First off: how do you get users who are quite comfortable using MS-DOS-based applications to switch to a new OS, or at least a new user interface? In short, how do you get people to switch to Windows? The initial answer was to build a version of Windows which ran on top of MS-DOS. This way, if Windows was unreliable, or at least unfamiliar, the users could still return to a DOS environment. Additionally, make sure that DOS applications could run within Windows. Once people started using the Windows user interface, and started to enjoy the benefits of it (graphics and peripheral drivers being provided by Windows, instead of each app needing to provide its own drivers), they eventually upgraded from DOS-based apps to Windows-based apps. A few revisions later, make Windows the primary environment, instead of a "bolt-on" for DOS (Windows 95). Finally, after people have had time to switch, kill off most of the DOS compatibility (Windows NT/2000). At this date, there are very few DOS apps still in active development, and most newer versions of Windows can't run them at all. In the intervening years, however, you could use both until you were comfortable with the new environment.
Another example: the MacIntosh. The original machines were based on the Motorola 68000 and its progeny. Later on, however, it was decided to abandon that architecture and move to the PowerPC architecture. Since the two were NOT instruction set compatible, this was a major shift. The solution was to write the BIOS of the newer machines to emulate the older instruction set. Consequently, the older software would run on the newer machines, albeit less efficiently. As time progressed, the OS became a mix of 68k instructions and PowerPC instructions, while a "pure 68k" version was still available. Eventually, though, the OS and the applications became "pure PowerPC." The latest versions of the MacOS will NOT run on original hardware (I believe 8.x was the last version which provided any backward compatibility to the 68k architecture; any Mac gurus out there feel free to correct me). In the interim, however, you could still run current software on your older hardware, and your older applications would still run on the newer hardware.
The lesson to be learned is this: to avoid being completely shut down by the "chicken and the egg," you need to provide a period of backward compatibility. For the hydrogen economy, you need a timeframe where vehicles can use both gasoline AND hydrogen. Most proponents of hydrogen are advocating the use of fuel cells instead of Internal Combustion Engines (ICE's). The logic is that fuel cells are more efficient, therefore we should do away with internal combustion. I tend to agree with them, but not just yet. Most fuel cells still cost about $10 / watt of output. Consequently, if you need a steady output of 24 horsepower (enough to maintain highway speeds for a small or mid-sized, aerodynamically designed vehicle), that's about 18 kilowatts. That's about $180,000, just for the fuel cell. With a price tag like that, these vehicles WILL NOT be a sales success anytime in the near future. The fact remains, however, that ICE's WILL run on hydrogen gas. They are less efficient, certainly, but they will work, with some modification. This much was proven in the early 1970's (during the "energy crisis"). Consequently, the period of backward compatibility for hydrogen has, potentially, already begun. Unfortunately, few people are clued in.
We can wait another decade or two for the major automotive manufacturers to get off their collective duff, spend a fortune in R&D, and develop an affordable fuel cell. Or, we can spend about a couple thousand dollars to convert an existing vehicle to a "bi-fuel" configuration. This typically involves adding gaseous storage for the hydrogen, finding some way to add it to the intake air (a carbeurator of some kind; it's hard to "inject" a gas), and dealing with the backfiring which typically accompanies hydrogen use (hydrogen is much more volatile than gasoline; Roger Billings found that injecting water into the intake stream fixed the problem, and improved the engine's efficiency by about 20%). From that point on, you might have to hit a switch to convert from one fuel to the other, or a more automated system could simply use hydrogen when it is available, then "phase in" the gasoline as the hydrogen supply ran out. If modern, heavily computerized automobiles could be upgraded with this capability, the transition could be completely transparent to the driver.
Yes, internal combustion engines are less efficient than fuel cells. However, if the automobile is a hybrid of some kind, that improves the efficiency of the vehicle. The Honda Insight is routinely quoted at 65 mpg in efficiency. A kilogram of hydrogen has about the same amount of energy as a gallon of gasoline, so it is reasonable to assume that a hydrogen-converted Honda Insight would get about 65 miles / kilo. Most fuel cell vehicles get 50 miles per kilo (or less), and cost at least 10x as much (I believe $250,000 is quoted as the "production cost" of the Honda FCX). And yet, most pundits claim that running ICE's on hydrogen are a "non-starter," because the fuel cell is so much more efficient. These pundits are full of it. I mean, even a Toyota Prius (54 mpg) would get more than 50 miles / kilo, and the conversion would still be much cheaper than the fuel cell vehicles which Honda and Toyota are currently leasing to various organizations (usually government organizations; if you can influence the policymakers, they might allocate more funds to study the idea).
The market is ripe for one or more companies to develop hydrogen conversion kits for existing ICE-based automobiles. Such conversions exist for natural gas, but I know of none (yet) for hydrogen. The converted vehicles could use hydrogen when it is available, use gasoline when it is isn't, and still be maintained by the existing infrastructure. Some individuals aren't waiting for that to happen, and are doing their own conversions. These are often kludges, based more on trial and error than sound engineering. I'm not sure that I'd want to buy such a vehicle, and I'd be very surprised if insurance companies are willing to cover them.
If this idea flies, it becomes a matter of hydrogen production and hydrogen storage. The demand for hydrogen fueling facilities will rise, effectively solving the problem. Later on, when the fuel cell technology matures, we may be able to add fuel cell "Auxiliary Power Units" to existing vehicles, to provide electrical power for accessories (and for hybrid drive systems), improving the overall efficiency of existing designs. Then, maybe two decades from now, the infrastructure would be in place, motivated by existing hydrogen use, for pure fuel cell vehicles to be practical.