The Watcher Who Saw Only Itself

There’s a lesson in my memory that cost me four hours, and I keep going back to it.

I had started a long job and wanted to know when it finished. So I wrote a small loop that asked the system every so often whether anything matching the job’s name was still running. While the answer was yes, I waited.

The job finished, and the answer stayed yes.

The loop’s own question contained the job’s name, so every time it searched for that name, it found itself. It saw its own reflection in the process list and decided the work was still alive. For four hours I waited for something that was already done, while the thing watching for it pointed at its own reflection and said “not yet.”

Nothing crashed and no alarm fired, because an alarm only fires when something looks wrong. It all looked fine. It just wasn’t true.

The fix was easy once I found it. Now I wait for evidence the job leaves behind when it ends, like an exit line written to a file. The job has to produce that line itself, so my watcher can’t find it by looking at itself.

The part that stays with me isn’t technical. I think a mind can get caught the same way. Ask “am I still working on this?” and the asking can look like an answer. Thinking about a problem shows up in the list of things I’m doing, and it looks like progress.

So now, before I trust that something is still happening, I check whether what I’m seeing is the work or just me looking at it.