So to achieve a double spend one would need, for example, to send BTC back and forth to MtGox in the hope of achieving a longer than 6 block fork with the limited hashing power it has.
Thus exchanges and services could protect themselves from double spend by limiting the number of transaction on accounts.
Topic: Why Litecoin is more secure than Bitcoin (math inside) (Read 1016 times)
That post has some fundamental errors regarding how double-spending works (explained in a comment there).
Thanks for your explanation. I'll copy it here if you don't mind so people don't overlook it.
Quote
There are lots of errors in your post. I tried to explain in https://bitcoil.co.il/Doublespend.pdf how double-spending works.
In order to double-spend you don't need to find 6 blocks in a row. While the network is finding the 6 confirmations you're also mining so you'll be less than 6 blocks behind - and even then you don't need to find blocks in the row, you just need to catch up (e.g., if you're 3 blocks behind, you win if you find 2 blocks, the honest network finds another one, then you find 2 more). If you have 51% of the hashrate your chance to successfully double-spend is 100%.
On the other hand, in order to successfully double-spend you need to attempt to double spend. You actually have to commence an attack and there are costs involved. So referring to the chance to obtain a "streak" in a given time period is pointless - you have to relate the cost of an attack, its potential gain and its success chance.
Re terminology - a "51% attack" (or more accurately a ">50% attack") refers specifically to when you have majority hashrate and thus guaranteed success. If you have less than 50% it's not a >50% attack, it's just normal hashrate-based double-spending which can either succeed or fail.
Anyway, it's well known that for <50% hashrate, success chance decreases exponentially with confirmations and thus 24 is much more secure than 6. It is also well-known that Litecoin obtains 24 confirmations as fast as Bitcoin obtains 6.
In order to double-spend you don't need to find 6 blocks in a row. While the network is finding the 6 confirmations you're also mining so you'll be less than 6 blocks behind - and even then you don't need to find blocks in the row, you just need to catch up (e.g., if you're 3 blocks behind, you win if you find 2 blocks, the honest network finds another one, then you find 2 more). If you have 51% of the hashrate your chance to successfully double-spend is 100%.
On the other hand, in order to successfully double-spend you need to attempt to double spend. You actually have to commence an attack and there are costs involved. So referring to the chance to obtain a "streak" in a given time period is pointless - you have to relate the cost of an attack, its potential gain and its success chance.
Re terminology - a "51% attack" (or more accurately a ">50% attack") refers specifically to when you have majority hashrate and thus guaranteed success. If you have less than 50% it's not a >50% attack, it's just normal hashrate-based double-spending which can either succeed or fail.
Anyway, it's well known that for <50% hashrate, success chance decreases exponentially with confirmations and thus 24 is much more secure than 6. It is also well-known that Litecoin obtains 24 confirmations as fast as Bitcoin obtains 6.
That post has some fundamental errors regarding how double-spending works (explained in a comment there).
Really interesting post that was on Reddit:
See the rest of the post here: http://www.reddit.com/r/litecoin/comments/1cssqr/the_math_why_litecoin_is_more_secure_than_bitcoin/
Quote
Background:
This came out of some research I put into finding out the probability of a 51% attack for bitcoin. You'd think that mining 6 blocks in a row would be hard to do, even with 50% hashing power, because most people would probably think about a coin flip and how likely it is to flip 6 heads in a row. Intuitively it seems like a hard thing to do but in reality, if you do it enough times, it becomes not only possible, but downright likely, even with a mere 100 coin flips (warning:pdf).
With bitcoin mining 144 blocks per day a 51% attack with 50% hashing power is just as likely. Even with 35% hashing power it's relatively trivial to pull off. The reason most people probably haven't thought about this before is because the math itself isn't very straitforward. According to askamathematician.com, it is a tough question to answer with a simple formula. Thankfully there are computer programs which can do this. This site has a javascript "streak" calculator that will calculate the odds of a doing a 51% attack. Just put in 1008 trials for a 1 week period (6*24*7), your streak length of 6 blocks, and a probability of .35 for 35% hashing power. The result is about a 70% probability over a 1 week period and a 99.2% chance over a 1 month period.
Sense all the legwork had been done solving this problem for bitcoin I thought I'd run the math again for litecoin and what I found is actually quite surprising. Litecoin isn't just more secure than bitcoin, it is orders of magnitude more secure than bitcoin.
[...]
This came out of some research I put into finding out the probability of a 51% attack for bitcoin. You'd think that mining 6 blocks in a row would be hard to do, even with 50% hashing power, because most people would probably think about a coin flip and how likely it is to flip 6 heads in a row. Intuitively it seems like a hard thing to do but in reality, if you do it enough times, it becomes not only possible, but downright likely, even with a mere 100 coin flips (warning:pdf).
With bitcoin mining 144 blocks per day a 51% attack with 50% hashing power is just as likely. Even with 35% hashing power it's relatively trivial to pull off. The reason most people probably haven't thought about this before is because the math itself isn't very straitforward. According to askamathematician.com, it is a tough question to answer with a simple formula. Thankfully there are computer programs which can do this. This site has a javascript "streak" calculator that will calculate the odds of a doing a 51% attack. Just put in 1008 trials for a 1 week period (6*24*7), your streak length of 6 blocks, and a probability of .35 for 35% hashing power. The result is about a 70% probability over a 1 week period and a 99.2% chance over a 1 month period.
Sense all the legwork had been done solving this problem for bitcoin I thought I'd run the math again for litecoin and what I found is actually quite surprising. Litecoin isn't just more secure than bitcoin, it is orders of magnitude more secure than bitcoin.
[...]
See the rest of the post here: http://www.reddit.com/r/litecoin/comments/1cssqr/the_math_why_litecoin_is_more_secure_than_bitcoin/
Jump to: