Why Traditional Coding Classes Fail and How “Prediction‑First” Can Transform Learning

Traditional programming courses often follow a repetitive pattern: introduce terminology (variables, arrays, objects), show example code, and ask students to copy the code until the syllabus is completed. Jeff Olsen argues this method is outdated and proposes a more active learning strategy.

Prediction‑First Teaching Method

Olsen calls his method “Prediction‑First”. Students are first given a short code snippet and asked to predict its behavior before seeing the output. This forces them to think about language features and encourages discovery through experimentation.

He outlines five standards for effective coding exercises:

Can a student who does not yet know the syntax infer the purpose of the code?

Does the snippet resemble problems they might encounter later?

Are variable names meaningful and correctly chosen?

Does the task focus the learner on the most important part of the code?

Is the example understandable without requiring extensive prior knowledge?

Instead of presenting concepts first, instructors pose a challenging problem that incorporates the day’s learning objectives. Students then create or modify code to solve the problem, learning terminology and constructs in context.

Concrete Examples

Hello World example:

name = "Tamara"
print("Hello" + name)

Students predict the output ("HelloTamara" without a space) and then explore how to adjust the code to include a space, thereby learning string concatenation and the print function without explicit definitions.

1. Conditional Statements

Traditional example:

if 1 > 2:
    print("1大于2")
else:
    print("1不大于2")

Olsen replaces it with a scenario involving age verification, prompting learners to predict the output and answer follow‑up questions about the missing number, how to change the message, and the role of indentation.

age = 15
if age >= 18:
    print("你可以购买R级电影票")
else:
    print(f"再过{18 - age}年你才满18岁")

Students answer: the program prints “再过3年你才满18岁”.

2. Arrays

Traditional example prints the third element of a list:

a = [3, 2, 7, 5, 3, 9]
print(a[2])  # returns 7

Olsen rewrites the example with a list of foods, then asks learners to predict the output, retrieve the first and last items, and consider what happens when an out‑of‑range index is used, thereby exposing array indexing and error handling.

favorite_foods = ["寿司", "玉米粽子", "披萨", "烤鸡肉"]
print(favorite_foods[2])

3. Objects

Traditional class example uses a Dog class with a speak method. Olsen replaces it with a realistic User class containing email, age, and password fields, then asks questions about specific attribute values and program output, guiding learners to understand object creation and attribute access.

user1 = User("Ty", "[email protected]", 16, "Pa$$word")
user2 = User("Sarai", "[email protected]", 15, "Kangaroo!")
user3 = User("Osu", "[email protected]", 12, "12345abcde")
print(user2.age)

Through these problem‑driven examples, students actively engage with concepts rather than passively copying code.

The article concludes that the “prediction‑first” approach benefits both teachers and self‑learners by fostering deeper comprehension and encouraging learners to think critically about code before execution.