[Bitcoin-development] F2Pool has enabled full replace-by-fee

justusranvier at riseup.net justusranvier at riseup.net
Fri Jun 19 15:39:15 UTC 2015

Hash: SHA512

On 2015-06-19 15:11, Peter Todd wrote:
> If you ask me to pay you 1BTC at address A and I create tx1 that pays
> 1BTC to A1 and 2BTC of chain to C, what's wrong with me creating tx2
> that still pays 1BTC to A, but now only pays 1.999BTC to C? I'm not
> defrauding you, I'm just reducing the value of my change address to pay
> a higher fee. Similarly if I now need to pay Bob 0.5BTC, I can create
> tx3 paying 1BTC to A, 0.5BTC to B, and 1.498BTC to C.
> Yet from the point of view of an external observer they have no idea 
> why
> the transaction outputs reduced in size, nor any way of knowing if 
> fraud
> did or did not occur.

If there are two transactions which spend the same inputs, and each 
transaction has completely different output scripts, then this is prima 
facie fraudulent. https://en.wikipedia.org/wiki/Prima_facie

If the two transactions have identical output scripts, and one output is 
reduced in value to increase the transaction fee, that has the 
appearance of honest dealing. There is a possibility that the payer has 
chose to under-pay their payee in order to over-pay the miner, but 
that's not what a reasonable observer would assume at first glance.

Adding outputs to a transaction, while keeping all the existing outputs 
exactly how they are is another way of increasing the transaction fee of 
a transaction and is prima facie non-fraudulent.

Note that child-pays-for-parent has none of this ambiguity.

> What do you think of Bitcoin XT then? It relays double-spends, which
> makes it much easier to get double-spends to miners than before. In
> particular you see a lot of zero-fee transactions being replaced by
> fee-paying transactions, relayed through Bitcoin XT nodes and then
> mined. Is that encouraging fraud?

I haven't closely looked into the features of Bitcoin XT because I'm 
hoping that it never becomes relevant. I do want to see a heterogenous 
implementation network develop, but Bitcoin XT doesn't really count 
since it's a derivative of the Bitcoin Core codebase.

In general, I think every signed Bitcoin transaction sent between 
different parties is part of a valid, enforceable contract (using common 
law definitions which predate any particular legal jurisdiction). 
Handling contracts and money is Serious Business and so the decision of 
how software should respond to double spends should not be made 

Version: GnuPG v2


More information about the bitcoin-dev mailing list