Miles-per-gallon ratings have always been less useful to consumers than other measurements of fuel consumption, but as we move to using biofuels and to driving cars powered by electricity, they are becoming completely useless.

What's needed instead is a system that tells consumers how much they can expect to pay to operate a car and how often they'll have to fill up. The system should also tell regulators how much gas will be consumed (if the point of a policy is to reduce oil imports, say) and how much pollution will be emitted. The figures given would be dollars per month and stops per month for consumers, and regulators would be told the pounds of key pollutants emitted per year and gallons of gas consumed per year. The figures would be personalized based on basic information on driving provided by consumers.

There's more on how this could work below, but first here's what's wrong with miles-per-gallon (mpg) ratings.

What's wrong with miles per gallon?

The recent debate about General Motors' claim that the upcoming Volt plug-in hybrid will get 230 mpg is a case in point. The number is nearly meaningless, in some ways overselling the car and in some ways hiding its potential benefits. The fact is that the car won't use any gas at all for most commutes, relying only on electricity stored in the battery--but it will still consume energy and result in power plant emissions. And on long trips, that fuel economy will drop to 50 mpg or less as a gas generator switches on to recharge the battery.

There are similar problems with flex-fuel vehicles that can burn either gas or ethanol. They will get far fewer miles per gallon running on ethanol than running on gasoline. And then you pair a flex-fuel vehicle with a plug-in hybrid and things get even worse. Consumers won't know what they're getting with a miles-per-gallon rating.

Of course, this is something experts have been talking about for years. Some have proposed switching to a miles-per-gallon equivalent rating, which accounts for both the gas burned and the electricity consumed. But this doesn't take into account the fact that this rating will change depending on how far someone drives in a plug-in hybrid or whether this person uses biofuels.

And it doesn't address a fundamental problem with miles-per-gallon ratings that holds true even if everyone's using gasoline. The point of a car is not to consume gallons of gasoline. The point is to traverse miles. That's why some have suggested (and some countries have implemented) a fuel- or emissions-per-distance measurement. That would tell me how much gas it will take to travel a certain distance, say 100 miles. A variant, meant to account for the possibility of using electric power or gas, is to put that in terms of energy used--kilowatt-hours per 100 miles.

But this measurement still doesn't work when I'm switching between gas and electricity. And it doesn't directly tell me what I really want to know, which is how much the car will cost me to drive or how often I'll need to stop for gas.

It really shouldn't be that difficult to get these relevant figures, or at least get estimates that are close enough to be useful. Here's a very rough outline of how a system for doing this could work.

The new system

A person considering buying a car would go to a website, something like fueleconomy.gov, and provide a little information: home address, work address, number of days commuting. Total miles traveled per day and per month could be calculated using Google Maps or something like that. Tack on estimates for weekend driving and errands based on surveys. The system would then use data from vehicle tests and from local electricity and fuel prices to calculate cost per month of daily driving and how often refueling will be needed. It would provide numbers for costs using gas or biofuels. It would also automatically provide information about the cost of several longer trips. For someone in Boston, for example, it could give the cost of a trip to New York, a trip to a ski resort in northern Vermont, and a cross-country road trip to LA.

The details of the vehicle tests would have to be worked out. They would have to measure the rate of energy consumption continuously until the car runs out of its stored energy, whatever form or forms that's in, and over a couple of standard drive cycles like the ones used now to calculate city and highway miles per gallon. The data would have to include rate of electricity use and gasoline use, for example, at 30 miles when a plug-in hybrid is still running on batteries, and at 100 miles, when it is using gasoline. And it would have to include data from the use of alternative fuels, if these have significantly different energy density than gasoline (as ethanol does).

In the future the output data could be optimized using actual driving data harvested from GPS data about actual driving conditions in a given area, or even during different times of the day. Consumers could offer more information if they wanted to have a better estimate (including info about where they shop, when they drive, and even driving styles gathered from GPS data). Or they could keep the information given to a minimum if they're concerned about privacy. Even if the estimates are rough, they would be more useful than miles-per-gallon figures given today.

If someone decides to purchase a car, figures (without personal information) would be sent to regulators so they can estimate gas consumption and vehicle emissions.

The system I've described no doubt has flaws and can be improved upon. But the general point I think remains. It's not that hard to calculate personalized figures that can be far more useful than miles-per-gallon ratings, especially as we get a greater variety of vehicles in the coming years.

Automakers may balk. It may be more difficult for them to predict whether their vehicle mix will satisfy regulations without some one-size-fits-all standard. So I propose that at first they're given a backstop--some standard they can use to meet regulations. If the key regulation is carbon emissions, then grams per mile, as is done in Europe, makes sense. Miles per gallon equivalent could also work, since it accounts for electricity use as well as gas or biofuels use. Something like the draft EPA standards GM used to derive its 230-mpg rating could also be useful, if the real concern is petroleum consumption, for an aggregate sense of fuel consumption (although it's useless for individuals).

My guess, however, is that, at least after a couple of years of experience, automakers will be able to estimate whether they can meet regulations. And they may find that it's easier to meet them when consumers actually know how much they're paying to operate a car. I think I'd be more likely to buy a fuel-efficient car if, rather than vague miles-per-gallon ratings, I were provided with clear dollar amounts to consider.