Field notes
The PROBE Log
Notes on how PROBE measures results, explains privacy signals, handles unclear responses, and keeps its 322 functional sources current.
Audit Your Own Username Footprint in 15 Minutes
A step-by-step self-check: list your handles, search them, review what you find, cut the details that link accounts, and retire what you no longer use.
Choosing a Username That Doesn’t Give You Away
Practical rules for choosing handles that do not reveal your name, age or location, keeping separate identities separate, and changing a handle without leaving gaps.
Honest by Default: Why PROBE Says Who It Is
Why PROBE identifies itself in every request, never poses as a browser or search crawler, and keeps some sources manual because their terms say no.
What We’re Building Next
A census of every manual and non-deciding source: what blocks each one, the four sources nearly ready, nightly canaries, and the infrastructure and measurement work ahead.
How to Search a Username Across the Web
A practical, step-by-step guide: choosing username variants, reading found, not found and unknown results, and verifying a match responsibly.
The Internet Isn’t Forever
We treat online footprints as permanent, but platforms vanish and historical profiles dissolve. What digital evaporation means for OSINT.
The Soft 404 Epidemic
When modern single-page applications return HTTP 200 OK for every path, naive scanners find phantom profiles everywhere.
Decay of the Open Web
Over half of historical username targets no longer exist. Why catalog size is a vanity metric that harms scan fidelity.
When a Profile Field Stays Empty
A blank bio is not a missing account. How to read unavailable profile details without turning them into identity claims.
Reading an Accuracy Snapshot
772 checks, 620 decisions, and only 129 calibration-eligible rows. A practical guide to the saved campaign's denominators and limits.
Profile Details, Carefully
Found results can now show a picture, name, and short bio. How each value is proven before it is shown, why coverage is partial, and what is never stored.
Checking the Checks
How re-testing our own answer key raised accuracy to 98.69% and recall to 95.87%, and which part of the gain came from a smaller catalog.
What Your Username Doesn't Hide
Why separate handles are only the first privacy layer, and how behavior, metadata, and context can still connect accounts.
Improving Accuracy Without Guessing
How a full-catalog evidence cycle raised coverage and recall while preserving zero observed false matches.
Measuring Username Search Accuracy Honestly
What we learned from testing PROBE against 947 known examples, and why one accuracy number never tells the whole story.
Why PROBE Says “Unknown”
A blocked page is not a missing account. Here is why PROBE sometimes says it cannot tell instead of making a guess.
Maintaining 322 Active Sources
How we keep PROBE's functional sources accurate as platforms change.