Merged Discussion of femr's video data analysis

[LashL]

femr2: Please answer directly. Does your "video data analysis" relate to any 9/11 conspiracy theory or not?

If so, just identify the 9/11 conspiracy or conspiracies to which it relates. If not, then this thread (which was started by tfk, not you) may have been started in the wrong sub-forum.

It's a simple question. Please answer it.

 

[femr2]

femr2: Please answer directly. Does your "video data analysis" relate to any 9/11 conspiracy theory or not?

If so, just identify the 9/11 conspiracy or conspiracies to which it relates. If not, then this thread (which was started by tfk, not you) may have been started in the wrong sub-forum.

It's a simple question. Please answer it.

It is, as is all discussion within the entire realm, associated with the question: Demolition, or not ?

The *conspiracy* as you put it, is demolition, or not.

What else do you think it would be related to ?

What conspiracies do these relate to ?

• What are the finest examples of 9/ll Truther stupidity that you've ever encountered?
• Engineers named Mike who think Truthers are nuts
• "Try clapping your hands 110 times in 10 seconds"
• WTC Attacks not 911 in fact are 711
• Any 9/11 conspiracy debunkers in Bristol, UK?
• Look Here For a Commentary on the Latest News from AE911Truth

Does a thread which *debunks* a *twoofer* claim belong in this sub-forum ?

Does a thread which *debunks* a NIST claim belong in this sub-forum ?

 

[NoahFence]

It is, as is all discussion within the entire realm, associated with the question: Demolition, or not ?

It took a moderator's threat to get you to even type that word.

So have at it - answer your own question - Demolition or not?

 

[W.D.Clinger]

Show me where NIST presented their derived acceleration formula or used it at all...

NIST's derived formula for acceleration expresses the slope of their curve in NCSTAR 1-9 Figure 12-77, which they checked by performing a linear regression for the 2.25 seconds NIST defined as Stage 2 "for discussion purposes". That was used in turn to justify point (2) of the last bullet item in Section 12.6.

That's how little this matters. For all femr2 has written about NIST's acceleration curve in this thread and elsewhere, nothing he's said could possibly affect any of NIST's conclusions outside of that one small sentence fragment in that one bullet item.

They are very different. We disagree. Shock.

Let's show everyone exactly how much difference it makes:

As shown by that graph, the differences between NIST's nonlinear model and femr2's preferred Poly(10) compression are less than the noise in femr2's data throughout most of the collapse. There's a small disparity in the neighborhood of 13 seconds, and another small disparity that begins around 16 seconds. (The disparity in the neighborhood of 11 seconds involves femr2's Poly(10) approximation, not NIST's model.)

Those disparities become more pronounced in the acceleration curve because differencing amplifies disparities just as it amplifies noise. Although femr2 would have you believe that none of us can estimate the noise in acceleration curves, that is untrue. Because the disparities between NIST's model and femr2's data are hidden within the noise of femr2's data throughout most of the collapse, we know that the disparities between NIST's acceleration curve and any accelerations that can be reliably inferred from femr2's data are likewise within the noise throughout most of the collapse. femr2's ability to pretend otherwise depends upon his personal inability to estimate the noise in his acceleration graphs.

 

[femr2]

NIST's derived formula for acceleration expresses the slope of their curve in NCSTAR 1-9 Figure 12-77

Figure 12-77 is a velocity graph, not acceleration, and shows the non-linear formula for VELOCITY, the adjacent difference point data and the LINEAR acceleration regression.

There is nothing in the figure suggesting any use of a further derived formula for acceleration from their non-linear model.

As shown by that graph

Your graph is unreadable. Try using colour. Also, trying to make your point with a view not zoomed is ridiculous.

(The disparity in the neighborhood of 11 seconds involves femr2's Poly(10) approximation, not NIST's model.)

Again, you are deliberately using data outside the bounds of applicability for the Poly(10) curve. Tsk. How deceitful of you.

 

[W.D.Clinger]

NIST's derived formula for acceleration expresses the slope of their curve in NCSTAR 1-9 Figure 12-77

Figure 12-77 is a velocity graph, not acceleration, and shows the non-linear formula for VELOCITY, the adjacent difference point data and the LINEAR acceleration regression.

That's why I wrote "the slope of their curve". As should be obvious to readers who understand calculus, you're just complaining that NIST overestimated the intelligence of its audience.

(The disparity in the neighborhood of 11 seconds involves femr2's Poly(10) approximation, not NIST's model.)

Again, you are deliberately using data outside the bounds of applicability for the Poly(10) curve. Tsk. How deceitful of you.

I produced that graph before you had stated any bounds of applicability for your Poly(10) approximation, which you were then touting as a model worthy of comparison with (and, in your view, superior to) NIST's. Judged as a model, your Poly(10) model is wildly inaccurate in the neighborhood of collapse initiation.

It was only later that you accepted Myriad's characterization of your Poly(10) approximation as a lossy compression of your data. We can only speculate as to why you thought that compression would be useful, but I agree that it is unfair to judge a lossy compression outside of the interval it was intended to compress.

I have wasted enough time on your posing without wasting more in producing a new version of that graph that omits your Poly(10) approximation. You may not like that, but it's a consequence of your original misrepresentation of Poly(10) as a model rather than a lossy compression.

 

[femr2]

That's why I wrote "the slope of their curve".

You were trying to imply that NIST USED their non-linear model derived to acceleration. Obviously you failed to do so, as they did not. They, as I have said repeatedly, used their linear regresion for accelerarion, and only their linear regression for acceleration.

I produced that graph before you had stated any bounds of applicability for your Poly(10) approximation

So ? You know now, and knew when you posted your recent post that you were *outside the bounds*. Tsk, tsk. Deceitful.

which you were then touting as a model

Nonsense. I have repeatedly highlighted that I was not building a model at all, simply containing the data in a form which could be differentiated.

