What does "taint analyis" on blockchain.info really mean?

[Amitabh S]

I have always wondered what they meant by that. Clicking on it gives some bunch of addresses and a "taint percentage". Is there any proper document explaining these terms?



I guess its been asked before. So please point me to the link.

[1715293456]

1715293456

[1715293456]

1715293456

[1715293456]

1715293456

[1715293456]

1715293456

[Amitabh S]

https://blockchain.info/taint/1JArS6jzE3AJ9sZ3aFij1BmTcpFGgN86hA

HTH

Thats not exceptionally clear. How is the percent calculated? Is it the number of times, or the amount of coins?

[dogie]

Here is the full quote:

The taint is the percentage of funds received by an address that can be traced back to another address. An example of a taint analysis is: https://blockchain.info/taint/1dice6GV5Rz2iaifPvX7RMjfhaNPC8SXH

Tldr its the proportional origin of the coins (expressed in %). So if A sent to a fresh address B, the taint analysis for B would be 100% A. If C then sent the same amount to B, the taint analysis for B would then show 50% A and 50% C. [Unless I have also misunderstood it? Does it go back multiple levels?]

[madonnino]

with this function you can see what is the address that exchanged bitcoin with your wallet , it also serves to test if mixer works well

[laurentmt]

Dogie is correct about the interpretation and unless things have changed, the metric goes back to multiple levels.

Unfortunately they don't seem to explain it very clearly themselves but my guess is that what really matters is the % of BTC in terms of UTXOs that can be clearly traced between the two parties (also it is not clear if that includes "levels" of indirection).

So if Alice was sending to Bob a total of 1 BTC but used a mixer to do this assuming that the mixer did its job correctly the % of Alice's coins that can be directly traced to Bob should be no bigger than of basically any other involved party (say 1%).

Taint is a broken concept for cryptocurrencies like bitcoin (and let's forget monero, zerocash, ...).
Computation of taint requires that you can determine with certainty the link between inputs & outputs of a transaction. Systems like mixers or coinjoin break this link. The best you can get is a probabilistic metric.

[gmaxwell]

The computation on BC.i is darn near a random number generator. The main thing it accomplishes is making people think they're more (or, infrequently, less) private than they are.

[OROBTC]

...

I hear and respect what gmaxwell says, but perhaps the best use of a current (existing) & decent mixing service is that it provides some privacy.  I am not worried about hiding my BTC trail from .gov (perhaps they can track BTC very easily or even own / can hack the mixers).  But, the mixers make it easier to hide the trail from casual or even Junior Varsity Snoops (mafias maybe too).

*   *   *

Is there anyone else who does "taint analysis"?  It would be interesting, to say the least, to see if both analyses would arrive at a similar percentage of taint between BTC addresses.