you do realise that making "empty blocks" is not about making the average 10 minute block empty.
its about giving somthing to ASICS to work on during the minute or 2 that the pool server is validating a competitors previous block solution. and then when validated, the pool server sends the next cluster of unconfirmed transaction hash to the asics to fill the empty block with data.

its just by pure luck that some of the blocks are solved in a couple minutes and it has become visible.
which if blocks were solved in more then 2 minutes no one would know the difference..

the only issue is that sometimes the asics get lucky and have a solution in under 2 minutes before they get a chance to handle real data. which again is just luck. so chill out

atleast its better then making all the pools wait 2 minutes before doing anything, which would bring the average solution to more than 10 minutes and causing the difficulty to drop.

which if a pool server has to validate segwit+payment codes means pools are taking even longer to validate, causing again blocks to take longer until there is a viable solution. which again reduces the difficulty.

now imagine this:

if we had all pools waiting 4-8 minutes before actually hashing. with a lower difficulty. it only takes 1 pool with less hashpower to do an empty block to get a quick solution because they have a 4-8 minute headstart against the competitors.. which is another vector of the 51% attack theorum..

also add to the mix that a malicious pool sends out false blocks to waste competitors time making then validate a block for 2 minutes because they know by sending out false blocks delays the ethical competitors by 1-2 minutes per false block, ths a malicious pool can endlessly keep doing it for hours... sending  false blocks every 30 seconds so that competitors never even get to hash anything..  causing ethical miners to waste many many minutes validating and ultimately rejecting the false flag blocks, giving the malicious pool more of a head start.

its better that all pools follow the same rules where no one has an advantage by not leaving themselves open to abuse using delays... rather than hoping everyone delays their own efforts leaving a risk for abuse..on the pure hope and blind faith that no one malicious would try.

also remember.. hashing an empty block does NOT mean it will get solved in 1-2 minutes.. its just that sometimes they get lucky before they had a chance to fill blocks

id rather have 20 pools that occassionally got lucky then 20 pools that get delayed by 2minutes-2 hours due to a malicious attack vulnerability. and have only 1 pool solving blocks with malicious intent