Pressure ladder and human mocks: rehearse up to being watched, and check your own score against the interviewer's

Rehearse up a ladder until you are watched, and test your feeling of a round against the interviewer's score.

Before you read: predict

A backend engineer at a chat company has her first human mock, a 45-minute design round, in ten days. For three weeks she has practised group-chat designs alone, silently, on paper, and finishes each one with time to spare.

Which one change to her sessions rehearses what the mock will add?

The 30-second version

Being watched while you solve is the hard part, so silent practice skips it. Climb one condition at a time: private, timed, spoken, observed with follow-ups, then a full mock with a human. Your feeling of how it went is weak evidence: write your score first, then compare.

Rehearse up a ladder until you are watched, and test your feeling of a round against the interviewer's score.

How it works

Step 1 of 13

Watched is the hard part. Being watched while you work, not talk alone, is what stresses; pair-narration-think-aloud step 6 owns its cost. [S1]

Analogy: A comic working up a new set

A stand-up comic writes her new five minutes at the kitchen table. Then she runs it against a timer, then says it aloud to her phone, then tries it at an open-mic night, where the room stares back and a heckler cuts in. Only then does she play the showcase where a booker decides. Before each set she writes down which three jokes will land and seals the note; afterwards she counts the laughs on the recording. A comic who only ever rehearsed at the table forgets her opening line the first time a room watches. (The rungs and the showcase are the scene's own.)

In the comedy club	In the interview
Kitchen table, timer, phone	Private, timed, spoken
Open mic with a heckler	Observed, with follow-ups
The booker's showcase	A full mock with a human
The sealed note	Your score, written before feedback
Laughs counted on the recording	The interviewer's score

Breaks down when: the comic hears the room laugh or go quiet as she works, while an interviewer often gives no sign. Wrong inference: “I'll know how it went”: your feeling is weaker evidence still.

In real terms: add one condition at a time until you are watched, and trust the logged gap over the feeling.

Try it

Answer first, then check. Nothing is saved.

Question 1 of 7

A chat engineer logs three paid human mocks, each with her own 1–4 score and the interviewer's. Mock 1 (design, a group-chat service): hers 2, the interviewer's 3. Mock 2 (coding, a message-search box): hers 2, the interviewer's 4. Mock 3 (design, read receipts): hers 3, the interviewer's 3.

What do the three mocks say about her own scores?

Question 2 of 7

A chat engineer books a 45-minute human mock. The interviewer runs the design for 40 minutes, then spends the last 5 telling her what went well and what didn't; his 1–4 score follows by email tonight. She wants to learn whether her own sense of a round can be trusted.

Going by interviewing.io (read 2026-10-09), when does she write down her own 1–4 score?

Question 3 of 7

Observed mock: the interviewer asks a chat engineer to build an unread-message counter. She finishes it, and he extends it three times: per conversation, then across her devices, then with muted threads. She stalls on the last one and rates herself 1 of 4.

Going by interviewing.io (read 2026-10-09), how should she read the stall?

Question 4 of 7

A chat engineer rehearses a message-retry queue on two rungs. In three private sessions she traced an input through her code by hand before saying done, and caught a bug each time. In three observed sessions, coding aloud while a friend watched, she said done with no trace, and two of the three had a bug. Going by the trial of Behroozi et al.

(read 2026-10-09), what should her next rehearsals change?

Question 5 of 7

The interviewer's last words are “Tell me what you would change before the real thing.” Before she answers, she logs the mock.

Write the log entry as five lines, each starting with its label: Rung:, Sealed:, Interviewer:, Gap: and Before the real thing:.

Your answer (optional, stays on this page)

Question 6 of 7

On interviewing.io's 1–4 scale (as of 2026-10-09; its data as of August 2016), a score of about 3 or more is roughly the hire line. A chat engineer's log of four human mocks, her sealed score first and the interviewer's second: mock 1, 2 and 3; mock 2, 3 and 3; mock 3, 2 and 2; mock 4, 3 and 4.

In which mocks did her sealed score land on the same side of the hire line as the interviewer's?

Question 7 of 7

A chat engineer's real round is a 45-minute design round. She plans two rehearsals for it. In rehearsal 1 she designs a presence service silently on paper, untimed. In rehearsal 2 she designs a read-receipt service aloud into her phone's recorder, on a 45-minute clock.

Which conditions of the real round does rehearsal 2 have that rehearsal 1 lacks?

Go deeper

Pro tip

Record the first five minutes of every observed rehearsal and play them back. In the trial, several watched participants found the opening the most stressful part.

Checked against the tools and docs as of 2026-10-09.

Predict first