Video card · Nate Curtiss · open in Notion ↗
1 College Dropout vs 10 Haters (White Room)
Cut from B cam, A_0007C205H260828_053602CL_CANON.MP4. It has the room audio, A cam does not. B cam runs 37 seconds ahead of A cam. Every timecode below is B cam.
No voiceover. Cut entirely from tape. Montage first, then the intro take lands it. Whole hook is 45 seconds max, montage plus intro.
Montage. 32 seconds hard cap. Nine clips, fast, no air between them. Trim each to the punch. He opens and closes, and every shot from the room gets answered on the next cut.
Attribution warning. The transcripts have no speaker labels and merge turns inside a single cue. Confirm every THEM clip against picture before locking.
Then the intro take. 00:02:33 – 00:02:46. "I'm Nate Curtis. I'm a college dropout. I quit school to pursue my dream of doing social media full time one year ago. And today we're back at my old college to debate 10 college students." Fourth and final take, the only one where he says Nate and looks down the lens. Do not use an earlier take. Cut on "students," before the "bang."
Straight into the rules.
Note: the title normally has to be confirmed in the first line. Here the montage runs first and the intro take does that job on the back end, which is deliberate.
The story: he goes back to defend leaving. The room's real argument is that it is not repeatable. He ends on the fact that one of them already did it and quit. That is the TikTok live girl, and she is why the video ends where it does.
Two biggest room moments go in the first eight minutes. Texture alternates instead of stacking four argument rounds at the front.
Seed her early. She does not speak until 00:53:00. Use cutaways of her from the first twenty minutes so she is not a new face at minute thirty-five.
Keep the losses. The hospitality kid, the college experience concession, and her shot at his numbers. The format dies if he answers everything.
Two checks before locking. People get flagged out and come back, so every reorder junction needs a look at who is in frame on both sides. And he announces prompts on camera, so most of those lines now name the wrong prompt and get replaced by the graphic.
Neither line is in either transcript. The ASR dropped both. This has to be done by ear.
What to listen for: a student answering back when he says "be around ambitious people," along the lines of "was being around Lily Vowell ambitious?" And separately a remark about sleeping with her friend.
Candidate window, 00:29:45 – 00:30:09. He is mid-sentence on "if you want to be around ambitious people, build skills, and" when something cuts across him. He loses the thread, says "what was the third one here?", then "I'm sorry," then restarts the identical sentence at 00:30:04. Both cameras show the derail, neither transcribed the cause. If it is there, cut from the interruption to 00:30:04 and pick up on the restart, which is word for word the same line so the join is invisible.
No candidate for the second. Sweep the gaps where the ASR gave up: 00:09:27 – 00:10:38, 00:33:17 – 00:33:39, 00:44:00 – 00:45:50, 00:50:15 – 00:50:40, 01:14:19 – 01:15:23.
No version ships with her name or a line that points at her or the friend. Ambiguous lines go to Nate with a timecode.
Pull these as selects before assembling. If run time has to come down, take it out of the connective material between rounds, never out of these.
Tier 1
Tier 2
Tier 3, the ones that make him human
Shorts, cut first: 00:38:48, 00:21:11, 00:05:37, 00:32:28, 00:17:52, 01:05:00
body last edited in Notion 2026-08-31 18:26 UTC