I built a to-do list that is not allowed to prosecute me.
My previous lists kept trying. In the morning, each one recorded what I intended to do. By evening, it presented the same plan as evidence of what I had failed to finish. The day had changed. The task model had bravely preserved none of that information.
I built Leslie around a simple distinction: what you planned and what actually happened are both true, but they are not the same kind of truth. One is an intention. The other is evidence. A checkbox should not be asked to impersonate both.
Intent is not evidence
A plan points forward. It can have a date and an expected duration, but it remains an intention. A work log points backward. It says that something happened, when it happened, and where its notes belong. Time gives them opposite jobs.
In the task apps I used, a checkbox told me only whether something was open or complete. Partial and unplanned work had to hide in notes or disappear. The tiny square had acquired unreasonable authority.
Leslie keeps planned work beside a factual timeline. Completing a plan can create a log, but you can also record work that was never planned. The plan remains useful. It just stops being the sole witness.
A missed plan is ambiguous
A date passing proves one thing: the date passed. An unfinished item may mean you were blocked, chose another priority, made partial progress, forgot it, or decided it no longer mattered. Software cannot identify the true explanation by consulting the calendar with more confidence.
Automatic rollover looks helpful because nothing gets lost. It also turns every old intention into today’s obligation. The software has renewed a promise on your behalf and helpfully placed it back on your desk. Repeat this long enough and the backlog becomes a history of automated contract extensions.
What the research changed
The research did not produce a magical task app. It constrained three decisions, which was more useful:
- Keep the main path short. W3C guidance for helping users focus recommends short critical paths, limited interruptions, and removing unnecessary content. Leslie puts planning and logging in one composer. Neither path receives a ceremonial dashboard.
- Record progress as timeline entries, not points. A meta-analysis of 138 studies and 19,951 participants found that interventions designed to increase progress monitoring also promoted goal attainment. That supports making progress easy to record. The study did not compare timelines with points or streaks; I chose a factual timeline. The points system was never on trial.
- When a plan’s date passes, ask what happens next. In a diary study of 1,823 intentions, forgetting accounted for about 10–14% of non-completions, while situational and motivational reasons were more common. Four experiments on policy defaults found that people can read a default as a recommendation. Leslie therefore asks for an explicit choice when a plan no longer fits. It does not quietly vote on your behalf.
None of these studies tested Leslie. They gave me constraints for the first version, not a victory speech. Only regular use can show whether keeping plans beside a factual log helps anyone return to the list.
The sources, limits, and proposed interaction for automatic rollover are in Leslie’s plan-deferral research note. The current model keeps plans separate from factual logs; rescheduling is explicit, and completion is never inferred.
An estimate is not a promise
An estimate is useful in the same way a weather forecast is useful. It can shape the day. It should not become a sworn statement after the day changes. In Leslie, partial progress becomes a factual log. Yesterday stays visible, while today begins with an explicit choice.
Progress appears as timeline entries, not streaks, scores, or red overdue counts. The record does not need special effects to be true.
Leslie’s composer asks one of two questions: what needs doing, or what did you just do? Both answers land in the record where they belong. The checkbox has been relieved of its judicial duties.
Local-first is part of the design
A factual work log can become a private record of attention, energy, and unfinished thought. The current beta stores that record in a local SQLite database. It has no account and no cloud sync. The private diary of your attention remains on your computer, where diaries generally have excellent instincts about living.
Leslie is open source under MIT. You can inspect how it works, change it, or maintain your own version. If I stop maintaining Leslie, you can keep the software running. My future motivation is not a storage format.
Download Leslie v0.2.1 for macOS. The public beta is ad-hoc signed and not notarized. It has no cloud sync or automatic updates. You can also read the source on GitHub.
I do not want a to-do list to win an argument against your past self. It has a database, a calendar, and no sense of mercy. This is an unfair matchup. I want it to help you return to the present with a better record of the day you actually had.