From TFA: One declassified memo, which seems to be the conclusion of initial research and prototyping, says that Project 1794 is a flying saucer capable of "between Mach 3 and Mach 4," (2,300-3,000 mph) a service ceiling of over 100,000 feet (30,500m), and a range of around 1,000 nautical miles (1,150mi, 1850km). [...] According to the cutaway diagrams, the entire thing would even be capable of vertical takeoff and landing (VTOL).
VTOL, Mach 3++, 100,000ft - the only thing it was missing for it to be the perfect military aircraft is a Romulan cloaking device.
If the "flying saucer" could meet those specifications, it would still be classified. Q.E.D. it didn't work.
It would be one thing if we were reading leaked documentation about an experimental aircraft designed 10 years ago. But we're not. We're reading declassified info about an aircraft designed some 50-60 years ago. If it actually worked, a half-century would be a pretty reasonable timeframe from experiment to slightly more mainstream application. If not commercial application, than certainly military or scientific application. And its existence would have turned up by now. Alternatively, it would be so effective and groundbreaking that it would remain classified to this day, and nobody would have declassified any of this documentation.
EDIT: From the (extensively documented) Wikipedia article on the very similar Avrocar:
"In flight testing, the Avrocar proved to have unresolved thrust and stability problems that limited it to a degraded, low-performance flight envelope; subsequently, the project was cancelled in September 1961."
[1] http://en.wikipedia.org/wiki/Lockheed_Martin_F-35_Lightning_...
[2] http://en.wikipedia.org/wiki/Lockheed_Martin_F-35_Lightning_...
Those specifications were achieved in that era on other programs. I think the answer requires a deeper look.
A saucer wing has a very low aspect ratio, which has very low aerodynamic efficiency in cruise. The high-aspect trapezoidal wing we commonly see today is used because it produces much less drag while providing sufficient wing area to lift the aircraft. This has been understood since the 1930s, if not earlier.
I can't see why a saucer airframe would be desirable unless it spent a lot of time in backwards and sideways horizontal flight, where it might have better stability than a traditional wing. The complexity of the controls and thrust arrangement wouldn't seem outweighed by this though.
Actual test performance of Avro's saucers never exceeded altitudes of a few feet and speeds of a few mph.
"...Project 1794 is a flying saucer capable of “between Mach 3 and Mach 4,” (2,300-3,000 mph) a service ceiling of over 100,000 feet (30,500m), and a range of around 1,000 nautical miles ...
...the supersonic flying saucer would propel itself by rotating an outer disk at very high speed, taking advantage of the Coandă effect. Maneuvering would be accomplished by using small shutters on the edge of the disc ..."
However, the other handling qualities of these designs make them difficult to fly. Stability can be a real problem. I would guess that air handling problems were the issues that cancelled the saucer program. Perhaps the modern fly-by-wire computer control systems used today might make it easier to fly a saucer.
I disagree however with the last statement of the article: "If flying saucers were somehow faster or more efficient or capable of lifting heavier loads, we would almost certainly see them in a commercial setting."
Just because flying saucers aren't used commercially doesn't necessarily imply that they are inefficient. It just means that the flying saucer technology we have at the moment is more inefficient than the fixed wing technology. Fixed wing technology could simply be ahead because it has seen a lot more iterative development.
Endeavors of this kind, that carry a huge upfront investment in the initial technology compared to what's on the market now, tend to be things where a free market really performs badly.