The tripod theory seems ideal for a car with a very low centre of gravity. For most RTR cars I think it would be debateable.
The Brownlee Theory of Slot Cars is that you are trying to make the guide go around in the groove as fast as possible using two variable and one controlled contact patches (the rear wheels and guide respectively). In a real car you are trying to make the centre of gravity of the car go around as fast as possible by employing four variable contact patches (all four wheels) hence the benefit, at least in the good old days, of four wheel drifting through corners.
By the way, I say variable because the path of the tyres is not controlled by the track or groove the way the guide blade is strictly limited to travelling around in the groove.
I'd say the least critical job of the guide's two jobs is keeping the braids in electrical contact with the rails.
Yes, you need contact, but before we get too hung up about it let's consider that the braids have a bit of movement, or droop. On the GT40 in front of me that amounts to about 3mm. Not sure how far a sprung guide moves, but not much more. More importantly, the braid droops to about 1-2mm above the bottom edge of the guide blade, meaning that you'd need to be able to corner with only that little bit of blade in the slot before the lack of electrical contact is your worry rather than de-slotting.
Where sprung guides might help is in hard acceleration as the car tends to rise up at the front, although of course the lack of contact is its own self regulating traction control so is not entirely undesirable. The potential penalty of sprung guides is that at the point where the car has little or no effective weight over the guide due to the action/momentum of the car, over a crest or bump say, the sprung guide is pushing the car higher. Better hope thsi doesn't coincide with a corner. No absolute right and wrong here since on some tracks with some cars a sprung guide would be better and vice versa.
The real hard work done by the guide is steering the car around. Here is where real car and slot car dynamics vary fundamentally.
I think all would agree that you want as much weight on the guide during cornering as you can get to ensure the front of the car will follow the line of the corner and not de-slot and plough on and off. Only when the front/guide is in the groove can you start to worry about controlling the rear axel (via throttle inputs) to make your time in the corner as short as possible.
Slot racers can however borrow an idea form the 1:1 boys who, on cars with no aerodynamic down force, use weight transfer under braking to increase the grip of the front tyres and hence raise the entry speed into a corner. Too much and they understeer. On a slot car understeer is technically impossible of course, so braking into a bend can only add weight to the guide giving more than enough front end bite. I say more than enough since the rear will spin around long before you chisel the groove into a new, wider arc.
Back to tripods - bet you thought I'd forgotten! The move from stable straight ahead to stable cornering requires the weight transfer from front to right or left. If we assume a slot car has no suspension (big assumption, I know) then when the car begins to turn the mass of the car will act through the centre of gravity (CG). This should be as low as possible, but it must have some height. If the car tips up when cornering - assuming the tyre retains grip - then the CG can only get higher. The car will tip more, assuming speed, radius etc remain unchanged.
This twisting/tipping force must be counteracted by the outside wheel. Assuming the rear tyre has grip, the resisting force is acting through the distance from the wheel to the centre line. Therefore the greater the width of the car the more twisting force it can resist. Hence what we all know: low, wide cars have more grip.
A pure tripod slot car assumes the vast majority of the mass of the car is at the rear and therefore the rear axle is taking the vast majority of the twisting action. Think Vanquish MG base chassis and you'll get the idea.
However, add a body, interior, driver and front axle - a pretty heavy cosmetic part on most RTR cars - and the mass is more forwards, pushing the CG towards the centre of the car.
In a tripod car the resistance to tipping is all at the rear, but with the CG in the middle the car tips towards the front outside corner. It is at this precise point that having a front wheel touching the track and capable of taking some weight will stop the cat tipping up higher and the guide lifting from the slot.
Ideally it will be instantly effective i.e. already supporting weight, but in the absence of working, tuneable 1/32 springs, roll bars (sway bars) and dampers, most RTR cars make do with an axle riding just off the track, but coming into action as soon as the car tips.
At this point it is worth mentioning that the tiny amount of roll that is needed to bring the front axle into play might, if the braids were very stiff, mean the inside braid lost contact with the rail, but the previously mentioned droop compensates for this.
Ninco are master of this stiff and low front axle ride height trick and new Scalextric cars are pretty good too (although old Scalextic so-called big wheel saloons - TR7, 911, Rover 3500 - are the worst). The only downside is you need a pretty smooth track or else the low riding, firm front axle will lift the guide out of the slot on bumps.
Anyway, bed time.