[Bitcoin-development] Some PR preparation

[Gregory Maxwell]

On Tue, Mar 12, 2013 at 9:55 AM, Alan Reiner <etotheipi at gmail.com> wrote:
> I don't want to misrepresent what happened, but how much of that was really
> a risk?  The block was rejected, but the transactions were not.

Some but not much.  If someone flooded a bunch of duplicate
concurrently announcing both spends to as many nodes as they could
reach they would almost certainly gotten some conflicts into both
chains. Then both chains would have gotten >6 confirms. Then one chain
would pop and anyone on the popped side would see >6 confirm
transactions undo.

This attack would not require any particular resources, and only
enough technical sophistication to run something like pynode to give
raw txn to nodes at random.

The biggest barriers against it were people being uninterested in
attacking (as usual for all things) and there not being many (any?)
good targets who hadn't shut down their deposits.  They would have to
have accepted deposits with <12 confirms and let you withdraw. During
the event an attacker could have gotten  of their deposit-able funds.

On Tue, Mar 12, 2013 at 10:35 AM, Peter Vessenes <peter at coinlab.com> wrote:
> Can some enterprising soul determine if there were any double-spend attempts?
> I'm assuming no, and if that's the case, we should talk about that publicly.

There were circulating double-spends during the fork (as were visible
on blockchain.info). I don't know if any conflicts made it into the
losing chain, however. It's not too hard to check to see what inputs
were consumed in the losing fork and see if any have been consumed by
different transactions now.

I agree it would be good to confirm no one was ripped off, even though
we can't say there weren't any attempts.