Hi,

On 28/12/13 12:52, jaume wrote:
>> Hey, I'm not sure that always reusing addresses has such a drastic
>> effect on privacy like we always assume. I'd like to see more evidence
>> for this assumption.
>> https://wiki.unsystem.net/index.php/DarkWallet/Paranoia#Address_reuse
>> The vast majority of addresses in the blockchain are uniquely used.
>> We can see this by comparing both those graphs.
>>
>> I'm also interested in other data and stats we can collect to improve
>> privacy and freedom of users to transact. Also usage stats and things
>> that can improve software utility for the usecases we're interested in.
>
> Thanks for sharing, Amir :) I'm very interested in all these stats, but I
> didn't know how to generate them myself. The day before yesterday I
> downloaded a postgresql dump that you shared back in 2011.
> https://bitcointalk.org/index.php?topic=38246.0
>
> Could you ellaborate a bit on what is shown on the plots and why it
> evidences that addresses are uniquely used? Could you share the raw data
> and/or explain how can obtain it myself?

I wanted to see some info about the number of inputs compared to the
number of outputs in transactions (inputs vs outputs) to give me some
idea about the proportion of money flows. That's the first graph where
we see a lot of clustering in the bottom left corner. There's a lot more
transactions with very many outputs coming from a single source, than
transactions with many inputs going into few outputs.

The next graph is grouping the inputs/outputs by unique address (so
counting the number of unique addresses in inputs/outputs). Apart from
some transactions with many inputs, it remains relatively unchanged.

You can generate it from here:
https://github.com/genjix/txfillstat/tree/master/addr
The source files generate the data needed for the gnuplot .plot files.

If it would be useful, I can generate another .sql file but they take
ages to create (especially now the blockchain is much larger). Or I can
document the database formats and provide some Python files to show how
to load and parse them.

>
> I read the paper "A Fistful of Bitcoins: Characterizing Payments Among Men
> with No Names" and they mentioned that they use Andresen's bitcointools
> (What is the unsystem alternative?). The authors propose techniques to
> link addresses that belong to the same user. If I understood everything
> correctly, those techniques will no longer be useful as soon as some
> people start using CoinJoin and DarkWallet.
>
> It would be extremely interesting to gather stats about the use of
> CoinJoin in the blockchain. I expect an increase in the number of CoinJoin
> transactions when DarkWallet is available.

https://github.com/darkwallet/python-obelisk
see examples/
example wallet:
https://github.com/darkwallet/dat.wallet

>
> Regarding address re-use, my understanding was that disposable addresses
> that are used only once are a necessary but not sufficient condition for
> anonymity and privacy. Am I right?

Yeah, I'm questioning this conventional wisdom. If we want to have a
rigorous case to base dev on, then we can't be basing our beliefs on
superstition or intuition. I'm interested to help provide the data or
tools so people can start this discussion, research and evaluation. But
I need to know what's needed first.

>
> Sorry for the avalanche of questions. Keep up the great work in writing
> awesome free software and educating the community.
>
> Jaume
>
>
> _______________________________________________
> unSYSTEM mailing list: http://unsystem.net
> https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/unsystem
>