newbie questions: fee estimation, file size prediction, ISP blocking

[zyb]

Hi, I'm using bitcoin for a few days now. I read a lot in the forum but am certainly still missing a lot of points I guess. So, here come a couple of questions that are puzzling me, and which are probably totally clear to you:

1. I see that there is a reason for transaction confirmation being relatively slow, where the main limitation comes from the protocol which allows procession of a block every 10 minutes. As far as I have understood, paying a fee increases the probability for a transaction. I sent out three orders at the same time (same origin, same recipient, same amount), but one with zero fee, one with 0.1 BTC and one with 0.2 BTC fee. All three transactions got their first confirmation after ca 1h, and it took ca 2h to get 6 confirmations. Shouldn't those transactions with fees have been processed faster? What fee should be paid to "ensure" that the transaction will be in the next block? Is there a way to estimate procession probability as a function of the fee? Does the amount of coins transferred (in my case three times 0.01 BTC) matter?

2. Do I understand that correctly, that blk0001.dat contains the complete global history of transactions?

3. My blk0001.dat file is 122 MB. When the bitcoin network is growing transactions will probably grow exponentially. Will that cause the file size also to grow exponentially? If so, isn't that a showstopper?

4. Can't it just happen that an ISP blocks all bitcoin connections, let's say, if a government decides that bitcoin is illegal an forces ISPs to do so? Or are the bitcoin packets designed in a way so that they are not (easily) discoverable by an ISP?

Thanks for clarifications!

[theGECK]

I will answer to the best of my ability.

1) The transaction fees only matter when the room for free transactions is filled up. Transactions can be given priority in 2 ways, by transaction fee and by age. The older a transaction is, the higher priority it is given. This applies to both free and fee transactions. If you pay a transaction fee, your transaction is placed into the section of the block that can contain hundreds of transactions. If that is full, the size of the fee and the age of the transaction determine which gets into the block first. All three of your transactions were confirmed at the same time because there was room in the block for them, so priority didn't matter.

2) Yes

3) The file size shouldn't grow more than a few hundred MB a year at full capacity, and that is if you are downloading the full chain. There are options for making the size smaller, just using headers, etc that are not necessary to implement at this time.

4) Bitcoin uses IRC, so an ISP would have to block IRC to stop it.

[theymos]

4) Bitcoin uses IRC, so an ISP would have to block IRC to stop it.

Bitcoin doesn't use IRC for communication. It uses TCP over a fixed port, which makes it very easy to block. You can use Tor to bypass this, though.

[MoonShadow]

1. I see that there is a reason for transaction confirmation being relatively slow, where the main limitation comes from the protocol which allows procession of a block every 10 minutes.

Bitcoin's transactions confirmations are relatively fast, actually.  The banks and credit card companies maintain the illusion of instant transaction processing by simply making an advance guess on the likelyhood that your transaction isn't fraudulent, but then takes days to confirm.  Bitcoin could use a similar method to have instant transactions as well, but doesn't necessarily need such a method.

As far as I have understood, paying a fee increases the probability for a transaction. I sent out three orders at the same time (same origin, same recipient, same amount), but one with zero fee, one with 0.1 BTC and one with 0.2 BTC fee. All three transactions got their first confirmation after ca 1h, and it took ca 2h to get 6 confirmations. Shouldn't those transactions with fees have been processed faster?

This shows that, at present, there is no gain to adding voluntary transaction fees.  It is strange that your transactions took so long to process.  Was your client connected to the Bitcoin network at the time?

What fee should be paid to "ensure" that the transaction will be in the next block?

There is no way to garantee that any particular transaction will be in the next block.

2. Do I understand that correctly, that blk0001.dat contains the complete global history of transactions?

At present, this is an accurate understanding.

3. My blk0001.dat file is 122 MB. When the bitcoin network is growing transactions will probably grow exponentially. Will that cause the file size also to grow exponentially? If so, isn't that a showstopper?

