Here's the article:

http://physics.ucsd.edu/~tmurphy/papers/lumens-per-watt.pdf

It's true an incandescent lamp gets hot and a lot of the energy goes into heat, but note almost all of it is in the IR radiation range. Our eyes can't see this radiation. Theoretically, if someone could change the electron configuration of the atoms inside the material used to make a filament, may it be carbon or tungsten or some other material, so that only the visible range is radiated, a 100w incandescent lamp with such an engineered filament would be as bright as an ordinary 1000w incandescent lamp! Of course it is currently not possible with todays knowledge of science.

According to Wikipedia , an ideal blackbody radiator at 6600K (6300°C or 11,400°F) can reach 95 lm/w. But this is impossible to heat the filament to such high temperatures, since there are no known materials that remains solid at these temperatures, which are hotter than the sun surface.

I think that infra-red energy from incandescent bulbs is what helps make a room feel warm and inviting (especially during the winter). L.E.D.s (even warm colored models) don't have that welcoming feel, while the new halogen bulbs do.

If some kind of phosphor was developed or discovered that converted infra-red to visible light, we could have the efficacy of fluorescent lighting in an incandescent bulb!

@DetroitTwoStroke: But such phosphor is not possible: To become a color conversion phosphor, higher energy photon have to rise the electron few energy levels up and it is it's return via all the intermediate energy levels, what could turn the original energuy from the high energy photon into multiple lower energy photons.
When you excite the electron by a low energy photon, it may jump up to the one step higher level, but then it will return back to the original one, emitting the same photon back. So the only thing you get is a scattered light.

@"100% efficient ibncandescent": All that is based on one simple theory: Hot surfaces radiate only on wavelengths, which they are absorbing. Very frequently is cited the "black body radiator", what mean the black surface absorb all the radiation, so when heated up, it radiate in all wavelengths.
So to get higher efficiency from incandescent, you will need something, what could be called an "IR body", so a surface reflecting all wavelengths (mainly IR), except the visible range. When this would be heated to the 3000K range (and assume it retain it's color even on such temperatures), it will radiate only on wavelengths, where it absorb the light, so in visible.

But the main practical problem is: We are just happy our technology allow at least some materials to somehow endure the required temperature, so no one even dare to challenge the surface "color" (so the spectrum of reflection vs absorption/radiation) on such temperatures. And it happen, than all known materials basically just turn black on such high temperatures...

Some materials could be discovered with required color at the high temperatures, but it will take time and quite huge investments.
And it will never be perfect, so efficiency gain in 80% reduction in IR emission would already be a huge scientific success.

But in the mean time the illumination technology development turned more towards cold quantum generation, where the already attained efficacy is already way higher. And the spectrum is of no problem either: Technically it is only a question of efficacy trade off. And it is the mass market, what steer this trade-off to Ra80 as sufficient for most application, so prefer the higher efficacy instead of higher Ra.
But technically it is no problem to make a LED or fluorescent with a visible spectrum part equivalent to e.g. an incandescent, it will be still way more efficient than the incandescent itself, but less efficient than the counterpart with only Ra80, so therefore it will be way less demanded on the market.

There are phosphors out there that reduce the wavelength of the light they give off relative to the exciting light.  They need multiple exciting photons to give off an emitted photon, and I don't know what their conversion efficiency is.

Long live the incandescent lamp as far as I'm concerned. It was cheap to make and in it's basic form left no negative impacts on the environment in the recycling process. They should never have been banned nor anyone have the alternatives forced upon them by the hand-wringers of this world. I keep buying them whenever I can.

After I censored a part of laugh in my video of the Tungsram TC-100 on Eltam EHID electronic ballast (Thing that allowed me to hear what we said before the censored part), I've noticed that when Eltam engineer said that a 100W incandescent lamp is equal to a radiant heater and produces mainly heat, I've said that it produces mainly infra red radiation, and it almost don't produces convection heat (Compared to CFL and LED of course).
The fact that the incandescent lamp produces almost no real heat and mainly electromagnetic radiation (Infra red)(This is true also for radiant heaters btw), what most people don't know, is why this theory of 100% efficient incandescent lamps actually has been raised.

I haven't seen any incandescent energy efficients anywhere. Some of the incandescents get too hot and dangerous too. It is true that incandescents can warm a house. For me, I would stick with 50W or less for incandescents. I am also a fan of safety.