[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Dead Duds - sorry, but this just isn't physics...

Greetings All,

Probably not the most propitious of topics over which to stop lurking, but... 
all of the talk of "it's physics, end of argument" is so much baloney, I'm 
afraid. I don't want to debate the _model_ Scott has described, as that really 
isn't the issue for me. Let us _assume_ that the model, in its entirety, is 
logically coherent, and that the empirical information he's entered into it is 
accurate. Does that mean that the model's predictions are correct? No.

A model, no matter how complex, is a very poor substitute for reality. As we 
can't mathematically deal with the whole of reality in one go, models are a 
very valuable tool for analysis, but only if one understands and accepts their 
limitations. No model of the Torsen, no matter how elegant, can fully take into 
account all of the factors which determine the progress of a car through a 
turn. The model Scott advocates appears to imply the transfer back and forth of 
large amounts of torque, which, within the model he uses, leads to a rapid, 
violent succession of understeering and oversteering. Fine and dandy, but only 
for that model, and only under specified conditions.

Dave, and others, believe (as far as I can judge from their posts, with 
apologies if I misrepresent them) that, regardless of the model's internal 
logic (or otherwise), other factors, beyond the scope of Scott's modeling, 
leave the car safe and predictable through the same maneuvers. Can the circle 
be squared? Certainly. They _observe_ that, in reality, under the circumstances 
in which they have tested their cars, the model's conclusions are inaccurate. 
Other's may observe that under differing circumstances, the model's conclusions 
are valid, and the spiders run wild.

If the physical property suggested as a cause of spider bite was overwhelming, 
then every aggressive driver would see it on a regular basis. Apparently not 
so. That doesn't mean the physics are incorrect.

Take a bumblebee. Measure it, test it, apply (entirely reasonable, and usually 
accurate) aerodynamic models to it. (1) Theoretical conclusion: The bumblebee 
cannot fly. It weighs too much, its wings are too small and beat to slowly. (2) 
Practical observation: Bumblebees fly. (3) Logical corollary (assuming we 
accept the physics of the aerodynamic model): (i) other factors, beyond the 
model's scope, assist flight and/or (ii) the aerodynamic model does not apply 
well to small round insects with black and yellow stripes.

What Scott seems to be doing is making a series of flying logical leaps:

1) From the existence of the physical property (neither accepted nor debated 
here) _within_ his model
2) To the idea that this modeled property is accurate in the real world (quite 
literally debatable)
3) To the belief that whatever element of that property remains valid under 
real world conditions is significant enough to overwhelm other physical factors 
mitigating the property (likewise debatable)

At the end of the day, given Scott's experiences (and others') and given his 
math (and the tech papers etc), it seems likely, from the evidence presented, 
that rapid transfers of torque are possible, and hence that spider bite _can_ 
exist under certain circumstances. This belief is aided by the physics model, 
but it can only be _proven_ by the model under rigorous application of the 
ceteris paribus assumption, i.e. never in the real world.

Dave's experiences (and others') and a little thought to the imponderables 
which lie beyond the math and the modeling suggest that, under a very wide 
range of circumstances, the condition _does_ _not_ apply.

Personally, as far as I have been able to replicate Scott's description of the 
circumstances (backed up by Dave's method of going round and round the 
roundabouts), it doesn't happen to my car, under any circumstances I have yet 
been able to generate. To me, it's more important how the car actually drives, 
rather than how, in theory, it might.

For those for whose reality involves getting bitten, that's none too cheering, 
but whatever property of the Torsen (if indeed it's the guilty party) generated 
their circumstances, that property reliably fails to generate the problem for 
other people with other setups or at other places and points in time. Physics 
holds the car to the road, physics pushes it off, but I'll accept the "it's 
physics, end of argument" assertion only when I see a model that contains every 
imponderable, every parameter, every circumstance. Until then, Scott's argument 
will remain not "The Physics", but just "One possible physical element".


'94 S4, which no doubt just adds another datapoint to Qhip's dislike of the car 
and its drivers.