Tune
{% include "_nav.html" %}
{{ '%.1f'|format(per_day) }}notifications/day
{{ counts['alert:important'] }} importance · {{ counts['alert:interest'] }} interest ·
{{ counts['digest'] }} digest · {{ counts['drop'] }} dropped
Would notify
{% for a in alerts %}
- {{ a.lane.replace('alert:','') }}
{{ '%.2f'|format(a.score) }}
{{ a.headline }}
{% else %}
- Nothing clears these thresholds.
{% endfor %}
Would go to the digest
{% for d in digest %}
- {{ '%.2f'|format(d.score) }}
{{ d.headline }}
{{ d.reason if d.reason != 'digest' else '' }}
{% endfor %}
Sliders are a simulation only — nothing here writes to config.toml.
Three constraints compete, and which one actually binds is not obvious.
Measured on this corpus: at the shipped budgets (3 interest, 2 importance)
the budget binds and the subject quota is inert — moving the
quota from 1 to 3 changes nothing. Raise the budgets to 8 and the
quota takes over, at which point 1 / 2 / 3 yields 10 / 11 /
13 alerts. The threshold binds last in both cases: there is
almost always more above-threshold news than there are slots, so it works as
a quality floor for slow days rather than as a volume control.
So to get more notifications, raise the budgets first. To get more
variety at the same volume, lower the budgets and raise the quota.