Did you not watch the video? There is no consistency.
There was no consistency between the 'total' input lag, but that doesn't mean the controller inputs don't have their own consistent latency, at least imo, they are being mixed into his larger numbers.
I assume, that when inputs come in from the joystick there is going to be a specific period of time it will require before the hardware/driver will register it as key press.
During the update() phase of the game, the code will poll the input values and update the controller inputs as required and then act on them.
Where it gets interesting is whether or not your controller inputs are verified by the driver before they are polled again.
For the sake of discussion, assume it takes 7ms for wired controller inputs to be verified by the driver, and 11 ms for wireless controller inputs to be verified respectively.
That means if your inputs are delivered 7ms before the next frame, your wired inputs will be captured for next frame. But for the wireless controller, which won't be ready for an additional 4ms, will also miss 'next' frame as well. Ultimately, we will be measuring the number of misses in this case, and not exactly the 'ms' time it takes, though that is probably the most ideal thing to figure out.
NX Gamer worked with average latency as I understand to provide an overall view of what to expect as a whole from our controller behaviour. Clearly game code and OS has a much larger factor in inputs, thus he went to discuss that. But because we're working with intervals close to factors of 16.6 and 33.3 + display lag, you're unlikely to see any other numbers crop up as your total input time.
I'm actually much more interested in the histogram of latency values (Readied, Missed, JIT) than I would be the average in this case. If after 1000 button pushes and I found that wired controllers have significantly less frequency with 'Missed' say 1:50, that's 20 frames worth of more inputs than the wireless controller that player got in every 1000 frames. When you got 1 million dollars as the prize pool, this doesn't seem so trivial to me when you're playing through millions of frames per game.
when people ask between wired and wireless I assume they are trying to get an edge or at least limit disadvantage. if measured correctly there should be a consistent difference between the two, even if it's marginal. That consistency should be proven through a variety of titles and it's entirely possible that different controllers have different latencies as well.
