Over one week this summer I recorded the same idea three separate times. Not three related thoughts. The same idea, arriving each time with the small thrill of being new, each time slightly better argued than the last, and each time with no apparent memory that I’d been standing in exactly that spot four days earlier.
I do my thinking on morning runs and talk into a recorder while I do it, because otherwise it’s gone by the time I get home. So all three are on tape.
Then I went looking through my own files for something else and found it. The same idea, written out in more detail than any of the three recordings, in a plan I’d written two months before.
The obvious reading of that is wrong, and the wrong reading is the interesting part.
It isn’t that I hadn’t written it down. I do write things down. Notes, records, reflections, decisions, the reasoning behind the decisions. The capture side worked exactly as built. It caught the idea in June and it caught all three recordings faithfully.
What I had never built was any way of reading it back.
And here’s the part I find least flattering and most useful. I think the reason I built one half and not the other is that capturing feels like work and reading back doesn’t. Writing something down is satisfying. It produces an artifact, it looks like diligence, and you can point at it afterward. Going back through what you already wrote feels like it produces nothing. It takes real time, and most of what it turns up you half-remember already. The payoff is real, but it’s occasional and it doesn’t look like anything while you’re doing it.
One of those gets built. The other gets intended.
The write side working well is what convinces everyone the knowledge is safe. Nothing in the chain looks broken. The wiki is full. Every account in the CRM has notes on it. The retros happened and somebody typed them up. The vendor evaluation from two years ago is in a folder with the reasoning intact. All of the writing worked, which is exactly why nobody goes looking for the failure.
A record nobody reads isn’t knowledge, though. It’s a receipt.
And then a team spends six weeks arriving at a conclusion that is sitting, fully formed, in a document written by someone who has since moved to another department.
The tools have made this worse in a way that’s easy to miss. Capturing costs almost nothing now. Meetings transcribe themselves, notes summarize themselves, and the pile grows faster than it ever has. Nobody built the other half at the same speed, so the ratio keeps sliding while the dashboard keeps looking better.
So the question worth asking of your own organization isn’t whether you can find things. You can. You have search, you have dashboards, you have a quarterly review where somebody pulls the numbers.
The question is what reaches you without your having asked for it. Search requires you to already suspect the thing exists. Surfacing doesn’t.
Nearly every organization has the first. Far fewer have the second working, and the gap stays invisible precisely because search works. Most of the ones that tried to close it bought a tool and aimed it at everything they had. That fails too, just more expensively.
Either way, the fix is smaller than it sounds and it isn’t a new tool. Find a moment that already reliably happens… a Monday planning session, a standup, a monthly review, the briefing somebody already writes. Attach one act of retrieval to it. One thing surfaced from what you already have.
One. Not twelve. This matters more than it looks. A system that surfaces twelve things every morning gets skimmed and then ignored, exactly like the wiki did, and you’ll have rebuilt the original problem with better tooling. Restraint is the feature.
And the limit, because there is one and it changes the advice. I am not telling you to look everything up before you’re allowed to think. Re-deriving something is often how it gets better; the second pass is usually sharper than the first, and a team that checks the archive before it thinks will produce very tidy, very small ideas. Thinking it through again is fine.
The waste isn’t the re-thinking. It’s not knowing you’d been there before, so you begin at nothing instead of continuing from where you stopped. Three runs got me back to roughly where I already was in June. What I lost wasn’t the idea. It was two months of building on it.
A postscript, from October.
I wrote everything above in August. In September I built the other half.
It was a small piece in my AI workspace that, every time I gave it a task, noted which of my old notes it would have put in front of me. To keep it honest I had it show me nothing yet. It just logged what it would have surfaced, for me to read through once a week.
I never attached that weekly read to anything I already do. Nearly three weeks later I hadn’t opened it once, and I only learned it was there when my AI assistant mentioned it in passing.
I don’t think that’s a story about one tool. Looking back over the last few months, I suspect it’s a pattern. Some of my own routines depended on me remembering them, and while I kept building new things, they drifted. Nothing broke. Nothing told me. Building them was rewarding and keeping up with them wasn’t, which is the same imbalance this whole piece is about, one level up.
So I’m trying the advice on myself. This week I started actually running the routines I’d built and left alone. A maintenance pass every Monday, and a terminal setup I’d configured months ago and never switched to. The read now sits inside the Monday pass, and it shows me three things, not a log of thousands. I’ll be honest that I’m not sure yet it holds. Ask me in a month.
I’d be curious whether you’ve seen the same thing where you work… the tools built to keep a discipline, still running, and nobody looking at what they say.
That’s the thing an organization actually loses, and it never shows up as a loss. Everybody is working. Everything is documented. And the same ground gets covered again by people who would gladly have started further along, if anyone had told them they already had.