[Bitcoin-development] Coinbase reallocation to discourage Finney attacks

[Andy Parkins]

On Wednesday 23 Apr 2014 12:45:34 Mike Hearn wrote:

> OK, sure, let's say most Bitcoin users will be honest (we hope). But
> unfortunately in a situation where fraud is possible users wouldn't
> necessarily distribute evenly over transactions.

That's true, but even in the worst that that 5% hashing power attack means 
that 95% of the time, your attack fails.  That means you end up paying for 
what you bought.  Also, you're again changing the comparison basis -- your 
CC figures were for the entire industry, not the most badly affected 
merchant.  You can't say "one particular bitcoin merchant suffers 5% fraud, 
therefore that's worse than the 2% fraud averaged across all CC merchants".

> If a merchant is selling something of value repeatedly, then a small
> number of scammers can go back and try their luck over and over. I'm not
> sure how many trades fall into such an exploitable category, though.
>
> Also, there's the philosophical question of how honest people really are
> when there's no consequences to their actions. For instance, if most

There _are_ consequences though: 95% of the time, you end up buying 
something and paying for it.

Viewed another way, if I buy something repeatedly from an at risk merchant 
(and there won't be many; as you pointed out, mail order is completely 
unaffected as you can simply wait for your confirmations) that costs, say 
0.01 BTC per item, then I have to buy 100 of them to get 5 of them for free.  
Do I really want 100 of them?  Even if I do want them, then I've had to 
supply capital of 1 BTC to earn 0.05 BTC in kind.

If what I'm buying is another form of money (as with exchanges, or perhaps 
casinos) when that "in kind" is just as liquid as the BTC, then fair enough, 
there is a risk, but that just incentivises the merchant in those cases to 
not allow withdrawal/deposit until 6 confirmations have been received.  
Those merchants then move from "at risk" to "not at risk".

I'm still struggling to see how bitcoin could ever be as bad as CC fraud.



Andy

-- 
Dr Andy Parkins
andyparkins at gmail.com