Common mistakes

On Friday, April 23, Owen took the practice assessment a second time, after the seven domain reviews. He scored 44 of 50, up from 37. More useful than the score was his error log, which now held every miss and every lucky guess from both attempts, 21 entries in all. Priya asked him to sort them not by area but by cause. The result surprised him: only six entries were things he simply had not known. The other fifteen were mistakes of a few recurring kinds: confusing two similar features, misreading the question, and overthinking. This lesson collects those kinds of mistake, with the pairs of features candidates confuse most often, and the habits that prevent them.

Why candidates lose marks

A foundations exam rarely fails people on obscure facts. It catches them on three things. Confusable pairs: two features that sound alike, where the distractor is the one that fits a slightly different situation. Misreading: missing a "not", choosing one answer where two are required, or answering the question you expected instead of the one asked. Overthinking: inventing conditions the question does not state, such as assuming a company must be on an Enterprise plan, and talking yourself out of the plain answer. Owen's log showed all three.

The confusable pairs

Priya and Owen built a single table of every pair that had caught him or that she knew catches people. Each row gives the one sentence that separates the two. Owen read it aloud each morning for the last week.

Reading the question

Five of Owen's misses were reading errors, and they followed patterns that exam writers use deliberately. The fix for each is a reading habit, not more knowledge.

Owen's worst example was a question asking which features are not free for private repositories on GitHub Free. He read it as "which features are free" and chose exactly the wrong set. After that, he adopted Priya's habit of restating every question in his own words before looking at the options: "So they want the ones that cost money on private repositories."

Overthinking

The opposite failure is reading too much into a question. Foundations questions describe simple situations, and the answer is usually the most direct feature. If a question does not mention an enterprise, do not assume one. If it says a team wants to discuss an idea, the answer is Discussions, not a project with custom fields. If two answers seem right and one is simpler and fully meets the need, the simpler one is usually intended. Owen's two overthinking entries were both cases where he chose an elaborate answer, rulesets with bypass lists and a GitHub App, when the question needed only branch protection and a webhook.

Guessing well is a skill too. When Owen genuinely did not know an answer, he eliminated options that contradicted something he did know, such as an option claiming github.dev runs tests or that a secret gist is private, and chose among the rest. Removing even one impossible option from four raises the chance of a correct guess from one in four to one in three, and removing two raises it to one in two.

Time mistakes

Time is the other source of lost marks. The commonest pattern is spending four or five minutes on one hard question early on, then rushing the last ten questions, which may have been easy. The defense is the pacing plan from lesson 2: answer, mark and move on when a question takes more than about two minutes, and check progress at the checkpoints. The second pattern is changing answers on the review screen without a reason. Research on exam behavior suggests that changed answers are more often changed from wrong to right than the reverse, but only when the candidate has a specific reason, such as noticing a missed word. Changing an answer because of a vague feeling is closer to guessing again.

One typical trap in each area

Priya ended the session by naming the single trap she sees most often in each area of the outline. Each is a confusable pair or a misreading in disguise, and each has a quick check that settles it.

The quick checks are worth practicing until they are automatic, because under time pressure the brain reaches for the familiar word rather than the right one. Owen found that saying the quick check silently before choosing an answer took two or three seconds and removed most of his confusable-pair errors in the final practice set.

A final category deserves a mention: questions about features that change quickly, above all Copilot plans and features. Here the mistake is usually trusting a memory of last month's interface over the documented purpose of the feature. When a Copilot question seems to depend on a detail Owen was unsure of, his rule was to answer from what the feature is for and how it is offered, individual or organization, and to move on rather than burn time trying to recall a menu.

Mistakes outside the exam room

Some mistakes happen before the exam starts. Candidates book with a work account and lose the record, fail the online system test on the morning, leave notes on the desk and have the exam ended by the proctor, or arrive at the test center with an ID whose name does not match the booking. Lesson 12 turns these into a checklist. Owen's rule was to treat the logistics as part of the exam: checked once a week before, and once on the day before.

Turning the error log into practice

An error log only helps if it changes what you do. For each entry Owen wrote the cause and a one-line rule, such as "Closing keywords only work on merges into the default branch" or "Restate negative questions before reading options". He reviewed the rules, not the questions, on each of the last four days, and grouped the confusable pairs into flash cards with the pair on the front and the separating sentence on the back. By the final session he could recite the separating sentence for every pair in the table without hesitation.