PPS puts all risk on the pool operator, because miners will get paid for submitted shares regardless of whether or not the pool finds a block. That's why any PPS pool charges fees - to help mitigate some of that risk.
PPLNS transfers risk to miners. You are only paid for N shares, regardless of how long it took the pool to find a block.
When I first got into mining I did a lot of calculations on payment methods and my conclusion was that Eclipse's DGM at 0% fee was the best compromise. I'm not going to pull all the math out again because it isn't relevant. What is relevant is the trust you have for the operator, which killed Eclipse. With f2pool, antpool, and btcc PPS you have a maximum of two days loss if they skip a payment and then fail. That's relatively little risk, so they are easier to trust. A larger pool that can prove itself by making regular and frequent payouts is also easier to trust. Its harder to trust a pool with a lower hash rate simply because you are paid less frequently. A pool like kano.is can go multiple days without a block / payment and then if the pool were to fail you would have a loss of multiple days of hash. Consider the way HHTT shut down. That was a good pool and honest, but people still perceived that they lost something in the end.
The only reason anyone could have used logic to mine here was by weighing the risk to the reward. The first block finder reward was the reward that balanced out the risk for some, and there were also promises of extra payments for others in the top however many miners. Aside from that there was too much risk to mine here.
You can make arguments against the Chinese pools all day, fees, etc, but in the end people are going to trust them more than low hash rate pools simply because of the lower risk. It is more than variance that factors in. Nexious is a case in point.
BTW, even if the operator comes back with some kind of explanation at this point and pays the miners out, it would not be logical to continue mining at this pool. The risk of the pool's stability and dependence on the operator has been proven too great. The pool is dead.