DSST Computing and IT common mistakes
DSST Computing and IT · Understand the traps and how to avoid them.
When asked which software license permits linking proprietary code with open-source components, the candidate answers from the label alone ("GPL means free, so anything goes") without tracing the copyleft obligation to the combined work.
Candidates memorize license names and slogans (GPL, LGPL, MIT, Apache) but never learn the distribution-trigger clause, so they cannot reason about derivative works or linking exceptions under exam pressure.
What to do instead
For each major license, write one line covering (a) whether source disclosure is required on distribution, (b) whether linking proprietary code is allowed, and (c) the patent/attribution terms. Practice classifying three concrete scenarios per license: static linking, dynamic linking, and SaaS-only use.
On a question about which development model fits an unstable, fast-changing requirements set, the candidate picks waterfall because the question mentions "documentation" or "phases," rather than matching the model to the requirement volatility described.
Candidates learn life-cycle models as a list of definitions instead of as decision tools, so any keyword in the stem triggers a reflexive answer.
What to do instead
Build a comparison table with columns for requirement stability, delivery cadence, customer involvement, and risk profile, then answer every scenario question by filling those columns from the stem before looking at the options.
When asked to choose between a relational database and a NoSQL store for a given workload, the candidate selects NoSQL simply because the data is described as "large" or "web-scale."
Marketing framing around NoSQL leads candidates to treat volume alone as the deciding factor, ignoring schema stability, consistency requirements, and query patterns.
What to do instead
Decide using three explicit tests: is the schema stable, are multi-record transactions required, and are access patterns known in advance. If all three are yes, relational is the default; justify any NoSQL choice with a specific access-pattern or scaling reason.
The candidate answers a privacy question by naming a regulation (GDPR, HIPAA) as the whole answer, without stating the specific obligation the scenario triggers, such as purpose limitation, data minimization, or breach notification timing.
Candidates study regulation names and penalties but skip the operative principles, so they cannot map a fact pattern to the rule that actually governs it.
What to do instead
For each major privacy regime, memorize the core principles as short imperatives and rehearse mapping a one-paragraph scenario to the single principle it violates. Always state the obligation, not just the statute.
On intellectual-property items, the candidate treats copyright, patent, trademark, and trade secret as interchangeable, for example saying an algorithm is protected by copyright rather than by patent or trade secret.
The four regimes are taught together and share the word "protection," so candidates collapse them instead of distinguishing what each protects and for how long.
What to do instead
Memorize a four-row table: subject matter protected, term length, registration requirement, and typical infringement test. Then classify ten concrete assets (source code, logo, invention, customer list) into the correct regime before answering any IP question.
When a question involves a network topology or medium choice, the candidate answers with the name of a technology ("fiber," "mesh," "Wi-Fi") without connecting it to the stated constraint such as distance, interference, or single point of failure.
Connectivity topics are learned as vocabulary lists, so candidates recognize terms but do not evaluate them against requirements.
What to do instead
For each topology and medium, note two strengths and two failure conditions. In practice, underline the constraint in the stem (distance, mobility, redundancy) and eliminate any option that fails that constraint before choosing.
The candidate answers netiquette questions by asserting personal opinion about what is polite rather than applying documented conventions such as quoting only relevant text, avoiding all-caps, and respecting list norms.
Netiquette feels like common sense, so candidates do not study it and default to modern social-media habits that differ from established network conventions.
What to do instead
Review the classic RFC 1855 conventions and note where modern practice diverges. For each item, identify the specific convention being tested rather than reasoning from personal preference.
On telecommunications-law items, the candidate conflates regulatory categories, for example treating a common carrier's tariff obligations as identical to those of an unregulated information service, or ignoring the distinction between interstate and intrastate jurisdiction.
Telecom regulation is layered and historically contingent, and candidates learn only headline concepts such as "common carrier" without the jurisdictional split that determines which rules apply.
What to do instead
Draw a decision tree: what service is offered, is it classified as a telecommunications service or information service, and is the traffic interstate or intrastate. Walk the tree for each scenario before reading the answer options.
When an AI question asks about a system's limitations, the candidate answers with the popular label ("hallucination," "bias") without identifying the mechanism or the appropriate mitigation for the described situation.
AI concepts are absorbed from news coverage, which supplies vocabulary but not causal understanding, so candidates cannot match a failure to its remedy.
What to do instead
For each AI failure mode, record the cause, a concrete symptom, and one mitigation (for example, retrieval grounding for fabricated citations, or representative evaluation data for bias). Answer by naming the mechanism and the mitigation together.
On globalization items, the candidate recommends a single global rollout and treats localization as translation only, missing legal, payment, date-format, and data-residency requirements.
Candidates equate globalization with language, because that is the most visible layer, and overlook the regulatory and infrastructural layers that drive architecture decisions.
What to do instead
Use a checklist of five layers for every globalization scenario: language, law and data residency, payments and currency, formats and calendars, and connectivity assumptions. State which layers the scenario stresses before choosing an approach.