Dynamic rather than static Block Size

[clearbit]

There is a lot of debate over the correct block size, and the rate of increase of the block size. Rather than hardcoding the size or the time period for doubling, why not let the network decide what the correct block size should be? If the network starts to get congested, as some people suggest, then the network would simply increase the block size based on a predetermined rule. Or is it impossible to come up with a rule?

[achow101]

There are many proposals which are dynamic block sizes. BIPs 100, 106, and 105 all are dynamic block sizes.

[clearbit]

There are many proposals which are dynamic block sizes. BIPs 100, 106, and 105 all are dynamic block sizes.

BIP 100 is a consensus voting algorithm for miners. It is not necessary to have a vote - the system can simply decide when the block size needs to be increased or decreased and do it automatically. No outside input would be required.

[achow101]

There are many proposals which are dynamic block sizes. BIPs 100, 106, and 105 all are dynamic block sizes.

BIP 100 is a consensus voting algorithm for miners. It is not necessary to have a vote - the system can simply decide when the block size needs to be increased or decreased and do it automatically. No outside input would be required.

Therein lies the problem. What is the correct block size? What does it need to be? That is where all of the debate is centered. What is considered right and how is that determined.

[mezzomix]

In my opinion a dynamic maximum block size is not necessary. Each miner can still decide about the real size of his block. All users have to decide about the maximum to keep a balanced system. Without a maximum size or with a high limit strong miners will be able to take over the mining and full node business by creating large blocks. All miners that cannot afford the network costs will be forced out. The full nodes make no money - the higher the cost the more of them are forced to switch of their node.

A dynamic maximum size is not necessary. Keep it simple, stupid (KISS).

[teukon]

If the network starts to get congested, as some people suggest, then the network would simply increase the block size based on a predetermined rule.

But then what would happen were demand for block space to far exceed what can be supplied by the network?

I maintain that a limit should be tied to what can be supplied, not what is being demanded.  A purely demand-driven block size limit would be as pointless as the US debt ceiling.

[DannyHamilton]

- snip -
Rather than hardcoding the size or the time period for doubling, why not let the network decide what the correct block size should be? If the network starts to get congested, as some people suggest, then the network would simply increase the block size based on a predetermined rule.

If you are not going to "hardcode" a predetermined rule into the network that it can use to "increase the blocksize", then how will the network be able to increase the blocksize?  There needs to be some sort of predetermined rule, the vast majority of node operators need to agree on the rule in order to reach consensus, and that rule needs to be coded into the nodes so that they can use it.

Or is it impossible to come up with a rule?

People have come up with lots of different rules.  Everyone has their own idea about what that rule should be.  The really difficult part is getting everyone (or at least the vast majority of node operators) to agree on what the rule should be so that the network can reach consensus about which blocks are valid.

[Smith480]

A purely demand-driven block size limit would be as pointless as the US debt ceiling. Thanks