Chapter 7 of 13 · 2 min read
An email address is not an argument
What the contact detour taught us about extraction, identity and evidence.
The Search for 100 Thinkers · Published 1.0 · 2026-09-22 · Moses Sam Paul
Contact research seemed like the practical part. Find the people, find the profiles, extract the links. Surely the spreadsheet could have this one.
It promptly introduced us to generic homepages, blocked pages, incomplete extraction and the possibility of matching the wrong person. The practical part had practical opinions.
First the person, then the page
The workflow had to become explicit: resolve identity, inspect an official profile, extract its actual links, follow candidate-specific destinations, and keep a source for every populated field. Search snippets and plausible URLs were not enough.
- Confirm the person and official profile
- Inspect linked pages and explicit contact fields
- Store exact sources or an explicit blocker
For the same reason, missing does not automatically mean “not found.” A page may be blocked, a destination uninspected, or an identity unresolved. Those states lead to different next actions.
The repair that made the list smaller
At the 11 September checkpoint, the addendum recorded 61 work-email fields, 90 verified profile states and 10 blocked profile states. These are stored audit states, not a guarantee of deliverability or an independent fresh check of every legacy record.
A repair then examined 33 disputed fields across 20 records: 14 retained, two corrected and 17 withdrawn. Previous values and reasons were preserved. A generic platform homepage could no longer masquerade as somebody's professional account.
JSON, CSV and workbook agreement passed. The contact provenance validator passed within its audited scope. Neither result means the entire contact task is finished.
Keep the address book in its lane
Reachability helps with an eventual invitation. It tells us nothing about whether a candidate can test a model. No email address was inferred, and contact availability must not decide portfolio membership.
The remaining contact work is an operational queue. It should not hold the intellectual investigation hostage. We came looking for criticism, not a particularly well-formatted mailing list.