ASICMINER: Entering the Future of ASIC Mining by Inventing It

[tinus42]

I think it will be solved very soon, for example, offshore miner farm in HK with better internet connection environment may be considered and built.

Shenzen (where they are located) is only 10 miles away from Hong Kong, so that would make sense.

Quality of network latency and reliability between China and outside of China is not necessarily related to geographic distance. HK uses different ISPs and is outside the great firewall. Moving to HK for example, would be much better in terms of networking speed but I'm not sure that the regulatory concerns would be completely solved by that.

Wasn't thinking about network latency. Just simple logistics, build blades in Shenzen and have them up and running a few hours later in Hong Kong without much delay. Don't know anything about the internet in China and Hong Kong nor about the regulatory environments there.

[binaryFate]

It's true that latency can have a small effect on block mining rewards (higher probability of orphaned), but should not impact the number of transactions that are known to the mining node (except for those that arrive in the last moment before a block is found), so it cannot explain these low numbers.

[runeks]

It's true that latency can have a small effect on block mining rewards (higher probability of orphaned), but should not impact the number of transactions that are known to the mining node (except for those that arrive in the last moment before a block is found), so it cannot explain these low numbers.

I agree. We're seeing a 40% difference in the number of transactions included per block. Unless the Great Firewall is picking out transactions, I don't see how this is possible. One explanation is that ASICMiner has a custom filter to decide which transactions to include.

[binaryFate]

It's true that latency can have a small effect on block mining rewards (higher probability of orphaned), but should not impact the number of transactions that are known to the mining node (except for those that arrive in the last moment before a block is found), so it cannot explain these low numbers.

I agree. We're seeing a 40% difference in the number of transactions included per block. Unless the Great Firewall is picking out transactions, I don't see how this is possible. One explanation is that ASICMiner has a custom filter to decide which transactions to include.

The custom filter is a good explanation. It's the only one I can think of actually. Great firewall appart...
Anybody has feelings, or even better, statistics on the type of transactions included in their blocks, in particular on their category (free or not), and the fee per KB (for the non-free), and their priority?
I think I saw something along that lines recently, don't remember if it was here or the speculation thread, or about AM at all.

EDIT: That would make a good official question for friedcat! I'm really curious about these filter policies and their motivations, if they do exist.

[chkgk]

EDIT: That would make a good official question for friedcat! I'm really curious about these filter policies and their motivations, if they do exist.

+1

I would very much like to see that question included!

[canth]

The custom filter is a good explanation. It's the only one I can think of actually. Great firewall appart...

I hope this is not the case.  Filtering transactions is not good for bitcoin.  It is also not good for ASICminer shareholders.

I personally have had several transactions that were not initially confirmed when ASICminer found a block.  I had to wait for the next block to be found by someone other than ASICminer to get an initial confirmation.  I included a .0005 fee on each of those transactions, which I thought was enough to get my transaction included.

I doubt it's this - I would say it's more likely that connections from within China are able to get to fewer connections outside of China. It's not just the government firewall, peering is just not straightforward when it comes to this country.

[binaryFate]

The custom filter is a good explanation. It's the only one I can think of actually. Great firewall appart...

I hope this is not the case.  Filtering transactions is not good for bitcoin.  It is also not good for ASICminer shareholders.

I personally have had several transactions that were not initially confirmed when ASICminer found a block.  I had to wait for the next block to be found by someone other than ASICminer to get an initial confirmation.  I included a .0005 fee on each of those transactions, which I thought was enough to get my transaction included.

I doubt it's this - I would say it's more likely that connections from within China are able to get to fewer connections outside of China. It's not just the government firewall, peering is just not straightforward when it comes to this country.

Poor peering introduces latency, and you might get transactions 10s after the average rest of the world, but you'll get them anyway. Unless you're totally cut on a sub-network, poor peering should be still good enough.

[canth]

Poor peering introduces latency, and you might get transactions 10s after the average rest of the world, but you'll get them anyway. Unless you're totally cut on a sub-network, poor peering should be still good enough.

OK. There's no logical reason to intentionally filter transactions, except those that don't include fees, right?

[AMuppInTime]

Poor peering introduces latency, and you might get transactions 10s after the average rest of the world, but you'll get them anyway. Unless you're totally cut on a sub-network, poor peering should be still good enough.

OK. There's no logical reason to intentionally filter transactions, except those that don't include fees, right?

Might it be they are trying to limit incoming traffic to not attract attention?

[binaryFate]

I doubt it's this - I would say it's more likely that connections from within China are able to get to fewer connections outside of China. It's not just the government firewall, peering is just not straightforward when it comes to this country.

In that case, then the mining operations of ASICminer need to be moved outside China.  Long term, transaction fees are going to be a susbtanial portion of ASICminer's mining income.  In order to ensure a steady stream of mining income, they need to make sure that they are getting all of the transactions fees that are available.

As it is now, if you look at blocks found immediately after ASICminer finds a block, you will see a larger number of transactions and thus a larger amount of transaction fees in that block.  ASICminer is giving away those transaction fees to the next miner to find a block.  It may not seem like ASICminer is giving up much now, but if the value of bitcoin increases substantially, then these missed transaction fees are going to add up.

