Twenty-five years of notes, a 10 GB archive I never open, and what AI taught me about the real cost of "remember everything".
I have a bad memory. Many things simply don't stick, so I have always kept notes. I am also a systematic person: thinking feels good to me when concepts sit in a coherent structure I understand. Sometimes that is a strength - it keeps me focused, and I know what matters and what doesn't. Sometimes it is a weakness - I cannot give an expert judgment until I have turned an idea over in my head for a few days, or at least a few hours.
Before smartphones, my notes lived in thick paper notepads. Then Evernote released its Android app - I took the bait the very same day. Save anything that might ever be useful in one click - find it and use it the moment you need it. The promise was seductive: my own private Wikipedia, or at least a personal body of knowledge with full-text search and tags - everything paper could never give me. I believed that accumulating knowledge this way would not only make me faster, but would let me examine future problems from more angles. So Evernote became more than a place for current notes and tasks. It became my knowledge-accumulation system. For the knowledge I needed - or so it seemed at the time.
I left Evernote long ago, for a much more mature and secure tool: Joplin. Its creator clearly had Evernote users in mind and closed Evernote's obvious weaknesses. I migrated all my notes without any trouble and have not regretted it once in all these years.
But the trap - save everything that might be useful - moved with me, and it has been biting harder every year. To be clear, the problem is not the database. Mine is over 10 GB now, end-to-end encrypted, and Joplin syncs it without complaint - which says a lot about how well the system is engineered. The weak link turned out to be me. Me and my life.
Things that seemed important
Things that seemed important 25 years ago are not important anymore. Worse: they turned into noise that gets in the way of thinking.
Tags, my supposed safety net, did not help either. Say I want to recall how to analyze user registrations with Python and a Jupyter notebook. I type in the tags and get a pile of courses I once took and exercises that have nothing to do with my question - everything except the note I need. And when I do find a note I once considered important enough to save, its usefulness usually turns out smaller than I remembered. I would have solved the problem faster by simply solving it, instead of searching my past for a shortcut.
Here is a more painful example. I work with teams a lot: I assign tasks and track their execution. For years I believed it was essential to keep a short record of every meeting - who committed to what - so that before meeting each person I could quickly restore their list. It turned out to be my management mistake, and its name was micromanagement. If you delegate a task, you delegate all of it - without constant checks and reminders. People don't work that way, and frankly, I discovered that I like working with partners, not subordinates. So my careful record of meetings and decisions turned out to be a fiction too. The less recorded information I keep in front of my eyes, the more focused I am on what actually matters.
An archive I never open
Over time I had to split my notes into three notebooks: Work, Family, and Archive. Work and Family hold only what matters here and now; everything that "might someday be useful" goes to Archive. And you know what I noticed? My archive turned into a trash bin. I move finished and stale notes there and never come back. Worse: on the way in, I spend time cleaning them up - linking, structuring, fixing tags. I spend the time, and then I never return. So why keep them at all?
For several years now I have tried to stay brief and not multiply information I can get lost in. Still, my archive notes number in the thousands. A place I never visit, because there is never enough time.
I see the same pattern on the Joplin forum. Evernote users switch to Joplin and drag thousands of notes behind them - and then panic when their first mobile sync with encryption enabled takes hours, sometimes days. This is not really a Joplin problem: I remember Evernote itself barely coping with that volume. And I catch myself not even wanting to reassure them that only the first sync is slow. Because I can see they are carrying along the same broken approach to information that I carried - and, honestly, still carry.
Then AI sent me the bill
AI put this whole problem under a magnifying glass.
First, LLMs are graphomaniacs by nature. They produce a lot of text nobody needs, and we skim it, trying to fish the useful essence out of the stream of words.
Second, I connected AI agents to my notes: Claude Code and Codex work directly with my Joplin through alondmnt's joplin-mcp server. This should have been the moment my "second brain" finally started to work - an assistant that can actually read everything I ever saved. Instead, the archive started costing real money. The agent burns tokens reading text that never needed to exist - and still under-reads: it gives up halfway, skims the table of contents or the first paragraphs, and calls it a day. Every useless note I saved decades ago now slows down the very assistant that was supposed to put it all to use. My archive found a way to bite me here too.
