It looks like you're new here. If you want to get involved, click one of these buttons!
By 2043, a micro-SD card (or equivalent device) will have a storage capacity of more than 500 billion gigabytes - equal to the entire contents of the Internet in 2009.
So in 2043 a MicroSD card could have a storage capacity of 500,000,000 TB (assuming we follow the same exponential increase in storage capacity that we've seen in the last 50 years).
The largest MicroSD card I've seen on the market today is 1TB.
500,000,000 / 1 = 0.000000002
So the average consumer drive today is theoretically 0.000000002% the size of what it will be in 2043.
If we had 1MB blocks still in 2043, it would be laughably small. If we had the same ratio of storage capacity to block size limit today, blocks would be 500x smaller than a single byte. So what is the ideal blocksize? Take a look at BIP 101 by Gavin Andresen. It was designed to automatically increase the block size limit to conservatively follow the same exponential increase that we've been seeing for the last 50 years in storage technology, starting at 8MB blocks today.
An 8MB block limit today would equal a maximum of 420GB per year added to the blockchain. Although this is an acceptable size, in reality it would be nowhere near this because the average blocksize is just passing 1MB now. It's good to have extra space to allow for burst activity, keep THE mempool clear (addressing Greg and his cop-out arguments, everyone's mempool increases when a limit is hit), and also to keep fees lower.
BIP 101 doubles the block size limit every 2 years. Following this increase in the block size limit and storage capacity, we would see the following results...
BSL = Block size limit. ADC = Average drive capacity.
As you can see the projection of BIP 101 vastly underestimates the average drive capacity of the future (it regulates for 16PB instead of 500,000PB) making it a conservative, yet progressive and safe solution.
But why do we want to scale on-chain instead of off-chain? Simply put, the blockchain works as intended and will scale as intended. This was the original intention of satoshi when he created bitcoin. There is absolutely no reason to move transactions outside of the bitcoin network. It will only stagnate growth and lead to further complications. Even if we wanted to scale off-chain, a 1MB limit would still be cripplingly slow. Lastly, the blockchain is evolving to handle much more than just bitcoin transactions. Projects like counterparty and rootstock intend to use the bitcoin blockchain for smart contracts and "dapps", and a larger block size limit is vital for their success.
The only other counter argument to large blocks is internet bandwidth and speeds. And for that there is Edholm's Law. BIP 101 accounts for this the same way it does for larger storage capacity.
Increasing the block size limit is not a "temporary" solution. It is the only inevitable scaling solution. If anything, segwit and the lightning network are impractical, temporary and a complete waste of time. They will eventually be seen as obsolete and be naturally replaced with a larger block size limit. This can also be safely done right now, so why bother with blockstream's "solutions"?
In the future, everyone will have a copy of the internet in their pocket. It will be called the blockchain.