Common Mistakes That Cost Marks in GATE CSE and How to Avoid Them

Many of the marks lost in GATE CSE have nothing to do with hard questions. They go to predictable errors: a blind MCQ guess, one extra tick in an MSQ, a unit slip in a NAT answer, or a textbook reflex applied to a question that quietly changed the rule. Each one has a fixed correction. Learn the corrections, practise them in mocks, and you keep marks you have already earned.

Key takeaways

• Never guess an MCQ with no option eliminated. The expected value is zero. Do guess when you can rule out one or more options.
• Treat each MSQ as four separate true-or-false decisions. One over-selection loses the whole question.
• Answer every NAT and MSQ before time runs out, since neither carries negative marks.
• Unit slips are, by the book's analysis, the single largest source of lost NAT marks. Write the unit next to every number.
• Many subject traps come from a rule with a false converse or an unstated convention. Learn the rule, not the pattern.

Mistakes in how you attempt the paper

Guessing MCQs blind

Only MCQs carry negative marks: −1/3 for a wrong 1-mark answer and −2/3 for a wrong 2-mark answer. Here is the expected value of a guess:

A blind guess gains nothing on average. Guessing after eliminating even one option is worth it. Leaving a NAT blank while guessing a 2-mark MCQ blind has the order backwards. The NAT guess costs nothing, while the blind MCQ guess gains nothing.

Over-selecting in MSQs

An MSQ pays only when you pick exactly the correct set of options, with no partial credit. At least one option is always correct, and all four may be. Judge each option on its own and select only what you can justify. If one option cannot be decided, leave that question for last. The MCQ, MSQ and NAT strategy guide covers each type in depth.

Sinking time into one question

Spending more than five minutes on a single question in the first pass is the most common reason for an unfinished paper. Budget about 1.5 minutes for a 1-mark question and 3 minutes for a 2-mark question. If you have no plan within 60–90 seconds, mark the question and move on. Two 1-mark questions are worth the same as one 2-mark question.

Misreading under pressure

• Confusing "which is NOT true" with "which is true". Underline the negation.
• Reading only part of the last option. GATE often puts the deciding clause at the end.
• Changing a confident MCQ answer without a definite reason. Second-guessing under time pressure loses more than it gains.

NAT answer mistakes

NAT answers are typed on a virtual keypad with no options to check against, so small slips go uncaught.

• Units. Watch ms against µs, KB as 1000 against 1024, Mbps as 10^6, and bits against bytes. When a question fixes a convention, use it. When it does not, the usual defaults are 10^3 for data rates and 2^10 for memory sizes.
• Rounding. Carry full precision and round only the final answer to the stated decimals.
• Wrong quantity. Check you are giving what was asked: edges or nodes, faults or hits, waiting time or response time.
• Entry format. Type the number only, with a minus sign if it is negative. No units, commas or fractions.

The exam pattern and marking scheme article explains how each question type is scored.

Subject traps that repeat

Each trap below comes from GATE's recurring question patterns, with the rule that avoids it.

This table holds 21 traps. The book's last-minute revision sheet lists 85, subject by subject, along with 28 common mistakes, frequently confused concept pairs and edge cases. Its do-not-skip list also pairs each of its 160 concepts with the trap that most often costs the marks. See the full book.

When a familiar concept comes with a changed rule

Since 2021, GATE papers have often set familiar concepts in unfamiliar settings: a DP with an extra constraint, a graph algorithm with a modified rule, a scheduling policy with a twist. The mistake is to answer from memory of the textbook version. A safer routine:

1. Read the definition twice and restate it in one line.
2. Map it to the nearest concept you know, but do not assume the standard result applies.
3. Test the rule on the smallest case: n = 1, n = 2 or the empty input.
4. For a modified algorithm, run the modified rule on the given instance.
5. If it is still unclear after 90 seconds, mark it and move on. These are often 2-mark NATs with no penalty for returning later.

Some classic examples: Dijkstra can fail with a negative edge, but on a particular graph it may still be correct, so run it rather than reciting "fails". A syntax-directed translation may redefine operators or use right recursion, so evaluate it with the grammar's associativity, not arithmetic habit. And three-valued logic makes the intuitive answer to an SQL query with NULLs wrong, so evaluate row by row.

A small illustration

Here is a trap built on an off-by-one. Three processes each need at most 4 units of one resource type. How many units make deadlock impossible?

With 9 units, each process can hold 3 and wait for a fourth that never comes, so deadlock is still possible. The rule is n(k − 1) + 1 = 3 × 3 + 1 = 10 units. Stopping at n(k − 1) is exactly the mistake the question is built to catch. Many GATE traps work this way: the formula is familiar, and the marks go to whoever remembers the +1.

How to stop repeating the same mistakes

• Log every wrong, skipped and slow question in your mocks by subtopic, with the cause: trap, convention, unit, rounding, misread or time. After three mocks, the log is your personal revision list. The mock test strategy article explains the analysis in detail.
• Do every trace on paper. That means Gantt charts, K-maps, precedence graphs and cwnd tables, never in your head.
• State the counting convention before you count. Probes, comparisons and context switches all depend on it.
• In the last weeks, revise traps and formulas daily. The last-month revision plan shows how.

Accuracy work pays well. Turning three wrong 2-mark MCQs into skips gains two marks, and turning them into correct answers gains eight. The book's ten full mocks include a trap-heavy paper built around the misconceptions GATE re-tests (NULL in SQL, negative edges, Belady, MSQ over-selection, C precedence and IEEE-754 rounding), with full worked solutions. Get the book.