Aside from this practical reason why giant mechs will never happen, there is something more fundamental. The mass of any given organism/machine is proportional to the cube of it's scale. However, it's strength will only increase linearly or at most with the square of it's scale... This is why there are no huge insects or any other large creatures with exoskeletons. Every time the size of a creature/machine goes up by an order of magnitude or so, there must be a new design paradigm. Therefor, a 400 ton machine that walks like a person and has armor like an insect seems rather unlikely.

Oh, and just to be picky... "robot" controlled by gamers is by definition not a robot; It is a remote.

Partially true. A good example is that lobsters and crabs are, I believe, the largest creatures on earth with exoskeletons. However, a low gravity (or high bouncy) environment does nothing to modify the mass of a creature, only it's weight. You still have to deal with the full inertia of the machine, so it will be just as difficult to accelerate as it would on earth. If our hypothetical mech were running across the lunar surface, it would have to apply exactly the same amount of force to move forward or to stop as it would on earth, it just wouldn't have to spend as much energy to keep it from hitting the ground. Low gravity would certainly help though. A giant mech on the moon would just end up being ponderously slow because of the immense force it would still take to accelerate.

True, but that isn't the point. Energy requirements are a problem for something like this, but you could always use a jet turbine, right? This has less to do with energy generation than structural integrity. You can't just make a structure bigger and expect it's strength to scale equally with it's mass.

If you have a steel strut of a given size, and you double every dimension, it will be roughly twice as strong for any given direction of force applied to it, but it will also be eight times more massive! Because of this problem we use trussing systems and I-beams, but this is just another example of the new design paradigm required by a change of scale.

The same rules apply to a gasoline engine. Yes, if you double it's size you will also double it's energy output. However, the forces that the engine components can withstand will have increased more slowly than the forces being generated and you'll end up having to additionally strengthen all the components so it won't simply blow itself up. It amounts to the same thing in the end. You have something that will far more than double in weight when you double it's output, assuming that there is no significant redesign.

Besides, those huge mechs just make big targets. I frequently find that killing a small, fast-moving snark from HL1 is more frustrating than killing a Gargantua. They move quickly and are so small that they are extremely hard to hit.

As far as weapons, I imagine it would be cheaper to make a lot of little bots that move extremely fast than a large mech. Imagine you're surrounded by hundreds of small bots that shoot bullets at you. Your chance of survival is zero, even if you can smash some of them.