Much of this LED craze is due to a confluence of latest is greatest is the only way technofaddists and a push for the highest system efficiency possible with no regard to any other contributing factors like initial cost or reliability.  LEDs are best in low level illumination applications like flashlights and for most indication purposes.  Note that that takeover happened due to the benefits of LEDs in those particular applications and not due to some ham fisted scum of the earth bureaucratic myrmidon.  They tend to fall flat on their faces when it comes to unsuitable applications like kilolumen level illumination or anywhere high reliability or high operating temperatures are application considerations.
The only remotely close to universal illumination technology we have or that will exist for the foreseeable future is the incandescent lamp, as the myriad examples out there can attest.
In a sane world, LEDs would mostly stick to flashlights and indicators with various and sundry use in low level illumination a applications.  Interior illumination would be a mix of incandescent, fluorescent and HID lamps, the application specifics dictating the types of lamp used.  Street lighting would be an application dependent mix of incandescent and HID with some fluorescent and low pressure sodium in the mix.  Any other illumination application would utilize whatever technology or technologies were best suited to that application.  Note the common denominator of application specific appropriateness.

By the time the light reaches the ground it's only 25% output and the rest is glare.     I don't know why anyone thought LED is a good idea. 

The basic claim is, that LED is more efficient by better control of the light, due to the light being emitted right away into 2 PI sr angle ("180 deg in space") instead of 4 PI sr ("360 deg in space"). That means, that there is no effort required to handle the 1/2 of the light that goes backwards, and therefore none of the losses from reflecting it

How well the claim works can be seen well in fluorescent lamps :

A 36W T8 fluorescent lamp, complete with electronic ballast, takes about 38W to emit 3300 Lm of light (32W lamp + 6W ballast losses). Out of them 1650 go down, 1650 go up and are reflected off the lantern gear cover, ceiling etc. The reflection is lossy, a normal white surface reflects only about 2/3's of the light

A 20W 4ft T8 LED retrofit lamp is claimed to be a 36W T8 lamp equivalent replacement while cutting down 1/2 the energy use. The 36W lantern is destroyed by removing the ballast, and 20W LED lamp put in. then the user discovers that the LED is way too dim... He picks the package from the trash. Lo and behold. The LED lamp Lm output (in tiny letters) is 1650 Lm....

What happened ?

The LED manufacturer rationalizes that since the LED emit all the light down, all the light reach the target and therefore the lower Lm dont really mean lower resulting light levels. The concealed underlying assumption is that ALL the light emitted from the FL upwards is lost. This is convenient to assume as it gives the best figures to rate and promote the LED

The "problem" is that its not realy all that bad with the FL. White surfaces (lantern, ceiling etc) reflect about 2/3 of the light that hit them. So with the LED lamp the room get 1650 Lm, with the FL it get 1650 + 2/3*1650 = 2750 Lm - thats 2/3's more that he claimed....

By the way, the figures i used are a lot in favor of the LED. A 20W LED tube is visibly darker even than halophosphor 36W T8 that emits only 2500 Lm and not 3300

This explains why LEDs can be darker if chosen just by manufacturer claims rather than by actual Lm data and understanding. But it still does not explain the glare....



When looking up, you see an inch wide T8 tube emitting 1650 Lm. (divide that by 1 inch * 4 ft and youll get the intensity of the light that you see at the tube surface)

With the FL, the background behind the tube is the white ceiling emitting 2/3's of the other 1650 Lm, divided by surface (its not a constant, but integrating the reflection from all of the ceiling will give that 2/3's. Most of the light is reflected from the area near the tube)

With the LED, the ceiling is darker, so the tube is more glary on its background. To add insult, due to the main area of the picture (the ceiling area) being darker, your pupils are wide open to take the glare in



With 2/5's of the light being indirect (reflected from the ceiling), the shadows of everyting (book pages, ...) are 3/5's sharp and 2/5's smooth. Replace the FL with LED, and the shadows become "only" sharp



The spectrum content of LEDs causes higher eye strain, and one might well interpret that feeling as feeling of glare, even when it is not caused by glare but by flooding the eyes with concentrated band of blue light

Also, some led tube manufactures don't use diffusers and reflectors, they just encase the led diodes with a white circuit board backing in s clear tube and call it a day. The light emitted is just lines of dots instead of one smooth area of light (I can't think of a better way to explain this.) They also made fluorescent tubes where they had a reflective coat on the back half of it.

A super-efficient gas discharge lamp maybe.

With the bare LEDs having >140lm/W at 80CRI and still a room for improvement visible, there is no other light source type left in the efficacy race.
The fluorescents have the efficacy drawback of starting from rather high energy pohotons. Then because all has to be converted, you can not go much above about 30% energy efficiency of the light conversion and the real commercial products are already there. So no room to make it any better.
Compare to hypothetical 70..80% conversion efficiency of LEDs (by far not there yet, hence the anticipated room for improvement).
The initial UV efficiency of the discharge in fluorescents is already approaching the "100%", so not much improvement possible there either.
The bare LED efficiency has hypothetical limit of nearly 100% (to fundamental limit there), yet the real devices are just about around 50..60% efficiency.

With HID you need high temperature operation and that inherently means large power needed just to keep the temperature right. When going to fancier, more efficient mixes, it usually means higher cold spot temperatures, so even higher losses in maintaining that temperature. Plus the high arc temperature means high thermal (incandescent) radiation and that means high loses, leaving less energy for the real electron excitation jumps. So again not likely to improve much.

But where the LEDs are lacking the most is the reliability and cost. But both seems to be more related to the limited experience with them (~10 years of wide spread LEDs vs ~100 years of discharges in the industry really does make quite some difference), what should just improve over time. The industry is just in the learning curve there (the existing installations either use very early LEDs known to not perform wthat well or are very short time in service to prove anything yet), nothing else

The only thing I can see being a problem for led's in the short term (and electronics in general) is if there are rare earth materials needed to produce them and those go up significantly in price for various reasons. Otherwise I imagine you'll see what happened to discharge happen to LED's once the market is saturated. 10-15 more years for that maybe?

What's next? Compressed nuclear waste in a quartz tube, and radiated from the outside to keep the radiation in...ok just kidding.   

What's next? Compressed nuclear waste in a quartz tube, and radiated from the outside to keep the radiation in...ok just kidding.   

Yes with a outer bulb with phosphor