What I save now
I have arrived at a simple conclusion: most of what seemed useful to me was garbage, and I should save far less than I thought. Saving must not work the way the web clipper trained us - "this might come in handy, click, maybe I'll come back to it later." No. Saving must be deliberate. A note has to pass through me first. And there is one test for every candidate: will I be able to find this - or reinvent it - within 15 minutes, even 20 years from now? If the answer is yes, the note does not get saved. That single question kills most notes before they are born.
Today all my notes fall into three categories.
Momentary. Joplin to-dos - as many as needed. Things to do and things not to forget: tasks with due dates plus a backlog without due dates, but with priority marks. They hang there until done, and once done, I delete them and never think of them again. Not for a report, not for the boss, not for myself. Done - forgotten. I used to save completed tasks in case a client or a manager asked what I had been doing all month. Here is what I learned: a good client or a good manager never asked, and a bad one was always satisfied with a short answer built from the time-tracker report. Only rarely does something from a to-do deserve promotion into a real project note.
Product. Joplin notes, with one rule: one project or product - one note. A very short description of what the project is, who the key contacts are, where the main documents live, what the roadmap looks like. I try to keep this note as short as possible. It still grows as the project moves along - but slowly, and everything in it has earned its place.
Archival. Old projects: only the essential, and only what cannot be found on the internet or made again. It turns out shockingly little passes that filter after 25 years of collecting. What survives is short and to the point. The biggest surprise was that in many cases even my own work did not deserve saving. The courses I took, the documents I prepared, the articles I wrote - either I can find or reproduce them whenever I truly need them, or they were garbage all along, and their place is the trash bin. What they taught me lives in my head anyway.
We don't need a second brain
For many of you, none of this is a revelation. You delete without fear, keep nothing extra, and feel no urge to systematize chaos. That is exactly the point I finally understood: people are different, and each of us needs our own system of knowledge, notes, and reminders. My mistakes are partly a projection of my own personality - the urge to systematize, to hold on to every useful thing. And partly they come from following a marketing slogan. While writing this article, I opened evernote.com to check: it still greets you with "Your second brain" and a promise to "remember everything and tackle any project."
Evernote homepage, August 2026
As it turns out, I don't need a second brain. I need the only brain I have - my own - working at its best. And "remember everything" is simply the wrong goal: it leads to a mountain of garbage that slows down the very work it promised to speed up.
Why I built Cockpit
That is why I built Cockpit, my plugin for Joplin. It shows, side by side, my to-dos grouped by date and my notes grouped by project. I can switch between project folders in one click or see all my tasks and notes at once, search quickly, and set up profiles for different working contexts. A flexible plugin for flexible work. And one feature is there by design: Cockpit can hide chosen notebooks completely. The first thing I hid was my Archive - so that it pollutes neither my head nor my AI agent's context.
And the Archive itself? I still cannot bring myself to delete thousands of notes in one strike. So cleaning it up became a project of its own: I chew through it slowly, with the help of the same AI agent, whenever I find the time - and every cleanup session eats into my LLM subscription. Think about it: saving all that garbage took one click per note. Getting rid of it costs real money. If that is not the true price of "remember everything," I don't know what is.
I would love to hear how you handle this. What do you keep, what do you delete - and what techniques keep you productive instead of wandering through the garbage of your own ideas and records?
Sources:
[1]: https://joplinapp.org - Joplin
[2]: Joplin Plugins - Cockpit - Cockpit, my Joplin plugin
[3]: GitHub - alondmnt/joplin-mcp: MCP server for the Joplin note taking app · GitHub - joplin-mcp server by alondmnt
[4]: https://evernote.com - Evernote homepage, accessed 2026-08-21
I add mainly data, which I processed myself before. So not a lot clipping or copying around just in case it might be useful one time. Most data is already „personalised“, because abstract info is implemented in an individual context. And deleting data is not a big deal for me, because I do it just when I stumble over data, which clearly is redundant - mainly because my personal context is gone - e.g. info about projects that are done, stuff, that has been abandoned, technical info, that is outdated… But I admit, the first stage, the processing part, is time consuming, so it might be the opposite of what many other users do. This is just my concept…
; it was incredibly convenient for filtering notes without having to manually type out every tag.
