I'm looking for a numerical privacy scoring system that estimate the degree of linkage between the members of a key pool (or even just individual key pairs). Ideally, the method should be easy to understand and apply.

I'm aware of "taint analysis" as offered by blockchain.info. There's also discussion in these papers:

What else has been done to quantitate Bitcoin user privacy? I'm most interested in ways to compute a benchmark score that tells a user how closely linked the the members of a key pool are, or how closely linked two individual key pairs are.


Here is a great video in which they asked: given 1 address I know belonging to someone, how many other addresses can I associate with that particular wallet. It turned out to be about 75%.

The method detailed in the video uses a few assumptions about how wallets work.

  1. Multi-input heuristic: All inputs in a transaction belong to the same wallet.
  2. Shadow change heuristic: Change public keys have never been seen before on the blockchain.
  3. Consumer change heuristic: Transactions from consumer wallets have 2 or less outputs.
  4. Optimal change heuristic: Wallets do not spend unnecessary outputs. This means that a transaction with say 2 inputs of 0.5 each and 2 outputs of 0.6 and 0.4, the 0.4 is assumed to be the change address belonging to the same wallet. This is because if the intended payment is 0.4, the wallet would have chosen a single 0.5 input.
| improve this answer | |
  • The question is asking about a method to quantify how closely related two addresses are. Your answer is related to the topic, but doesn't address the question. If it is answered in the video, please include the relevant points in your answer. Otherwise, I think this post is to be classified as "not an answer" on this question. – Murch Apr 29 '16 at 8:23
  • Sure, I'll edit the response to answer the question. – mrkent May 9 '16 at 18:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.