That's also bad for the AM reputation. Us, we mostly care about missing fees for shareholders, but I can imagine many of the unrational anti-AM people out there would be happy to point this out as a subsequent problem of AM hashrate. Usually I don't care about their screaming, but on this I'd feel uncomfortable.

[twmz]

The custom filter is a good explanation. It's the only one I can think of actually. Great firewall appart...

Friedcat already said that they don't do anything special filtering of transactions...

We are using the same transaction policy of the official Bitcoin client's default behavior.

[binaryFate]

The custom filter is a good explanation. It's the only one I can think of actually. Great firewall appart...

Friedcat already said that they don't do anything special filtering transactions...

We are using the same transaction policy of the official Bitcoin client's default behavior.

Friedcat would agree that 40% of difference on average cannot be considered just statistical noise. There must be a technical explanation, either explicit policies or implicitly as consequences of some technical facts.

[canth]

If I understand correctly,

Can I buy (let's say) 20 shares of AM-PT on BTCT, and then transfer all of them immediately to be direct shares by asking burnside? (And then again once per month?)

Yes, but give it 3-4 days for the transfer to take place. Friedcat only does transfers 2x a week, so have a bit of patience.

[VolanicEruptor]

So let's say someone decided to mine a huge portion of the network, such as Friedcat does.  Then they get paid by someone who wants to destroy Bitcoin to deliberately slow their latency down so that they are able to solve blocks with a very minimal amount of transactions. 

You get a few people doing this, and it would destroy Bitcoin.

Much more creative than a 51% attack..

[twentyseventy]

So let's say someone decided to mine a huge portion of the network, such as Friedcat does.  Then they get paid by someone who wants to destroy Bitcoin to deliberately slow their latency down so that they are able to solve blocks with a very minimal amount of transactions. 

You get a few people doing this, and it would destroy Bitcoin.

Much more creative than a 51% attack..

I think a protocol change would be in order then, no?

[canth]

So let's say someone decided to mine a huge portion of the network, such as Friedcat does.  Then they get paid by someone who wants to destroy Bitcoin to deliberately slow their latency down so that they are able to solve blocks with a very minimal amount of transactions. 

You get a few people doing this, and it would destroy Bitcoin.

Much more creative than a 51% attack..

Who benefits from this sort of attack? If they have a reputation and fortune to potentially lose, would they risk a potential lawsuit by bitcoin owners, miners, businesses etc? The only entities that I could see doing something like this would be a government entity (potentially the BRIC countries) that decides bitcoin is too much competition for traditional banking and government controlled fiat.

[notme]

So let's say someone decided to mine a huge portion of the network, such as Friedcat does.  Then they get paid by someone who wants to destroy Bitcoin to deliberately slow their latency down so that they are able to solve blocks with a very minimal amount of transactions. 

You get a few people doing this, and it would destroy Bitcoin.

Much more creative than a 51% attack..

It would not destroy bitcoin.  It would drive up transaction fees until either not including them is too expensive or more honest miners come online.

[VolanicEruptor]

So let's say someone decided to mine a huge portion of the network, such as Friedcat does.  Then they get paid by someone who wants to destroy Bitcoin to deliberately slow their latency down so that they are able to solve blocks with a very minimal amount of transactions. 

You get a few people doing this, and it would destroy Bitcoin.

Much more creative than a 51% attack..

It would not destroy bitcoin.  It would drive up transaction fees until either not including them is too expensive or more honest miners come online.

driving up transaction fees could destroy Bitcoin as an alternate currency, seeing as how low transaction fees is supposed to be one of it's strongest points.

[notme]

So let's say someone decided to mine a huge portion of the network, such as Friedcat does.  Then they get paid by someone who wants to destroy Bitcoin to deliberately slow their latency down so that they are able to solve blocks with a very minimal amount of transactions.  

You get a few people doing this, and it would destroy Bitcoin.

Much more creative than a 51% attack..

It would not destroy bitcoin.  It would drive up transaction fees until either not including them is too expensive or more honest miners come online.

driving up transaction fees could destroy Bitcoin as an alternate currency, seeing as how low transaction fees is supposed to be one of it's strongest points.

I don't see low transaction fees as among it's strongest features.  More important are decentralized control, potential privacy, and censorship resistance.  In fact these features provide the business case for paying more for transactions than with fiat based systems.  Lower fees are just a nicety of the current economics.

[VolanicEruptor]

So let's say someone decided to mine a huge portion of the network, such as Friedcat does.  Then they get paid by someone who wants to destroy Bitcoin to deliberately slow their latency down so that they are able to solve blocks with a very minimal amount of transactions. 

You get a few people doing this, and it would destroy Bitcoin.

Much more creative than a 51% attack..

Who benefits from this sort of attack? If they have a reputation and fortune to potentially lose, would they risk a potential lawsuit by bitcoin owners, miners, businesses etc? The only entities that I could see doing something like this would be a government entity (potentially the BRIC countries) that decides bitcoin is too much competition for traditional banking and government controlled fiat.

If I was a rich asshole heavily invested in the banking industry, I could benefit..
Hell, maybe Satoshi bullied me in school and stole my lunch money.  I'm sure there's many reasons.  Terrorism even comes to mind.. what if I'm a kingpin meth distributor and Silkroad is interfering with my profits?  
I could go and on, so don't ever feel bulletproof based on the reason that the intent isn't there, because it very well could be..