The question didn’t sound particularly interesting. A developer wanted to know why one field behaved differently for a small group of employees. I was fairly sure we’d discussed it before, so I opened an old ticket, expecting to find the answer within a few minutes. Instead, the ticket pointed to another document. That document referred to a database column that no longer existed. Someone vaguely remembered discussing the change a couple of years earlier, although nobody could remember where the decision had eventually been recorded. By the time everything lined up again, most of the afternoon had disappeared.
The answer hadn’t been difficult. It had simply been difficult to find. After a while, I realised this wasn’t an unusual afternoon. It was how many projects quietly began. Before anyone designed a solution, wrote a query or opened an IDE, there was often a long period spent recovering what the organisation already knew. Searching through old tickets. Comparing different versions of the same rule. Asking around to understand why something had been built that way in the first place.
Curiously, nobody really talked about that part. Implementation was work. Testing was work. Writing documentation was work. Finding things somehow wasn’t, even though everything else depended on it.
The explanation that stopped fitting
For quite a long time, I thought the answer was documentation. If people kept searching for information, perhaps we simply hadn’t written enough of it. It felt like a sensible explanation, and it naturally led to an equally sensible solution: document more carefully, organise things better, keep everything up to date. It took a while before small contradictions started appearing.
Sometimes the documentation already existed. The ticket was there. The business rule had been explained. The decision had even been justified. People just couldn’t find any of it. Those moments were difficult to ignore because they didn’t fit the explanation I had been relying on. The knowledge hadn’t disappeared. It had become scattered, renamed, buried under newer discussions or disconnected from the place where someone naturally expected to look.
Without noticing, people often chose to rediscover something rather than spend another hour trying to locate it.
Experience began to look different
I used to think experienced colleagues moved faster because they knew more. Now I’m less certain. Many of them weren’t carrying dramatically more knowledge. They were carrying a map. They knew that one old project used almost the same business rule. They remembered which wiki page had quietly become obsolete. They knew that if a discussion involved occupational physicians, the real answer was probably hidden somewhere else entirely.
New people weren’t looking at a different system. They were looking at the same system without that map. That changed the way I think about onboarding. Learning a product isn’t only about understanding features. It’s also about learning where an organisation tends to leave its footprints. Some are in documentation. Others are hidden in conversations, habits, forgotten naming conventions or decisions that everyone assumes somebody else remembers.
The longer I work on complex systems, the more those invisible paths begin to feel like part of the system itself.
If I looked at the same project today
I don’t think my first question would be whether we need another document. I’d be more interested in something much simpler. If someone joins the team tomorrow, how long will it take them to find one answer they can genuinely trust? Not the quickest answer. Not the loudest answer. The trustworthy one. Because if that journey is confusing, another document probably won’t make much difference.
Making knowledge easier to discover feels like a quieter problem. It doesn’t usually appear on a roadmap, and nobody celebrates it in a sprint review. Yet it quietly shapes almost every feature that comes afterwards. I still don’t know whether every organisation eventually grows this kind of invisible maze. I do know that, whenever work starts feeling unexpectedly slow, I now spend a little less time asking what people know.
I’m usually more interested in what they can still find.