Code-like steps

Claim: an algorithm's steps are written as constrained, code-like pseudocode, one statement per line, not as prose.

• Grade: measured, in old and small experiments.
• Evidence: pseudocode review, strand 1; against.
• Check: algorithm-specs (exact): the Steps hold a code-like block; whether its statements are constrained is judgment.
• Reader styles: a model as implementer keeps it: a written plan helps a model by a few points.

What it asks

• Use assignment, conditions and loops only, one statement per line, with names from the interface.
• Keep the steps free of any one programming language, so any implementer can follow them.
• Put the explanation beside the steps, not inside them.

Why

• Code-like design text beat plain English prose for coding and debugging time in two experiments with professionals, with a trend in a third (Boehm-Davis 1983). Part of the gap was naming: giving the prose the code's names shrank it.
• Detailed flowcharts gave no reliable gain over text for writing, reading, debugging or changing programs in the older studies of the same review.

Limits

• The experiments are old and small.
• Structured flowcharts beat pseudocode for comprehension in two independent groups (Scanlan 1989; Andrzejewska 2022), and the first one's stimuli were disputed: the comparison is contested.
• The notation must match the conventions its readers expect; five exam boards each publish their own.