn’t.
What matters
A regular processing habit. Weekly. Non-negotiable. You sit down, review what you captured, and process the pieces that connect to active projects. Everything else can be imperfect. If you have this habit, your system will work. If you don’t, no amount of tooling will save you.
Output-orientation. Every note should be in service of something you’re producing. Not “might eventually be useful for something someday.” Connected to a specific project with a specific deliverable. If a note isn’t serving a project, it’s inventory. Inventory that doesn’t ship is waste.
Speed of capture. If it takes more than 30 seconds to capture an idea, you’ll stop capturing. Capture needs to be so fast that it doesn’t interrupt your flow. Daily notes work for this. Quick capture widgets work for this. Elaborate capture templates with metadata fields do not.
Longevity of data. Use a format and a tool that will survive your current enthusiasm. Markdown in a local folder will outlast any cloud service. If you’re in Notion and Notion shuts down, your export options are limited. If you’re in Obsidian and Obsidian shuts down, you still have markdown files that any text editor can open.
What doesn’t matter
The tool. Obsidian vs. Notion vs. Logseq. The differences between tools are tiny compared to the difference between “has a weekly processing habit” and “doesn’t have a weekly processing habit.” Pick one. Stick with it for at least a year. Stop reading tool comparisons.
Graph views. They’re pretty. They’re useless for actual work. I have never, in five years, used a graph view to find a note or make a connection I wouldn’t have made otherwise. They’re a visualization of your note collection, not a thinking tool. This is PKM’s most popular vanity metric.
Tagging taxonomies. I’ve seen people spend hours designing their tag hierarchy. Tags are useful for broad filtering (“show me all notes tagged ‘physics’”). They’re not useful for thinking. Links between specific notes are useful for thinking. Tags are organizational metadata. They feel like connecting ideas because they’re proximity-based, but proximity isn’t connection.
Backlink counts. A note with 50 backlinks isn’t more valuable than a note with 2 backlinks. It might just be a broad topic note that everything links to. What matters is whether the links represent genuine intellectual connections or just topical adjacency.
Templates. A template for your daily note is fine. Templates for every note type, with custom metadata fields and property schemas, is over-engineering. Most notes don’t need structure beyond a title and body text. If you’re spending more time filling out templates than writing, the templates are hurting you.
The system I actually use
I’ve settled on something I call “project-forward PKM.” It’s not a framework. It’s not branded. It’s just a set of habits that produce output.
Tool: Obsidian, with minimal plugins (Daily Notes, Templater for the daily note template, nothing else).
Structure: Three folders.
vault/
├── inbox/ # Daily notes go here. Temporary by definition.
├── projects/ # One folder per active project. Each has a _plan.md.
└── archive/ # Completed projects and orphaned notes.
That’s it. No Areas. No Resources. No elaborate PARA categorization. Three folders.
Workflow:
Monday through Saturday: capture into the daily note. Don’t organize. Don’t link. Don’t tag. Just capture.
Sunday: 60-minute processing session. Review the week’s daily notes. For each item, ask: “Does this serve an active project?” If yes, process it into a note in the project folder and link it in the project’s _plan.md. If no, leave it.
When starting a new project: create a folder, create _plan.md, pull in any relevant notes from the inbox or other projects.
When finishing a project: move the folder to archive. Leave the notes. If they’re useful later, you’ll find them via search. If not, they’ll rest peacefully in archive.
What I don’t do: Maintain a Zettelkasten. Build evergreen notes. Progressively summarize. Tend a digital garden. Read about PKM. Optimize my system. These are all legitimate activities for other people. They’re not compatible with my goal, which is to produce written and technical artifacts as efficiently as possible.
The uncomfortable conclusion
The PKM community has produced genuinely valuable ideas. Zettelkasten, PARA, progressive summarization, evergreen notes, digital gardens. These are real intellectual contributions that have helped people think more clearly and produce more work.
But the community has also produced an enormous amount of meta-work that displaces actual work. When the PKM Weekly newsletter has a larger readership than any artifact produced by its readers, something has gone sideways.
The test is always the same: what have you produced? Not “what have you organized” or “what have you captured” or “what’s your system.” What have you shipped into the world that other people can use?
If the answer is “a lot,” your system is working and you should ignore everything I’ve said.
If the answer is “not much, but my system is really well-organized,” then the system is the problem, not the solution.
The best PKM system is the one you barely think about because you’re too busy using it to make things. It doesn’t have a name. It doesn’t have a subreddit. It doesn’t photograph well for a YouTube thumbnail. It just works, quietly, in the background, while you do the thing you actually care about.
Stop building the factory. Start building the product.