Replace the names with codes, the same way every time ANON
Hashing a name does not anonymise it: there are only so many names, and anybody can hash all of them. A keyed hash cannot be reversed without the key, and is stable across every file.
The data
Read in this tab. Nothing is uploaded.
The key
Generated on this device and never stored with the data. Same key, same codes, across files and across years. A new key produces entirely different codes, which is how one release is separated from another.
Keep it somewhere the data is not. Without it you cannot re-link anything to anything. With it, anybody can re-identify everybody. Both of those are properties of the method rather than faults in it.
Which columns
What is left identifying
The result
How it works
ANON replaces names and other identifiers in a CSV with codes made by a keyed hash, the same code every time for the same value, so records can still be matched but cannot be reversed without the key. The result is pseudonymised, not anonymous. Lose the key and the codes cannot be turned back. Your data stays in this tab.
The guide: How to anonymise names in a spreadsheet, and keep the same person the same
Questions people ask about ANON
Why not just hash the names?
Because a hash of a name is not anonymous. There are only so many names, dates of birth and postcodes, so anybody can hash all of them and look yours up: the whole space is enumerable in seconds. A keyed hash cannot be attacked that way, because the attacker cannot compute the candidate codes without the key.
What happens if I lose the key?
You cannot re-link anything. Codes produced under a new key are entirely different, so a file released last year and one released this year could not be connected. That is a real consequence and it is the same property that makes the codes secure, so it cannot be designed away.
Is the output anonymous?
No, and this is the most important thing on the page. Pseudonymisation preserves exactly the structure that makes re-identification possible: if somebody establishes that one row is you, every row with your code is you. Treat the result as personal data and apply the same controls.
What is the uniqueness count for?
A combination of date of birth, postcode and sex that occurs once in the file identifies that row however well the name is coded. The count tells you how many rows are in that position, because coarsening those columns (a year instead of a date, an outward postcode instead of a full one) matters far more than the pseudonym does.
Why is the column name part of the code?
So the same value in two different columns gets two different codes. Otherwise a reference number that happened to equal a postcode would collide, and more importantly a code could be carried into a column it does not belong in.
Doing this often? OBSCURA Gnosis replaces names in transcripts with consistent pseudonyms, and codes them on this device. On its own, US$2 a month; with everything else, US$5 a month or US$48 a year.