I'd imagine more US companies than most of us realize have become much more 'metric' over the last 20 or so years. Obviously, complete metrification of an industrialized country certainly isn't a cheap or easy thing to do, and it can't (won't) happen overnight, as is easily confirmed by our Brit and Aussie friends here.
In some ways, the conversion would be simpler today than it would've been 30 or 40 years ago. I doubt any of us here own a single CNC machine that's inch-only. Push a button on the panel to change the default measurement system, or add a G word to change it for a specific program. Because such a great majority of machined parts today are cnc-made, we're at an advantage as compared to all manual shops of the past where every dimension on every print would've required conversion, and a single metric thread could've stopped lathe production dead in its tracks.
One of the more difficult aspects in our business involves dealing with metric parts made from inch dimensioned stock. We're definitely not going to just wake up one Monday morning and find all the mills now have 12x50 flat stock available instead of 1/2x2. At Michelin we dealt with inch-dimension stock in an all metric company every day, and if its approached sensibly, its not a big problem at all. "Sensible" means you have designers who have enough common sense to know you can substitute 3" round for 75mm on a shaft with 60mm bearing seats on its ends, and you don't need to take 1.2mm off the stock size on that other 4 feet of stock between the ends that's doing nothing but spinning in air. It means the designer realizes those 4 M10 tapped holes on a 75mm square pattern don't know whether they're sitting on 10x150 flat stock or 3/8x6 flat.
Working with what's available is nothing new in our trade. Its a matter of using your head to come up with workable and economical solutions. If we NEED a piece of plate to be .700 thick for some particular application, we start with 3/4 and mill or grind it to what we need. We don't start searching for a mill that'll roll us some .700" stock. OTOH, if we're looking at a fab drawing for a machine frame and the designer has spec'd 2 15/16 square tubing, we're likely going to ask if 3" will work, or else let him know he'd better start looking for a couple extra truckloads of dollars to get his frame made.
Economically manufacturable designs start with a knowledge of what's readily available, and add a minimum of labor to that material to create the finished product, regardless of whether the end result is Imperial or metric. If the designer is incapable of creating a metric product without having a full array of metric stock at his disposal, chances are his inch-dimensioned masterpieces of design ain't nothing to write home about either.
The desirability of creating foreign markets for your product would seem to be the most likely driver for conversion to metric. Getting the government involved would likely be one of the worst things that could happen, based on observing the results of their involvement in most anything else they decide to mess with. The idea that we'll just continue as is and the rest of the world can take it or leave it is obsolete from an export standpoint, mainly because we ain't the only dog in the show anymore.
Imagine a company in Germany has narrowed down the choice of some new machine to two candidates. One is US made and the other Italian. All their current machinery is European or Japanese. Both are about the same price, and the US one has the reputation of being the best and most reliable by some small margin. But, the US made machine brings the added baggage of needing to supply the maintenance department with $30K worth of inch-size tools that'll be necessary to maintain this machine, and useless on anything else in the plant. Chances are, simplicity of standardization and overall lower first cost will outweigh some rumored operational superiority, and the US manufacturer loses the sale to the Italians. OTOH, if the US machine was metric, and was indeed superior to the competition, the sale would be a relatively easy one at the same price point, and likely doable even if the price was a bit higher than the Italian machine. People are generally much more willing to pay a premium for perceived quality and reliability than they are to pay the same premium for extraneous support equipment because a machine is non-standard as compared to their other equipment.
For the most part, we're used to working on both metric and imperial equipment, and don't give much thought to one over the other. That doesn't mean the rest of the world is just dying to have the same mixed up mish-mash of overloaded toolboxes and stockrooms that many American companies deal with on a daily basis in supporting equipment from domestic and foreign origin.
In regard to Hu's theory on the Imperial gallon.... Likely true, and there's a parallel in the liquor industry in the States. Used to be, you could buy either a pint or a fifth. A fifth being a fifth of a gallon, or 4/5 of a quart. When the liquor industry went metric, you'd sorta think the fifth woul've been resized up to a litre, but it wasn't. Instead, it was downsized to 750ml. Predictably, "bottle engineering" made the visual difference between the 750ml and the fifth containers nearly imperceptible, and I didn't ever hear anything about a price decrease asociated with the volume decrease. No doubt the margins are real tight in liquor manufacturing, and most of them used that little windfall to keep themselves afloat
