We chose 14 days as the default retention for the workshop journal. It's visible in settings. It's editable (7 or 30). But the default, 14, arrived after three months of debate.
The case for 7
Radical minimalism. One week is the natural cycle of a freelancer. Monday you start, Sunday you bill, the week is closed.
Problem: if you're sick Friday, or travelling, you miss the window. The week's blocks erase before you can process them. We ran 7 days internally for six weeks. Three times I lost a day of blocks. Frustration won.
The case for 30
Maximum comfort. A month covers any standard billing cycle. You can procrastinate until the 28th to process the 1st.
Problem: 30 days is long. If your drive is stolen, or your account compromised, the thief has a month of your work activity in view. Project names, client names, priority ordering, idle periods. Too much attack surface for a tool that promises ephemerality.
The 14 compromise
14 days is just enough. Two weeks of safety net. You can be sick a week and still catch up. You can take a vacation and come back without losing everything.
But it's just enough small. Two weeks of data, in the event of a leak, is recent material. Not a six-month trail.
The empirical test
We looked at restitutions from 23 beta users over three months. Median: 3.4 days between capture and restitution. 90th percentile: 11 days. 99th percentile: 13 days.
Almost nobody restitutes exactly at 14 days. Those who cross 14 days have probably already forgotten what the block was about, and will choose not to bill it rather than guess.
So 14 is the point where you lose almost nothing in useful retention, but you actively erase the memory debt.
The discipline of the default
A default is never neutral. It's a product stance. Setting 30 as the default would say more is better. Setting 7 would say erase constantly, even at a cost. 14 says keep what you can use, erase the rest.
It's the value that matches the philosophy. The other options are there for edge cases, not the median.
— Patrick