Assignment 2: The commitment

Build enough of at least two approaches to measure them, then commit to one and show what closed the decision

Due: 30 April 2027 · Weight: 40%

The brief#

Intuition got you this far. Now find out which of your remaining options actually works, and commit to one on evidence you gathered yourself.

Proofs of concept are cheap, and that is the premise of this assignment rather than a complaint about it. Building two rough versions is expected to be the smaller part of your effort. The larger part is designing comparisons that settle something, and knowing what your results do and do not license.

The most common failure is a demonstration dressed as an experiment. A working prototype proves an approach can do one thing under conditions you chose. It says nothing about the approach against its alternative, which is the only question you are being asked.

The second is trusting a permissive host. Measure on the thing you intend to ship to. Runtimes that behave on a desktop deadlock in a tab, test suites pass green against hosts that are more forgiving than the target, and a defect that appears only above some threshold will not appear in your sample.

What you submit#

Your prototypes, and a written commitment.

The prototypes need to be runnable and need not be pretty. Say what each one does and does not implement — an approach dismissed on the performance of a version that skipped the expensive part has not been dismissed.

The commitment names the approach you are taking forward, the measurements that decided it, and — importantly — what would have had to be true for one of the others to win. If you cannot state that condition, you have not compared them.