Checkpoints will be replaced by compiled-in 'at THIS timestamp the main chain had THIS much proof of work.'

That is enough information to prevent attacks and still allow optimizations like skipping signature checking for ancient transactions.

I don't think anybody is proposing replacing checkpoints with nothing.

--
Gavin Andresen


On May 13, 2015, at 8:26 AM, Alex Mizrahi <alex.mizrahi@gmail.com> wrote:

Let's consider a concrete example:

1. User wants to accept Bitcoin payments, as his customers want this.
2. He downloads a recent version of Bitcoin Core, checks hashes and so on. (Maybe even builds from source.)
3. Let's it to sync for several hours or days.
4. After wallet is synced, he gives his address to customer.
5. Customer pays. 
6. User waits 10 confirmations and ships the goods. (Suppose it's something very expensive.)
7. Some time later, user wants to convert some of his bitcoins to dollars. He sends his bitcoins to an exchange but they never arrive.

He tries to investigate, and after some time discovers that his router (or his ISP's router) was hijacked. His Bitcoin node couldn't connect to any of the legitimate nodes, and thus got a complete fake chain from the attacker.
Bitcoins he received were totally fake.

Bitcoin Core did a shitty job and confirmed some fake transactions.
User doesn't care that if his network was not impaired, Bitcoin Core would have worked properly.
The main duty of Bitcoin Core is to check whether transactions are confirmed, and if it can be fooled by a simple router hack, then it does its job poorly.

If you don't see it being a problem, you should't be allowed to develop anything security-related.

If a node is connected to 99 dishonest nodes and 1 honest node, it can still sync with the main network.

Yes, it is good against Sybil attack, but not good against a network-level attack.
Attack on user's routers is a very realistic, plausible attack.
Imagine if SSL could be hacked by hacking a router, would people still use it?

Fucking no.
  
A 3 month reversal would be devastating, so the checkpoint isn't adding much extra security.

WIthout checkpoints an attacker could prepare a fork for $10.
With checkpoints, it would cost him at least $1000, but more likely upwards of $100000.
That's quite a difference, no?

I do not care what do you think about the reasons why checkpoints were added, but it is a fact that they make the attack scenario I describe above hard to impossible.

Without checkpoints, you could perform this attack using a laptop.
With checkpoints, you need access to significant amounts of mining ASICs.

------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
Bitcoin-development mailing list
Bitcoin-development@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bitcoin-development