Avoiding the turing tarpit: Learning conversational programming by starting from code's purpose

Conversational programmers want to learn about code primarily to communicate with technical co-workers, not to develop software. However, existing instructional materials don't meet the needs of conversational programmers because they prioritize syntax and semantics over concepts and applications. This mismatch results in feelings of failure and low self-efcacy. To motivate conversational programmers, we propose purpose-frst programming, a new approach that focuses on learning a handful of domain-specifc code patterns and assembling them to create authentic and useful programs. We report on the development of a purpose-frst programming prototype that teaches fve patterns in the domain of web scraping. We show that learning with purpose-frst programming is motivating for conversational programmers because it engenders a feeling of success and aligns with these learners' goals. Purpose-frst programming learning enabled novice conversational programmers to complete scafolded code writing, debugging, and explaining activities after only 30 minutes of instruction.