Miners are leaving transactions out of blocks today because the blocks propagate faster (ie, have more odds of being confirmed) if they are <250K. Odds on losing the block fee exceed the transaction fees available, and the miner leaves the transaction unconfirmed.
Raising the size of the largest allowed blocks will not change that.
Do we have an implementation yet of broadcasting only the block header, and then letting the nodes assemble the blocks out of transactions they've already received over the network? That would reduce miners' disincentives for including transactions, so wouldn't that would be the more immediate means of increasing the number of transactions per block?
I agree that the block size needs to be increased; I'm just saying that increasing the allowed block size won't help if miners still have a financial incentive to limit the actual block size to 250K.
As the block subsidy is reduced this will become less of an issue as a higher percentage of total block reward will be from TX fees.
Fore each additional second it takes for a block to propagate there is only ~1/600 chance (maybe more if the difficulty is increasing) that the block will be orphaned because of extra propagation time. If the amount of additional TX fees makes it worth the miners to take this risk then they will include the additional TXs and take the risk of their found block being orphaned