Overdriving LEDs Question!

I understand that many of us here overdrive our LEDs beyond the maximum rated current based on the CREE (or other mf'rs) datasheets, but has anyone ever contacted CREE to find out the reasons for not driving an LED beyond the rated current? What I am really asking, is the current/voltage (or wattage/amount of energy passing through the LED) the part that is damaging to the LED or is it the junction temperature that they are concerned about?

We're all familiar with the tests done on an XML, for example, that is directly mounted to a giant block of copper and actively cooled while driving it to upwards of 6A without many adverse affects. Additionally, an XPG driven at 1.5A without the proper heatsinking will also see some color shift over an extended period of time. This leads me to believe that it is the heat that is the main problem.

On the other hand, turning up the voltage to an LED will quickly damage the LED and cause color temperature shifts and other problems.

I'm really just trying to get a better understanding of how CREE or other mf'rs choose the "maximum current ratings" for a certain LED and why they choose that particular number (all ratings seem to curiously land on very round amperages -- 350mA, 700mA, 1.0A, 1.5A, 3A, etc.). Does anyone have any first-hand information from the manufacturers on this matter? Opinions and speculations are also welcome, however like anything else, taken with a grain of salt.

Thank you all in advance!

"Turning up the voltage" on an LED is misleading. LEDs are current-based devices. One XM-L may run brightly at 4.0v, and another will fail at 4.0v. Why? Because individual LEDs have different 'forward voltages' at test current. The current that will pass through an LED at a given voltage is different for each LED, and increases greatly with slightly increased voltage. So adding 0.1v to the voltage (With unlimited supply) may double the current.

Cree rates their LEDs to convenient values for production of power supplies. An LED is no good without one. Their published data are averages that are used to compare LEDs, so of course they are nicely rounded. The maximum ratings are a combination of best performance, efficiency, and useful life. A single XM-L carefully installed may accept 6 amps, but I don't believe anyone has data of what this does beyond 10 thousand hours. One major goal of big-shot lighting is 'lumen maintenance.' Fluorescent T5 tubes get around 90% maintenance at 40,000 hours (And worse at 50,000 hours). LEDs want to do as well or better, and higher currents make that less likely for several reasons.

Drive current directly changes tint. Temperature also changes tint. And yes, a 'burned in' LED has a long-term tint shift. Each of these effects are broadly:

More current: Whiter, bluer at extreme upper range.
Temperature: Whiter, bluer at extreme range.
Duration: Cooler over time.

Thank you AnAppleSnail for that very in-depth response and explanation! That gives me a better idea of what "goes on" in an LED, for lack of better words haha!

—the good news is, in our portable lighting hobby,
most of us will discontinue using our LEDs long before damage occurs because we will upgrade
for something with more lumens, more runtime, better tint, better CRI, or all of the above.

Of course, those in situations where lack of 100% reliability and robustness may put their lives in danger
will always want something at spec or even underdriven,
but for most folks,
we will not use our portable lighting devices until even close to the published MTBF.