This is something we will never agree on.

I expect not - but I'm hoping you might at least consider the fact that the figure you commonly cite for DGU's is certainly flawed, and way too high.

The statistical problems with these surveys are enormous. Their reported results are completely inconsistent with all the data on crime incidence (the respondents would have to experience crime rates an order of magnitude higher than what other people face in their area). They're completely inconsistent with data on gun ownership (Kleck & Gertz report that half of DGU responders are women, even though women are only 20% of gun owners and about 14% of justifiable homicide committers). The internal data is completely absurd (nearly 16% of defenders report firing at their attacker, and 8% report wounding them - both of which are ridiculously high on their own, but combine for a 50% hit rate which is beyond what we observe both police and criminal firearm discharges). And they're structured in a way that is guaranteed to be wrong (if you don't correct for telescoping bias, people will over report the frequency of X in their past - no matter what X is).

I have no expectation that this will change your mind on gun ownership - either generally, or for yourself personally. But perhaps it might lead you to be more skeptical of that two million DGU figure that you mention here from time to time.