Digital Technologies: applying concepts to unfamiliar scenarios
Digital Technologies is the single domain of the NZ Scholarship Digital Technologies examination, a written paper that asks you to apply core ideas - data representation, algorithms, systems, networks, security, and the human and social consequences of computing - to unfamiliar scenarios. Marks reward justified reasoning and trade-off analysis, not recall of definitions.
Core ideas to have ready
- Representation: everything reduces to bits; know why 8 bits gives 256 values, how fixed-width integers overflow, and how sampling (rate and depth) and compression (lossless vs lossy) trade fidelity for size.
- Algorithms: describe cost with Big-O; know that O(n log n) sorting beats O(n²) beyond a crossover point, and that hashing turns linear search into near-constant lookup.
- Systems and networks: layered models, latency versus bandwidth, and why the client–server boundary shapes security.
- Security: hashing is one-way (for stored passwords, salted); encryption is reversible (for data in transit). Authentication, authorisation and validation are different controls.
- Impacts: judge with evidence - stakeholders, access, privacy, sustainability - and propose mitigations.
Worked example
A sign-up form checks a new username against every stored name. With 50,000 accounts and 200 sign-ups a day, that is 200 × 50,000 = 10 million string comparisons daily - tolerable at microseconds each, but the cost per check grows linearly, so at five million accounts it is 1 billion comparisons a day. Reason aloud: current load is fine, growth is the risk. Replace the list scan with a hash set: one lookup per check, near-constant regardless of size. But hashing costs memory and needs collision handling, so state the trade-off rather than declaring a winner.
Traps that cost marks
- Answering "use the cloud" or "add encryption" without saying what problem it solves.
- Mixing up bits and bytes, or decimal (1000) and binary (1024) prefixes.
- Treating lossy compression as always wrong - for streamed audio it is the right call.
- Ignoring the stated constraint (budget, bandwidth, privacy) and the user who bears the cost.