GA — the genetic algorithm

Planning your own week
is where studying dies.

GA stands for genetic algorithm — the part of this product that decides which lesson you study on which day. This page is about why that job is harder than it looks, and how it gets done.

Your free hours never line up with the days you meant to study.

You re-read the subject you're already good at, because it feels productive.

The one you're worst at keeps sliding to next week.

By the time the exam is close, half the syllabus has had no time at all.

What if something drafted 50 versions of your week and handed you the best one?

In plain terms

That's all this does.

  1. 1

    It drafts 50 rough timetables at once, all different.

  2. 2

    It scores every one of them against what a good week looks like.

  3. 3

    It throws out the weak ones and combines the strong ones.

  4. 4

    It does that 100 times, and keeps the best week that survives.

Watch it happen

One week, sorting itself out.

Press Run to let it evolve, or Step through a generation at a time. Watch the sessions move off the days this learner isn't free, spread across the week, and put the basics before the hard lessons.

Generation 1 / 100
MonBusy
MathQuadratics
MathExpressions
MathGeometry
TueBusy
WedBusy
ScienceGenetics
ThuBusy
FriBusy
RLAMain idea
SatBusy
MathRatios
SunBusy

A random first draft. Three Math sessions stacked on Monday, two more booked on days this learner said they're busy, and half the week empty.

41

Planned by hand = 45 — the black line.

Six goals

Covers your weak spots25%
Most time where you're weakest20%
Fits your free hours15%
Basics before the hard stuff15%
Rotates your subjects15%
Spreads the hard days out10%

An illustration of how a run behaves, using one example learner's week — not measured output from the engine.

What counts as good

Six goals, pulling against each other.

This is why one plain rule can't do the job. Spend every hour on your weakest subject and you stop rotating; rotate perfectly and you neglect the weak spot; do both and you may book a day you're not free. Each draft is scored on all six at once, and the weights decide which compromise wins.

Covers your weak spots

Anything under 60 out of 100

25%

Weighted highest because a plan that skips a whole lesson can't be fixed by doing the others well.

Most time where you're weakest

The lower the score, the more time

20%

Exam time is finite, so hours are worth more spent on a lesson you're failing than one you've already passed.

Fits your free hours

Never books a day you're busy

15%

A session booked when you're at work is a session that doesn't happen — an unrealistic plan scores nothing in practice.

Basics before the hard stuff

Expressions before quadratics

15%

Studying a lesson before its groundwork wastes the session, so order carries the same weight as fitting your calendar.

Rotates your subjects

Not one subject all week

15%

Mixing subjects across a week is harder in the moment but retains better than blocking one subject solid.

Spreads the hard days out

No two heavy days in a row

10%

Lowest weight — it protects against burnout, but it matters less than studying the right things in the right order.

Does it beat doing it yourself

The same week, two ways.

A plan a learner would plausibly write for themselves, next to the one the algorithm settles on — both marked by the same six goals.

Planned by hand

45
MonBusy
MathQuadratics
MathGeometry
TueBusy
MathRatios
WedBusy
MathExpressions
ThuBusy
FriBusy
RLAMain idea
SatBusy
ScienceGenetics
SunBusy
  • Two subjects never appear at all
  • Books Wednesday and Saturday, both marked busy
  • Quadratics lands before Expressions
  • Four Math sessions in the first three days

Planned by the algorithm

83
MonBusy
MathExpressions
RLAMain idea
TueBusy
ScienceGenetics
MathRatios
WedBusy
ThuBusy
MathQuadratics
SocialUS civics
FriBusy
RLAEvidence
ScienceEnergy
SatBusy
SunBusy
MathGeometry
  • All four subjects covered
  • Nothing booked on a busy day
  • Expressions comes before Quadratics
  • Heavy days separated, subjects rotating

+38 points — and the gap isn't a matter of taste. Every point of it is one of the six goals below being met or missed, counted the same way for both timetables.