The protocol includes a couple differnet methods of mitigating the expansion of the blockchain, none of which are presently implemented.

4. Can't it just happen that an ISP blocks all bitcoin connections, let's say, if a government decides that bitcoin is illegal an forces ISPs to do so? Or are the bitcoin packets designed in a way so that they are not (easily) discoverable by an ISP?

Traffic is quite open, and easily identified or blocked.  However, there are other methods of hiding the traffic, such as TOR, I2P or just a SSH tunnel.

[MoonShadow]

but I pay $2/GB for my internet usage so I'm hoping around 500MB or so.

Why is this?  Are you on a wireless ISP plan?  Or do you herald from a nation with poor access in general?

[theGECK]

4) Bitcoin uses IRC, so an ISP would have to block IRC to stop it.

Bitcoin doesn't use IRC for communication. It uses TCP over a fixed port, which makes it very easy to block. You can use Tor to bypass this, though.

It connects to an IRC server though when it is downloading the blockchain. Does it change the port from there?

[MoonShadow]

4) Bitcoin uses IRC, so an ISP would have to block IRC to stop it.

Bitcoin doesn't use IRC for communication. It uses TCP over a fixed port, which makes it very easy to block. You can use Tor to bypass this, though.

It connects to an IRC server though when it is downloading the blockchain. Does it change the port from there?

The client does not require access to IRC to bootstrap, it's just quicker.  The clients communicate directly with one another in a peer-to-peer fashion, and do not require IRC or any other client-server protocol to act as a network intermediary.

[Rena]

I wonder if an attacker could harm the network by taking over the IRC server it uses?

[MoonShadow]

I wonder if an attacker could harm the network by taking over the IRC server it uses?

Not for long.  IRC is not a critical aspect of the system.  IT's not even part of the protocol, and is only used as a peer discovery method that is not exclusive.

[MoonShadow]

but I pay $2/GB for my internet usage so I'm hoping around 500MB or so.

Why is this?  Are you on a wireless ISP plan?  Or do you herald from a nation with poor access in general?

A first-world nation with third-world internet access due to a telco monopoly.

Do you mind telling us the nation in question?  I might have a solution for you to pursue, depending on where in the world you are.  And are you an urban dweller?

[MoonShadow]

but I pay $2/GB for my internet usage so I'm hoping around 500MB or so.

Why is this?  Are you on a wireless ISP plan?  Or do you herald from a nation with poor access in general?

A first-world nation with third-world internet access due to a telco monopoly.

Do you mind telling us the nation in question?  I might have a solution for you to pursue, depending on where in the world you are.  And are you an urban dweller?

New Zealand, and yes.

NZ is good.  Urban is probably good.  You might want to consider setting up a mesh node filtered for only Bitcoin traffic, and annouce in the area that you are doing so.  If users join the mesh, bitcoin traffic to the outside world can be shared (if you're careful about it) and most of the traffic that your personal node would see would be coming across the mesh.  The trick is to force your local client to only permit one or two connections from the ISP and the rest from the mesh, and as long as others agree to do the same and don't freeload (not a given), your own personal chargeable traffic from Bitcoin should be managable. You should throttle the ISP Bitcoin traffic anyway, down to modem speeds; or set it up to only connect to ISP side nodes at particular times of the day.  In the future if you have a portable client on a smartphone, set it up so that it only connects to the mesh clients via wifi and never over the cell network.

Better yet, try to convince your ISP to set up a node of it's own that you could connect to without charges.  If traffic beyond your ISP is what is getting you charges, they might be willing to do this because such a local node can function well as a caching gateway, permitting the ISP to block outbound Bitcoin connections without actually interfering with the Bitcoin network.  As the Bitcoin network grows, I have no doubt that ISP's are going to want to control bandwidth consumption.

Of course, I have no idea how receptive they might be to the idea, but it's worth an email.