Sampling with SQL

Based on what I glean about your scenario, I suspect you’d be better served by some kind of exponentially decaying weight based on age (i.e., age = current_timestamp ‒ created_at). For example, if you wanted to make your rows half as likely to be selected every time they grew another hour older, you could use `POW(2.0, -age / 3600)` as your weight, where age is given in seconds.

    duckdb> WITH Ages AS (
              SELECT age FROM UNNEST([0, 3600, 7200]) AS t(age)
            )
            SELECT age, POW(2.0, -age / 3600) AS w FROM Ages;
    ┌──────┬──────┐
    │ age  ┆ w    │
    ╞══════╪══════╡
    │    0 ┆  1.0 │
    │ 3600 ┆  0.5 │
    │ 7200 ┆ 0.25 │
    └──────┴──────┘