Ben Franklin invented the lightning rod, they told me in school, and ever since then we've had lightning rods to protect us.

So where are they all?

Looking around, one seldom sees any, even in lightning-prone areas. A few houses have visible lightning rods, but only a few. And while the wires for a lightning rod system can be concealed within a building, the rods themselves need to stick up from the top of the roof.

I don't have a definitive explanation, but after looking into it a bit I figure it's a combination of:

  1. The lightning rod installation industry has adopted a very high standard and thereby priced themselves out of the business, as far as most homeowners go, and

  2. Modern houses are full of electrical wiring which lightning will find and take advantage of, so the risk is no longer primarily to life and limb as it was in Ben Franklin's day, when humans were some of the most conductive things in the house and thus a preferred path for lightning. It still can happen that lightning comes through a roof and hits a person directly, but the chance of it is much lower; most risk is of using something such as a computer during a lightning storm. Even then, it's rare for someone sitting at a keyboard to be on the easiest path to ground. Now, lightning is quite powerful: even outside its main path one can still get injured by it. But even such injuries are rare indoors, according to a survey paper ("Lightning-Caused Casualties In and Near Dwellings and Other Buildings", by Ronald L. Holle, 2010), which also indeed reports a trend of them getting rarer:

The category of “Indoors” inside all types of buildings was found to account for 4% of U.S. lightning deaths and 12% of injuries from 1991 to 1994 (Holle et al. 2001, 2005) summarizing NOAA’s Storm Data. These rates are much lower than 100 years earlier, when 29% of deaths and 61% of injuries were indoors (Holle et al. 2001, 2005).

This does not mean that lightning rods are useless; far from it. A direct lightning strike on a house is likely to destroy everything electronic that's connected to the house's wiring, and is also likely to set the house on fire. But insurance pays for such damage. Of course fires can be deadly; per the same paper most of the fatalities included in the above 4% "were due to fires at night involving elderly and/or disabled people, or children". But fire is a different sort of death than instant electrocution by a lightning bolt; all the usual measures for fire safety still apply (such as smoke alarms, sprinklers, and available escape routes), and are even more important for other causes of fire, which predominate over lightning.

To further return to the same source,

Note that many dwellings catch fire every year, but do not involve any casualties. Holle et al. (1996) showed that over 300,000 claims for insurance compensation are paid to owners of homes and small businesses each year for lightning damage in the U.S.; the percentage involving fires is unknown.

But despite such a sizable number of claims, insurance companies don't seem to encourage people to install lightning rod systems. Likely the reason is that it's too expensive: offering people a few dollars off their bill in exchange for installing a $2000 system would have too few takers. And that seems like the probable way the math works out: 300,000 is a lot of claims, but still is only about one per thousand people in the US. If the average claim is for $10,000, the average savings per person is $10. Of course there's typically more than one person per insured house, so maybe the insurance company's savings would be $30 per insured house. Much of that would be eaten up by the cost of publicizing and administering the discount; maybe they could offer half of it to the insured. Fifteen dollars a year wouldn't come near to justifying a $2000 expenditure -- and that does seem to be the sort of cost figure that people throw around for installation of a lightning rod system. (Obviously these numbers can be argued with in all sorts of ways, and the $10,000 number in particular is just my wild guess, so feel free to adjust it to taste, or even better email me with the real number if you have a good source for it. But this is generally how the calculation should be done.)

Now, the sort of lightning rod system that Ben Franklin would have installed wouldn't be anywhere near $2000. But modern lightning rod systems are supposed to conform to NFPA 780. Having read it, it seems like a decent standard, but is rather demanding. The copper cable to conduct lightning to the ground, for instance, is supposed to be at least 29 mm^2 in cross-sectional area. That's the size of cable (about 2 gauge) that one might find in a car for the wiring to the starter motor; it's about ten times as much copper per foot as is in typical house wiring (12 or 14 gauge). Pricing for cables sold for lightning service is several dollars a foot. In contrast, the electrical power distribution industry is said to have decided that 6-gauge wire (13.3 mm^2, and about a dollar a foot) was suitable for most needs, as being enough to handle 96% of direct lightning strikes; for the other 4% it was more economical to just replace the destroyed equipment. (The source linked to is a copy of "Lightning Protection: Taming Thor's Thunder -- on a budget", by Kenneth J. Meyer, from Popular Communications magazine.)

Adding to the cost is that an ordinary electrician is not supposed to do the work; NFPA 780 is complicated enough that a specialized contractor is to be called in, meaning extra visits and higher fees. There are requirements such as that metal objects near the system be bonded electrically to it, with pages of rules detailing how to do so and when it is to be done. It generally seems like a good standard to follow if cost is no object, but generally cost is an object.

And it doesn't seem like there are any serious downsides to a weaker system, as compared to no system at all. Maybe if it were ridiculously weak, say, a system built with 22 gauge wire, it would attract lightning to the rods but then the wire would quickly vaporize and the arc would start a fire. Now, the industry's publicists like to say that lightning rods do not attract lightning, but that's ridiculous: that's how they work. There is an industry standard model for this: lightning propagates down in steps of 50 meters, with each step having a different randomly chosen direction though generally somewhat downward, and at each step hitting the closest grounded object if there is one within that range. NFPA 780 prescribes using that model to decide where to place lightning rods to attract lightning away from the rest of the building. Still, by that model, the chance of attracting lightning from the next house is minimal. So the publicity line is not entirely empty: they won't attract lightning from all over the neighborhood, or even probably the next house, but they do attract it locally (and then conduct it safely to ground, or in the case of a 22-gauge wire not so safely).

But even that 22-gauge system doesn't seem likely to cause harm: any lightning that hit it would most probably have hit the house anyway, and it's probably more benign to have it running down the outside of the house along the path of the former wire than it would be to have it going through the house's structure: a fire on the outside of the house is more likely to be extinguished by rain. The main downside of a weak system seems to just be insufficient coverage. In other words, a weak system still seems better than none.

I haven't delved into how exactly NFPA 780 is enforced -- what part of it is industry opinion, what part of it is regulation, and what part is lawsuits after the fact. No doubt such details vary from state to state. But it seems like a world where an ordinary electrician could just "run up a lightning rod while I'm at it" and do a half-assed job of it would be a world with less lightning damage.