ok now you make me wanna think about it.
lets say it drops 5% a year
year 0 33000 lumens
year 1 31350
year 2 29800
year 3 28300
year 4 26900
year 5 25500
year 6 24300
year 7 23000
year 8 21900
if youre replacing them once every two years you spend $90 average every 2 years and average 31350 lumens (actually prob just under 31000 as these bulbs drop to like 96% luminosity the first week)
diy cobs are approaching the $2/watt mark with drivers and sinks
so youd be buying 45W worth of cobs every two years = 8355 lumens youd be tacking on
so yeah looks like cob is a better investment than even the LEC bulbs alone. youd have to do a little better analysis of lumens/watt since you'd be running the already relatively inefficient LECs at an even lower efficiency at first (and pulling more power with LECs+COBs). so your lumens/watt might dip overall, but as far as lumens/dollar there is no reason to ever buy new LEC bulbs just supplement with cobs instead for the same money and you'll be way way ahead in overall lumens/dollar.
and yes i realize the above numbers are exaggerated because of the 2 factors i mentioned
(running an old inefficient lec bulb into the ground and also adding cob wattage to make more total wattage)
but it also doesn't consider that LEC bulbs are prob close to their price floor, where semiconductor tech should continue to drop 10% in price and gain 10% in efficiency on a 12-18 mo cycle.
maybe ill crunch the numbers later when i have more time i have a feeling that supplementing with cobs will still be way way more economical than spending money on bulbs with such a short relative life