The venerable incandescent lamp is under assault–rightly or wrongly–as an inefficient blight on our environment (note: we're not discussing the validity of that argument here). Some incandescent opponents say that CFLs (compact fluorescent lamps) are the way to go; others say the CFL is merely a stopgap, and LEDs are really the way to go and CFLs are only an interim solution. My view is simple: I don't know what will happen or the "best" way to go, period.
But LED-based lighting certainly has some real virtues, which I don't need to detail for this audience, you know it all too well. I do think that the way we'll really learn about using LEDs for illumination, as well as ramp up high brightness LED (HB LED) production and experience, is to use them in dedicated, embedded applications, where there is no need to retrofit to exiting sockets, power-line control, switches, and related infrastructure or worry about compatibility issues.
For example, check out this battery-powered, LED-sourced work light called the Might-D-Light (and related products) from Cooper Industries. If it works as promised, it's a very good application of HB LEDs. And since it is self-contained, there are no issues related to the LEDs being compatible with existing bulbs (incandescent, halogen, or other standard types).
Products like this take advantage of HB LEDs while not being subject to the challenges and stress of mass markets and the broad spectrum of user understanding, while working out high-volume issues, spurring engineering expertise, and developing mass-production. That's the way to go, it seems to me.♦