IB Computer Science — Study Hub

Two written papers and an internal assessment at both levels — there is no Paper 3. Paper 1 covers theme A and finishes with a section on a pre-seen case study; Paper 2 is applied, asking you to trace algorithms, write code and design or normalise a database. At SL each paper is 1 hour 15 minutes and worth 35%, leaving 30% for the internal assessment. At HL each paper is 2 hours and worth 40%, so the same internal assessment counts for 20%.

May 2027. The redesigned guide restructures the course into two themes — A, concepts of computer science, and B, computational thinking and problem-solving — and drops the old optional topics entirely. Papers written for the previous syllabus still test many of the same ideas, which is why our past-paper coverage maps the older questions onto the new micro-topics, but machine learning is genuinely new and barely appears in them.

Yes — the 2027 syllabus is taught in either Java or Python, and you sit the version of Paper 2 that matches the one your school teaches. The two versions test the same syllabus points with the same marks, and the course content itself is not specific to either language, but the code you read and write in the exam is real code: the syllabus names the constructs by language, such as try/catch in Java against try/except in Python, ArrayList against List, and HashMap against dict. This is a change from the previous syllabus, which examined IB pseudocode instead.

HL runs to 240 teaching hours against 150 at SL and goes further in almost every topic — pipelining and CPU-versus-GPU in computer fundamentals, routing and network segmentation in networks, transactions and data warehousing in databases, and a much larger machine learning strand covering feature selection, classification, clustering and genetic algorithms. HL also studies one whole unit SL never meets, B4 abstract data types: linked lists, stacks, queues, trees and hash tables built from scratch. In the exams HL sits the same two papers for two hours each rather than 1 hour 15, answers four case-study challenge questions instead of two, and gets HL-only parts added to the shared questions.

Paper 2, for most students, because it asks you to produce rather than recall. Tracing an algorithm correctly means writing out every variable after every step instead of guessing, and writing code means deciding what goes inside a loop and what goes outside before you start — and in Python the indentation is that decision, so a misplaced line changes what the program does. The other common difficulty is justification: choosing a data structure, a topology or a normal form earns few marks unless you also say what would have gone wrong with the alternative.

Answer the command term exactly — outline wants the shape of an idea, explain wants the mechanism and its consequence. Show every conversion and every trace step, because method marks survive a wrong final answer. Test your algorithms on the boundaries: the empty list, the single item and the value exactly on the limit are where marks are lost. And whenever a question asks you to choose, justify the choice against the option you rejected.