Topic: do miners zip blocks when trying to propagate them? - page 2. (Read 1854 times)

ya but they don't need any bandwidth  if their full node is running on a US server.

Let me rephrase it then:

Bitcoin, as it is implemented right now by most if not all miners, does not currently propagate blocks in their full size, only the tx hashes so your idea is not in touch with how Bitcoin network currently works.

Except that it is simply not possible to do this (unless perhaps you own your own internet infrastructure that has fibre cables between China and the outside world and which is not subject to government control and I've certainly not heard of such a thing).

Setting up a VPN or whatever method you want to use to get around the Great Chinese Firewall will stop things like website access restrictions but you can't do anything about the bandwidth restrictions (i.e. you are slowed down to the 1990s).

So it is quite easy to get around blocks to YouTube but forget about trying to watch a HD video.

chinese miners should setup a node somewhere with a better internet connection that will handle the validation while they hash away in chain. how hard is it for a chinese solo miner to run some software in the cloud? does his full node need to be next to him for it to be considered under his control?

"SPV mining has already been happening in China "

Good! problem solved.

Sure, let's use empirical notions and ignore reality 

Firstly zipping is unlikely to reduce the size much at all (as was pointed out) due to most of it being random bytes.

Next I would like to point out that unfortunately, when you use the internet in China, speeds for transmitting (or receiving) data outside of China is actually often much like that in the early 1990's (with home-grade internet at least and even with business-grade it is very slow compared to data transmission between western countries) and the majority of hash-power is currently located in China.

This bandwidth issue is why so-called SPV mining has already been happening in China (which I understand was what had led to the most recent fork that occurred). So even if the block size can be reduced by only including tx hashes the problem will be that some (or maybe many) of the miners in China might still have a problem in keeping up with the TPS (which will likely result in blocks with few txs or perhaps only the coinbase tx).

There is no rule in Bitcoin that says you need to include any tx beyond the "coinbase" (or block reward) and we have seen at times new blocks appearing with no other txs (wasting the potential TPS dramatically). So if the network bandwidth is being pushed too far don't be surprised to start seeing blocks with only one tx in them which would actually mean supporting much larger blocks might simply fail to get txs even included at all.

He seems to understand it better than you, and is more humble with what he claims to know too.  For example, you earlier claimed that the

"Bitcoin does not currently propagate blocks in their full size, only the tx hashes so your idea is not in touch with how Bitcoin network currently works."

which is clearly wrong; the Bitcoin P2P protocol supports the propagation of complete (non-encoded) blocks between nodes/miners.  

Some miners use a centralized service called the Relay Network to improve their propagation rate and reduce their orphaning risk. However, this is not Bitcoin; the members of the Relay Network can be thought of as an Industrial Trade Group less centralized than a mining pool but more centralized than an independent miner operating on the P2P network.

Furthermore, the validity of your claim that most miners use the Relay Network depends on the definition of "most."  The empirical network propagation impedance is closer to what you would expect if no miners used the Relay Network than if all miners used the relay network.

node can do the same damn thing....

1GB is pushing the limits i agree

but its totally reasonable to expect a typical home computer to be able to stream 100MB of TX within 10mins.

at 1MBps you can stream 600MB of data every 10mins

miners/full nodes would simply be expected to be able to keep up with the TPS happening on the network.

and another optimization would be that miners don't include TX's that are likely not to have fully propagate through the network yet.

wtf is wrong with that?

 

Did you forget about the nodes? 1 GB blocks ? 

Seriously, this stuff is just beyond your understanding, just stop it. 

why?
bigger blocks will no longer mean centralization, if a 1GB block and can be sent using 4MBs!

Corrected.

This doesn't solve any of the centralization or attack vectors concerns btw

right this method is already being used by some miners, but this isn't standard way to propagate blocks.
should it be?
is this the key to scaling bitcoin?
I think yes on both counts.
right now this method is netting miners >200X coding gain!

 

What novel idea, maybe you should apply to become a core developer?

Here:

http://lists.linuxfoundation.org/pipermail/bitcoin-dev/2015-August/009942.html

if minner could communicate 100MB blocks with 250KB of encoded data, this is what will allow bitcoin to scale don't you think.

my guess is they wouldn't even need to send the full 64byte TX IDs only a 4 byte hash of the TX ID should be enoght for other minner to identify the TX's included in the new blocks.

using this method a miner could communicate a block with 250000 TX's!!! ( >400TPS ) with only 1MB of data

Definitely not.
Random data does not compress. Any compressibility is a sure sign of non-randomness.

Onkel Paul

so which is it?

Bitcoin does not currently propagate blocks in their full size, only the tx hashes so your idea is not in touch with how Bitcoin network currently works.

Speaking generally, random data reliably compresses to 50% of original size.

I think a more appropriate term for this would be encoding - using codes to represent larger blocks of information that is already known by all parties. This is usually way more effective than blind data compression.

What is wrong with people asking questions and trying to learn?