My understanding is that the fraudster would either need to coordinate a Sybil attack, or he would need to pass the fraudulent double-spend over a non-public back-channel to an unethical miner that controls a great deal of global hash power.
The later is more likely. Grocery business is a very low margin business. It also has almost no credit card fraud (relatively speaking) and enjoys some of the lowest interchange fees among all business categories. The "real" cost (fraud, compliance, fees) for a grocery store to get paid by credit cards is <1% and may rival cash so these merchants have the least incentive to switch.
The "risk profile" for a business doesn't just depend on the price of goods or services but also its margin. Online casinos for example have very high fraud rates but they also end up taking 50% to 70% of player deposits or more. The high margin allows them to operate at a level of fraud which would bankrupt most other businesses. It doesn't take much fraud for a grocery store to lose money.
I imagine that in a "bitcoin future" there will be less zero-confirm fraud using bitcoin [malleability aside] than purchases with counterfeit cash or stolen credit cards. Do you disagree?
In general? Yes however every business is different.
Online digital goods has credit card fraud approaching 10%, when you add in the merchant fees, fraud prevention costs, and chargeback costs it can reach 15% or more of gross revenue. That is the low hanging fruit. Even if 5% of 0-confirm tx were fraudulent the companies would pocket billions in additional revenue. On the other hand some businesses (like grocery store) have very low margins, and have already negotiated very low interchange fees. That combined with low existing fraud rate means it is possible 0-confirm actually increases fraud rates relative to credit cards.
On a long enough timeline I think "fraudulent" miners are all but an inevitability. I could see an entity with say 20% of the network willing to accept out of band double spends for a hefty fee (either flat rate say $10 per tx, or a % of the tx amount). Would that impact retail stores? Lots of crimes are crimes of opportunity. In bad economic times and when gas prices rose the number of drive offs increased. If there is an app you download that gives you a 20% of having a free purchase with no risk and money is tight, I think the number of people that would rationalize their theft away is non-zero.
I don't want you think I am trying to predict the future. The risk may end up being overstated and maybe grocery stores just end up accepting zero confirm transactions straight on the blockchain with no other assurances. I honestly don't know. I am just pointing out that might not be the case. Among all possible businesses grocery stores probably have the least to benefit from Bitcoin and conversion on a large scale would require near zero fraud rates. I am not sure if zero confirmation (without a trusted third party) can guarantee that. Given the ease that the double spend risk can be mitigated with identification or backup payment I think that is certainly possible. Case in point many places ask for ID if you want to pay by check. A 0-confirm tx has a risk profile similar to a payment by check. Once confirmed it is like a cleared check. Of course waiting for confirmation at point of sale is no more possible then waiting for a check to clear. How do businesses today mitigate the risk of checks? They ask for identification.