I completely understand the frustration and anger that is often engendered by statistics such as these - as well as by anecdotal reports of abuse. With that said, these are complex issues. Add to that the fact that incentives to do or not do something are very tricky animals. They rarely exist without unintended and/or unexpected consequences.
It strikes me as less than ideal for soft drinks to be the most popular category of purchases in this program and it seems like a good thing to try to avoid it if possible. Unfortunately, as others have pointed out, one has to be very careful how you try to accomplish that goal. Any program is susceptible to being gamed in some way and it is likely impossible to design a program that can't be gamed in any significant way; however, I believe even the worst instances of abuse are relatively small in comparison to the benefit that is delivered as intended. I think we would all be thrilled if government in general was able to operate within 10-20% of the ideal effectiveness and efficiency (whatever that might be).
If you click the USDA report link and download the appendices, there is a far more detailed list of subcategory rankings. It is shocking how similar they are between SNAP and non-SNAP households - with one glaring exception: Infant Formula Starter/Solution ranks #10 for SNAP households and #190 for non-SNAP households. I think that item alone goes a long way to show that people who need help are being helped as intended and they are not simply wasting the support they are receiving.
