Put bluntly, this paper is wrong and biased:

• MPPS does pay out fairly over the long-term to fair miners, at least in practice.
• While it is true that hoppers can hurt fair miners with MPPS, they have no incentive to do so. (remember that any pool can be hurt by the block withholding attack, so this is not a significant flaw)
• Terminology: SMPPS does not accrue "debt", but issues fiat "extra credits" when it cannot pay Bitcoins.
• SMPPS, in theory, will always drift toward 0-buffer 0-credit, it does not have any inherent negative drift.
• In practice, SMPPS has proven to have a positive buffer much more than zero (on Eligius, we have only very briefly hit 0-buffer a few times)
• While in theory, people might "hop" off SMPPS when it has no buffer, this is in practice not a problem. There was no mass exodus from Ars when its buffer hit zero and began issuing extra credit, and by now everyone has observed that so long as the pool remains online, it will eventually recover. Furthermore, this kind of "hopping" does not benefit the hopper nor harm the non-hopper.

Note that I only read the "Attempts for risk-free pay-per-share" section, and am not attempting to cover the other methods here.