I'm wondering if/how Lightning Network relieves us of on-chain scaling on the road towards mass-level adoption. As I understand, payment broadcasts are deferred to the point in which the channel is closed, whereupon they are bundled and settled via consensus. Seems to me that although funds can be transferred hand to hand instantaneously, if we achieve global adoption, Lightning Network spends may still congest the mempool when they are finally broadcasted.
Discussion on page 52 of the Lightning Network whitepaper DRAFT Version 0.5.9.2 (after this was answered):
If we presume that a decentralized payment network exists and one user will make 3 blockchain transactions per year on average, Bitcoin will be able to support over 35 million users with 1MB blocks in ideal circumstances (assuming 2000 transactions/MB, or 500 bytes/Tx). This is quite limited, and an increase of the block size may be necessary to support everyone in the world using Bitcoin. A simple increase of the block size would be a hard fork, meaning all nodes will need to update their wallets if they wish to participate in the network with the larger blocks.
While it may appear as though this system will mitigate the block size increases in the short term, if it achieves global scale, it will necessitate a block size increase in the long term. Creating a credible tool to help prevent blockchain spam designed to encourage transactions to timeout becomes imperative.