A comment on the smart guns thread captures an argument I'd also seen from others:

As the owner of a self-defense gun, I want just one thing: A 99.9999+ percent probability that, when I pull the trigger, I will hear a 'BANG' and not a 'click.'

The existing, 100-year old, low-tech technology comes pretty close to delivering that level of confidence, and any 'smart-gun' innovations are only going to lower the probability of firing; so why take the chance?

Even if the probability of a successful firing is 'only' reduced from, say, 99.9999 to 99.9998; that's logically equivalent to doubling the rate of mis-fires. Why risk it? If you're worried about your kids, keep your gun in a safe -- or, better yet, take your kids to the firing range and educate them on safe firearms handling at an early age.

This, it seems to me, illustrates well a common problem that also appears in arguments for restricting guns (and, I'm sure, in arguments on many topics): A focus on one risk to the exclusion of others.

Keeping my gun in a safe, after all, reduces my probability of a successful firing by much more than 0.0001%. I'd rather have a gun I can keep unlocked and loaded by my bedside at night that's 99.9% reliable than a gun in a safe that's 99.9999% reliable but that I'll have a 5% chance of not reaching in time.

In a sense, this is the same sort of error that people make when they say "if gun control can save only one life, it's worth it," or when they use "safe gun storage" to mean storage of a gun in a locked safe. Those who make these arguments ignore the possibility that gun controls may cost lives (for instance, by preventing many effective self-defense uses), or that a gun that can't be reached to protect yourself in time isn't "safe[ly] ... stor[ed]." Likewise, the argument I quote focuses exclusively on the costs to self-defense of "smart guns," and ignores the costs to self-defense of the alternatives that the argument itself puts forward.

Of course, the argument also understates the costs of another alternative that it puts forward, by overstating the effectiveness of "education." Of course I'll "educate [my boys] on safe firearms handling at an early age." But I'm educating them to do lots of things -- not cry, not eat too many sweets, not fight with each other, and so on. With luck, over time the education will teach them to behave well. But there will always be lapses, especially when the children are relatively young. I'm not going to rely on safe firearms handling education to keep a 10-year-old boy from playing with a gun; it may decrease the chances of his doing so, but it won't reduce them to zero or to anything close to zero.

If I lived in an extraordinarily dangerous part of town, I might think the risk avoided by having an unlocked gun by my bed will exceed the risk created by it to my boys. (As I said, it's always a mistake to focus solely on one kind of risk.) But fortunately I live in a pretty safe place, so I'm not going to keep the gun outside the safe for many years to come.

It's only safe gun technology -- if such a thing is feasible -- that would lead me to keep the gun immediately handy. And, as I mentioned before, for that the reliability need not be 99.9998%. It only needs to be higher than the effective reliability of a locked gun, factoring in to the effective reliability the delay that unlocking the gun would create.