Surely, a great deal depends on the devices on offer, and how they can easily be installed?
With a move from incandescent to CFLs, did people generally go out and buy adapters to allow fitting multiple CFLs into each existing socket in order to consume the same amount of electricity?
Maybe if there had been lots of 250W-equivalent CFLS consuming ~50W, some people would have bought them to replace 60 or 100W bulbs with, but would everyone (or even most people) have done that even in the absence of a great price premium?
I might occasionally jump up a grade (putting a supposed 60W equivalent CFL in a desk light that had had a 40W in), but most of the time I'd tend to just do a straight lighting-equivalent replacement.
Presumably, when decent LED-based bulb replacements are available, they'll generally be used in the same way that CFLs were.
Also, at the moment, even though it's just about possible to make a bulb-sized LED unit about as bright as a regular ~60W bulb, there are heat issues even then, and in the forseeable future, it'd be very hard to fit LEDs consuming anything like 60W into that volume unless there was active cooling - there'd just be far too much heat generated.
In the short/medium term, it's also likely that unit prices for LED-based bulb replacements of a given quality would be charged vaguely in line with output, and that that up-front cost would tend to push many people away from stepping up in terms of output even if running costs were negligible or the people paid no attention to such costs.
In the longer run, we might end up with buildings designed to use LED lighting and keep the LEDs decently cooled, rather than retrofitting into existing wiring, but in such a situation, there'd be the likelihood of having some reasonably intelligent control systems.
In any case, it seems likely that a higher maximum light output would end up costing rather more than a lower maximum output in terms of space, LEDs, optics and electronics, and particularly in the case where lighting is distributed, at any one time the materials cost per installation seems likely to scale roughly linearly with total output.
Though it's easy to be too subjective, candles really are dim, oil lamps not great, but with electric lighting, people do generally seem to act as if they think a certain amount of light is about enough - people often use dimmer switches, or selectively turn off some light sources depending on what they are doing, and except maybe for actually doing manual work, often don't use supplemental lighting, except for reading lights when in a room that is otherwise dimly lit or unlit.
Even if lighting was effectively free, there's a limit to how brightly most people would want their living rooms to be when they're watching TV, or their office to be lit if they're using a computer with existing screen technology, and I'm not sure that that limit is radically higher than existing levels.