Notes as evidence: reading your own record six months later

There's a specific experience available to anyone who keeps a log, and it's the reason to keep one.

You're convinced a project has gone nowhere for months. You open the record expecting to confirm it, and find twenty-three entries, eleven of them substantial, including two decisions you'd entirely forgotten making.

The feeling that follows is disorientating. Both accounts came from you.

Memory reports the mood

Recall isn't a recording. It's a reconstruction, assembled at the moment of asking, and heavily coloured by how you feel while assembling it.

Which means that asking yourself how a project has been going produces an answer about your current state as much as about the project. Ask on a bad Tuesday and you'll produce a bad account, sincerely, and be able to supply supporting examples — because the examples that come to mind on a bad Tuesday are the ones that fit.

There's a further distortion specific to work. Difficult sessions are memorable and successful ones aren't. An evening where nothing worked leaves a strong impression; an evening where things went smoothly leaves almost none, because there was nothing to snag on. So the recalled version of a project over-represents the hard parts, systematically, for everyone.

A log doesn't do any of this. It was written on the day, by someone who knew what had just happened, and it says the same thing whatever mood you read it in.

This is why the record is worth most precisely when you trust it least. The instinct on a bad week is that the log will confirm the bad week, so there's no point looking. That instinct is the distortion talking, and it's the reason a lot of people keep records they never read — the reading gets postponed on exactly the days it would be useful, and on the good days there's no felt need.

What a record is for

Not nostalgia, and not accountability. Three practical uses.

Correcting the story. This is the main one. When your sense of a project and the record disagree, the record is usually right, and finding that out changes what you decide — whether to continue, whether you're capable of it, whether the approach is working.

Recovering decisions. Six months on you'll have forgotten which options you ruled out and why. Without the record you'll re-explore them, sometimes at length, and the re-exploration will feel like fresh thinking. Tried the queue-based version, doesn't handle retries saves an evening a year later.

Seeing patterns you can't see live. No individual entry shows you anything. Twenty read together show you that the work turned sideways in April, or that every gap follows a work trip, or that four months of sessions never once touched the difficult part. That's only visible in aggregate, and aggregate requires that the dull entries were written.

How to read it back

Not chronologically from the start, which turns into an hour of reminiscing.

Read the last twenty entries in one sitting and ask three questions. What was I actually doing — the verbs, not the intentions. Where did it change — the point where the character of the entries shifts, which is usually the interesting moment. And what's missing — the task that appears in none of them, which is nearly always the one that matters.

That third question is the one that pays. People are much better at noticing an absence when they're looking at twenty concrete entries than at any other time, and the absence is where the diagnosis lives.

Twenty minutes, every couple of months. It's a different activity from the weekly look at gaps, and it catches things the gap number structurally can't — a goal touched every week and going nowhere reads as perfectly healthy on the counter.

It also helps to read the entries for one goal at a time rather than everything in date order. Mixed together, twenty entries across five goals read as a busy period. Separated out, the same twenty might be sixteen on one thing and one apiece on the rest, which is a completely different picture and the one you'd want to act on.

The case against looking back

Two honest problems, and the first is real enough to take seriously.

Rereading your own record can be corrosive if you do it in the wrong frame of mind. A log read while you're already low becomes a catalogue of everything you didn't do, and the effect is the opposite of the one described above. The entries are neutral; the reading isn't. If you're reading it to confirm that you've been useless, you'll manage it.

The second is that a log encourages a particular kind of retrospective certainty. Reading back, the moment the project went wrong looks obvious, and it's tempting to conclude you should have seen it. You couldn't have — that's what makes it visible now, twenty entries at a glance, rather than one entry a week apart. Hindsight from a record is a genuine tool and a slightly unfair judge.

The mitigation for both is the same: read for what happened, not for how you did. The log is evidence, not a report card, and the moment it becomes a report card you'll stop writing honest entries into it.

Eleven entries in a bad February

She's six weeks into a difficult consulting job and by the end of February is certain the month was a write-off.

The log has eleven entries. Four are ordinary progress. Three are dead ends — a data source that turned out unusable, two analytical approaches that didn't survive the actual numbers. Two say nothing, too tired. One says nothing, waiting on the client's finance team. One says two hours reading around the sector, didn't touch the model.

Read together that's a month in which a bad data source was eliminated, two approaches were tested and discarded, a dependency surfaced, and there was a single evening of avoidance. Anyone else reading it would say the work was progressing normally.

Her memory said write-off, because the three dead ends were the memorable evenings and the four ordinary ones left no trace.

The practical consequence: she was about to tell the client the timeline was slipping. What she does instead is send the dependency — the finance team have been sitting on something for eleven days — which is the actual problem, and which she could not have identified from her impression of the month.

What the record and the number do together

The gap tells you a goal has gone quiet. It's a fast signal and it's honest, and it can be read in a second from the top of a list.

The notes tell you what was happening when it did, and what you'd concluded before you stopped. That's the slow signal, and it's the one that survives six months.

Neither is much use without the other. A number with no record tells you something is wrong and nothing about what. A record with no number sits in a file that nobody opens, because nothing brings you back to it.

That's the whole reason the two live on the same screen in goaldrift. The counter puts a goal in front of you; the notes underneath it are what you read once it's there. Neither requires you to remember to do anything, which matters, because a practice that depends on remembering is a practice that stops in the first difficult fortnight.

---

Related: Logging beats tracking · One line is enough · Review by gap, not by list