What people say they learned

Retrospective accounts are unreliable in known ways, and they are still worth attending to when the same observations recur across a lot of them.

The recurring observations

The problem was visible early. Almost universal in these accounts, and it should be held loosely.

Hindsight makes past evidence seem more obvious than it was, which is a well-documented bias.

What survives the discount is narrower and still useful: the specific thing that ended the relationship was frequently present in mild form at the beginning, and was interpreted as a minor feature rather than a preview.

It was never said clearly. The most common regret in accounts of relationships that ended by drift.

People describe years of hinting, of raising things obliquely, and of concluding that the other person should have known.

The small things were the thing. Almost nobody describes a relationship ending because of a single event. They describe accumulation.

Which is consistent with the observational research, where the ratio of everyday warmth to everyday friction predicts more than any individual conflict.

Staying too long is the more common regret. In accounts of both marriages and long relationships, people more frequently describe having remained past the point of usefulness than having left too early.

This is subject to obvious selection — people who left early and regretted it are a different population — and the pattern is consistent enough to note.

The aftermath was less catastrophic than expected. Which matches the research on affective forecasting.

What people say they would do differently

The list is short and unspectacular.

Raise things when they are small.

Ask directly rather than infer.

Pay attention to how the other person behaves when there is a cost to behaving well.

Keep their own friendships.

Take the early pattern seriously rather than assuming it will change.

Notice the difference between wanting the relationship and not wanting to end it.

The pattern question

The most useful thing available in a retrospective, and the one people avoid.

Looking across several relationships rather than at one: what recurs?

The same complaint from different partners. The same point at which things deteriorate. The same type of person chosen.

Where a pattern is present across relationships with different people, the common element is you, and that is information rather than an accusation.

It is also actionable in a way that an account of one relationship is not.

The limits of hindsight

Worth stating, because retrospective accounts are treated as more authoritative than they are.

Memory is reconstructive, and the current state shapes the account of the past — measurably, in studies where earlier reports exist.

People in distress recall their relationships more negatively than they described them at the time.

Which means the confident account of what went wrong, delivered a year later, is partly a construction, and treating it as an accurate diagnosis is a mistake.

The parts worth trusting are the factual ones: what was said, what happened, what was never raised.

The thing nobody says

One observation that appears in these accounts and rarely in advice.

A relationship that ended is not thereby a failure. Some were right for a period and not for a life, and describing them exclusively in terms of their ending discards most of what they were.

People who describe their previous relationships this way — as things that were good and then were not — generally seem to have adjusted better than those who describe them as wasted time.

Which may be a consequence of the adjustment rather than a cause of it, and either way it is the more accurate description.

The thing people are glad about

Less commonly reported and worth including.

Accounts of ended relationships are not uniformly regretful. People frequently describe things they would not have had otherwise — children, a place they lived, a capability they acquired, a friendship with someone they met through it.

They also describe having become different people, in ways they endorse, partly through the relationship and partly through its ending.

Which is worth setting against the framing of a failed relationship as time lost, since that framing is both common and, on most accounts, not accurate.