The blockchain can grow at dynamic rate where each main or hyperlocal zone may add up to 10 megabytes to the TIM network during every block generation. Also there is the network state file for main network and each of the hyperlocal networks. The network state file is dynamic but of small size where a 1 kilo byte file can represent state of 10,000 zones.
The nodes in main network who want to add transactions for a zone would need to have all the zone files for previous block. The winning node adds current transactions, swarm defense signatures, network fork or merger points based on waiting transactions in current block for current zone, and awards for previous block. These details are verified by all other full-nodes or partial nodes for confirming the winner, and is an independently verifiable information sets.
The nodes in hyperlocal network who want to add transactions in hyperlocal dapp would need the main network state file and hyperlocal state file only. They don’t need hyperlocal zone files for previous block to add transactions, as any conflict will be referred to main network’s zone for resolution. Thus, the hyperlocal network would need to refresh themselves of 2 small files which would be less than a kilobyte, expectedly.
Repeating another historic assumptions like floppies, I expect that 1 kilobyte size for a network state file is large enough for all practical transactional needs and hyperlocal dapps extend that feature exponentially. Though time has proven all such assumptions are merely wrong in the vent of new innovations. Maybe in very long term, the main network would become conflict resolution media for hyperlocal dapps and hyperlocal dapps would carry all the functionalities. That way the main network will have highest performance. It requires a change of mindset from current notion of blockchains, and will take time for participants to realize the best usage of the network.