Or in miner terms
Every 30s you get new work to add to what you are doing - that you switch to for all new hashing you do.
Every network block change you get new work that invalidates all work you are currently doing, that you switch to immediately.
Each new work item is unique per miner instance - which of course it must be otherwise people would be duplicating hashes.
i.e. everyone on the internet is effectively always trying to solve a different block header - they just happen to have the same previous block hash and the same block number.
The ONLY competition comes where 2 miners solve the current block at around the same time.
The question then is who's network distribution of their block solution is the best to give the best chance of their block being confirmed by the NEXT block.
Or in 'pool' terms: does the pool you are mining to handle block changes fast or does it suck in terms of block change handling ...
and this is the real big problem about the big mb block. If we have an increase of mb in each block the pool that will solved will go ahead from others to solve the next block because of the network lag to transmit the previous blog