A New Brief That Silently Broke a Common Word
Before a medical-products deposition you import a set of case briefs. You define SPHAO* as 'spinal,' but your 6,200-entry dictionary already translates SPHAO* to 'spatial,' and you did not run a conflict check. Ten minutes into the job the witness says 'spatial disorientation' and 'spinal fusion' within the same answer, and the feed shows 'spinal disorientation.' You catch it live but cannot repair what has already displayed. The job runs another three hours, and 'spatial' will recur throughout the product-design section.
1. What do you do in the next sixty seconds, and how do you choose which word keeps the outline?
Read the worked answer
Pick one word for the outline and make the other obviously different, then test both in a scratch file inside a sentence - immediately, before the next answer arrives.
Which one keeps it? Weigh two things:
- Volume. 'Spinal' will dominate a spinal-implant deposition; 'spatial' is confined to the design section.
- Relearning risk. Whichever word loses the brief, you must remember not to use the old stroke for it, at speed, under pressure. If you keep 'spinal' on SPHAO*, then every time you mean 'spatial' you have to override a six-thousand-entry habit.
Given that 'spatial' already owned that outline before today, the lower-risk play is usually to revert SPHAO to 'spatial' and give 'spinal' a fresh, visibly different outline*, because you have three hours of spinal-heavy testimony ahead and a reverted habit is not something you have to actively suppress. What you must not do is leave the conflict live and hope to catch it.
2. The feed already shows 'spinal disorientation.' What do you do about the record, and what do you tell counsel?
Read the worked answer
Split the two questions.
The record is fixable and will be fixed: the audio backup resolves the passage, and the error is corrected during scoping against the audio, not from memory. Do not attempt to repair displayed text retroactively mid-job; that costs strokes and produces more errors.
Counsel gets told at the next break, briefly: 'One passage in the last ten minutes came out as spinal for spatial; here is the corrected line.' Disclosure is warranted because the error is substantive - a products case turning on spatial disorientation is a case where the wrong adjective misstates the testimony - and because counsel may already have read the feed and formed an impression.
Log it: time, the two words, the cause, the fix. If the same conflict produces a second error before the break, tell counsel immediately rather than waiting.
3. What process change prevents this, and why is the obvious check not sufficient on its own?
Read the worked answer
The check is a mandatory conflict screen on every imported brief, in both directions, before the import is saved. Obvious - and skip-prone under time pressure, which is exactly when imports happen.
The obvious check is not sufficient because a conflict screen only catches entries that collide on the same outline. It does not catch near-misses: a new brief one stroke away from an existing one, which translates correctly when tested and fails when written fast. That is why the second control is testing each imported brief in a full sentence in a scratch file, where spacing, boundaries, and near-misses show up.
The third control is a dated backup of the dictionary before any batch import, so a bad batch is rolled back in seconds instead of debugged live. Add all three and a conflict becomes a two-minute inconvenience rather than a corrupted live feed.