Re: [4+ EH] Slush Pool (slushpool.com); Overt AsicBoost; World First Mining Pool

They might not want to generate empty blocks, they might not even be trying to. They just are.
-Dave

This could be the case but it's still a poor excuse not to improve. Their entire service is based on mining blocks and the transaction/rewards to be as profitable as possible for their miners. Transactions are the biggest reason for a miner to choose a PPLNS pool, so not doing everything possible to avoid empty blocks is just letting down your clients. On the flip side of that Miners exist to process these transactions, so if there is even 1 in the mempool it should damn well be in the block. I haven't been on slush for quite some time, but I think the problem has been they are big enough they don't care. There have been a lot of complaints about things like the merged mining rewards, and nothing has come of it. Unfortunately a lot of people just get complacent and don't reevaluate often enough, and if they are still make mad bank on the fees for the blocks mined without facing any real consequence where's there incentive to improve.

I didn't say it was an excuse for them not to improve. IF it's anything I listed it's a WORSE reason not to improve.
Nothing was particularly difficult to fix. Hardware is simple, to stressed add more. Configuration check around till you find the issue.

I started on slush a looong time ago. Left and never looked back. As you said between the merged mining and things like this it looks like they just don't care.

-Dave

If the most important part of the pool code, to generate the work to be sent to the miners, was 'unfathomable' code to the 'new' people running the pool for the past few years, then that is an even greater reason to run away from slush

It's central to how the pool code works, it must work correctly every single time, it must generate a set of data (work) based on the transactions returned by bitcoind, to be sent to the miners.
It is the same base work to be sent to all miners (as slush also said in his argument about why work difficulty was not part of the strartum work data sent to the miners) that then a minor addition is added to differentiate the work for each miner.

If instead they have extra code paths to generating the work, then indeed they are purposefully generating empty blocks, since there really is only one way to generate it properly - based on the transactions available in the mempool.

So yes indeed it certainly must be "they don't care"