Thanks econ41, much appreciated.
I never understood how truthers came to the conclusion that 65% (or 2/3) of g meant zero resistance. Which to them meant columns removed by explosives.
I don't know if the following is something any Truthers claim, but here is a conceivable mechanism to solve your confusion:
Suppose the collapse is initiated by blowing out / making disappear by magic all columns (12 feet of column lengths) between two floors. The top then falls in free fall (ignoring resistance by 63x63 meters of slab pushing down on and out of the way ll that air below).
Then, juuuuuuuuust before the next floor is hit, again all columns (12 feet of them) are magically removed, such that the falling top hits a floor slab that is stationary, but without any vertical support.
And so forth.
Because the falling mass hits a floors slab every 12 feet that is (almost) stationary, and that slab has considerable mass, a momentum transfer will occur, within a very short time interval, where the single slab is greatly accelerated, and the falling top somewhat decelerated - before the whole mass continues in free fall.
This model would actually converge on a net acceleration of (and here I am blowing a secret) 1/3 of g (not 2/3!). But start out near g, and decrease as more and more floor slabs are consumed per second.
---
HOWEVER, for that to happen, it's not actually necessary to blow out vertical supports, and to blow out vertical support, you don't need to blow out columns!
Because most of the falling mass hits floor slabs, very little hits columns. The loads on the slabs is transferred to columns via floor truss seats, so all it takes is to blow out those seats.
And in reality, it didn't require any demolition to shear off the truss seats, nor would the shearing (necessarily) cause a deceleration in addition to the momentum transfer that happens anyway:
Momentum transfer in an inelastic collision - and these collisions are largely inelastic - always results in a decrease of Kinetic Energy (KE), and that KE will flow - into "destruction" - breaking, inelastic deformation, also a bit heat. That energy lost upon each collision is far more than what is needed to shear off the truss seats. And so they are sheared off (or some spot in the load path that is even weaker), without requiring any additional transformation of KE to destruction.
Therein lies one of the main thinking errors in Chandler's and Szamboti's and many a Truther's heads: They don't appreciate that the 1/3 or 2/3 of Potential Energy released by the mass simply going down is more than sufficient to bend, break, tear out the weakest elements in the load paths supporting the horizontal elements most affected by the avalanche of mass.
---
B&Z original, September 2001 argument was that even in the only theoretically conceivable event that no weaker load path than fully braced, perfectly vertical, undamaged columns were available, crushing them would eat up less energy than is available from gravity alone.
it turns out that they had their numbers somewhat wrong, and that substituting for better numbers would move their conclusions more into "undecided" territory.
But as
@econ41 never tires of pointing out, any model that builds on "column crush" is a wrong model to describe the reality of the WTC collapses.
---
Now as for the supposed "smoothness" of the acceleration curve and the lack of a "jolt", I think most counter-arguments have probably been mentioned in this thread already (I didn't read all of it), but here is my summary:
1. Their data is smoothed simply by choosing to sample at a low rate, i.e. only 6th frame (1 data point per 0.2 s in Chandler's video) and computing averages over those. Any jolt that might be there is thus mathematically certain to get smoothed out, possibly beyond recognition. If you look at the raw data, of course it is full of jolts. These Truthers say that raw data is too noisy, due to measurement error, and that is certainly true. BUT that means that, absent a robust error analysis, and possibly even
with a robust error analysis, it may not be possible to decide either way if there is a jolt or not.
2. They measure points on or near the roofline (antenna?). There is a lot elastic and complex structure between those points and the "crush" interface where jolts would actually occur, and so any real jolts will get significantly attenuated, if not entirely smoothed out, by the action of said intervening structure.
3. The tops of both twin towers tilted significantly before they descended as a unit. This alone ensures that there would not be a single jolt affecting the entire top structure - the only case where measuring the downward motion of a single point on the periphery would have any chance of being representative for the whole. Instead, each column would feel some resistance (if only that from truss seats as they are being sheared off) at its own point in time - there'd be many local "jolts" somewhere during every measurable time interval (video frame length).
3b. Showing the jolt at one Verinage demolition misses two important differences: a) That demolition is DESIGNED to have the top go down horizontally, evenly, without tilt, in order to maximize the force at which the structure below gets hit when it gets hit. It is DESIGNED for columns to hit columns (because they actually want to crush them; b) this was a concrete structure with concrete columns and concrete beams - which is probably why the demolition was planned that way: They wanted to maximize the amount of concrete that gets crushed by gravity.
---
Chandler's tracking point and plot, at around 55 s into the video already refered to above, suffers from several more problems
Source: https://www.youtube.com/watch?v=ZjSd9wB55zk&t=55s
1. The wall, the roofline of which serves as Chandler's target, is toppling towards the camera position, so there is some rotation in addition to linear descent. Because the wall starts out vertical, the trajectory of the target point due to rotation toward the camera, which is located lower, would make it look like moving
up, when in fact it starts to come down.
Later, as the wall leans seriously, the rotation may increase the apparant velocity (and acceleration) of the observed point. Although the rotation itself surely decreases later into the descent, which makes matters even more complicated - Chandler considers none of this
2. The straight line he plots through his 17 or 18 chosen data points does not fairly represent acceleration throughout those 3.4 seconds - but he pretends it does. He claims explicitly that the onset of fall is "sudden", that the 2/3 of g are there right from the beginning. But quite obviously, you can draw a straight line to best approximate the first 5 data points, and it would be considerably less steep, representing a lower acceleration. Similarly, you could draw a stright line through his last 4 data points, and get a higher acceleration. So this would make it look as if acceleration is actually increasing over time. (It probably isn't - this may be more an artifact of the rotation I discussed above)
3. As already pointed out, he does not differentiate between collapse stages - initiation (top starts to lean...), transition (all columns get severed over a brief interval) and vertical progression ("crush" throughout entire cross section), which are governed by different processes. If the collapse were initiated by blowing out all vertical supports, he'd have seen instant acceleration near g (and little to no rotation), but instead, during the first 0.8 seconds or so, according to Chandler's plot, "resistance" is the highest. This could be consistent with a lot of columns still having connection, and there being only tilt. Then the top releases - acceleration goes up; it appears to vary (go up and down) even in Chandler's data due to chaos - it is not actually smooth, despite his choice of rather long time intervals.