Most importantly, the vast majority of findings from this paper were first researched in late 2014 in MRL-0001. Since this research paper, improvements have been made, including increasing the minimum ringsize to 3, 5, and now 7. RingCT was added in January 2017, which hides the amount.
The paper has 3 recommendations:
1. Warn users who made early transactions that they may have had their ring signatures compromised. The stealth addresses would still work. This isn't a suggested improvement, and I believe that any user interested in maintaining their privacy should have considered researching Monero, including reading MRL-0001. I'm sure not everyone did this, but keep in mind that Monero was a much smaller network back then made of mostly enthusiasts. If you were transacting on AlphaBay, the least you could have done is seem what the limitations of your tool are. They were much higher at the time than they are now.
2. Improve the decoy selection algorithm. There is definitely room for improvement to make the most out of the decoys selected for the ring signature.
3. Consider avoiding public pool payout outputs, which are known to be used in the pool transactions. This is a fair point, though from the proportion of pool payout transactions to total transactions as found in the paper, this case is likely still adequately covered with ringsize 7. Monero can still consider avoiding these outputs as decoys though in some way.