Judged as a model

Repeating your own false assertion, again. Tsk, tsk. Naughty, naughty

your Poly(10) model is wildly inaccurate in the neighborhood of collapse initiation

Poly(10) is, AS YOU KNOW, meaningless before my stated T0. YET AGAIN you are manipulating context. tfk used to do that when he was struggling. Not a pretty sight.

It was only later that you accepted Myriad's characterization of your Poly(10) approximation as a lossy compression of your data.

ROFL. I never said the poly(10) curve was a model. That's your own false assertion. I had no problem with Myriad suggesting such, of course. It's interesting to watch you try and manipulate the discussion. Lots of useful prior discussion elements you are misrepresenting. Might be prudent ro remind you (and your avid readers) of the true nature of the discussion eh

We can only speculate as to why you thought that compression would be useful

See above. And expanded explanations have been given numerous times. Short term memory you have.

but I agree that it is unfair to judge a lossy compression outside of the interval it was intended to compress.

Then stop doing it

I have wasted enough time on your posing without wasting more in producing a new version of that graph that omits your Poly(10) approximation.

Leave it in, that's fine, but make it READABLE. Colour-coding is dead useful you know, as is higher resolution, as is looking at the data in much finer detail...aka zoom in

What was the phrase you used for the Poly(50) curve for displacement again...? (R^2 value )

You may not like that, but it's a consequence of your original misrepresentation of Poly(10) as a model rather than a lossy compression.

LMAO. I misrepresented nothing. Ironic that in a sentence you write accusing me of misrepresentation you are doing precisely such.

I did not present Poly(10) as a model.

 

[Oystein]

...
The use of such data refutes some assertions by bodies such as NIST, and also affirms others. The data both *debunks* some *Truther* theories, whilst also *debunking* other *debunker* theories.
...

In other words, NIST would have to revise some conclusion ("*debunks* ...theories")? Which?
See, you apparently claim that this thread is about debunking theories, but you explicitly refuse to state which! Which shows that thos thread is NOT about changing any theory, any conclusion at all! It is explicitly NOT about any theory, and NOT about any conspiracy theory, since you steadfastly refuse to discuss any and all theories and conclusions in this thread!

...
Which conspiracy do these address ...?

1. What are the finest examples of 9/ll Truther stupidity that you've ever encountered?
2. Engineers named Mike who think Truthers are nuts
3. "Try clapping your hands 110 times in 10 seconds"
4. WTC Attacks not 911 in fact are 711
5. Any 9/11 conspiracy debunkers in Bristol, UK?
6. Look Here For a Commentary on the Latest News from AE911Truth

...

(I changed the list type to have the bullets numbered)

Te answers are pretty simple and straightforward:

1. Addresses potentially ALL CTs, particularly the whackiest ones
2. Addresses the CTs promoted by AE911truth.org
3. Addresses any CT that claims all, or most, floors of the twin towers must have been attacked by whatever means; most usually explosives
4. Addresses/presents a particular, cabalistic/numerological CT
5. Addresses Richard Gage's European tour and his CTs
6. Again, addresses the CT claims by ae911truth.org

(AE911truth.org's CT is: explosive demolition of all three towers, using nanothermite and thermate plus ordinary high explosives)

 

[Oystein]

Discussion of the CONSEQUENCES and CONCLUSIONS from femr2's video data analysis

In another thread, femr2 has presented plenty of numbers and graphs derived from plenty of video of the WTC collapses.

Without going into detail, here is the drift of it, as I perceive it: femr2 has sampled the same videos that NIST used, but more finely, de-interlacing them and analysing sub-pixel information, and then applied several smoothing algorithms, among which he prefers the Savitzky-Golay.
There can be hardly any surprise that his data and graphs are not exactly the same NIST's, and we may agree, for the sake of debate, that his data and curve fitting is better than NIST's by some degree.

Let us agree in general, for the sake of this thread, that femr2's data analysis is precise and well done.

Now, in that discussion over there, femr2 tries to avoid at all costs to debate what conclusions can and ought to be drawn from his data analysis. Are the differences to NIST's analysis relevant? In particular, in what way does his data analysis warrant a change in NIST's conclusions? And finally: does this change go in the direction of making any conspiracy theory, such as explosive demolition, a more likely, or even the most likely explanation of ALL of the available observation?


Please do not debate the technical quality and physical, mathematical or engineering appropriateness of femr2's methods here! Let us stick to accepting his take on the data as given, and concentrate strictly on the consequences of femr2's data analysis!



Have fun

 

[ozeco41]

...Are the differences to NIST's analysis relevant?...

This is probably the second biggest issue of confusion or conflict - so much criticism directed at femr2 can be attributed to both sides failing to define "relevance"

For my money femr2's investigations show the potential to either confirm NIST or to differ from NIST - depending on which bit of collapse they are applied to - in either case at the detail level. Therefore NIST's global conclusions are not likely to be challenged, they may be confirmed OR a different explanation emerging which does not alter the global conclusion.

Some members here have made a lot of comments to the effect that if femr's work does not change global conclusions it is a waste of time. It may be for those persons who are only interested in the big picture. Arguably they should recognise that the details are not their area of interest and stand aside from discussion. However others have a genuine interest in the details of "how" and that interest is as valid for them as the "global only" interest is for others.

My own position was stated way back at the start of these discussions involving femr2's work. I'm interested in any outcomes but not interested in discussing the methods. (With a couple of disclaimers and extra points thrown in. One of them being that I would be satisfied that NIST demonstrated plausible mechanisms for the key stages of the collapses - I don't need 100.000% accuracy - I would not be at all bothered if it turned out that they were a bit out in the details.)

 

[NoahFence]

No.


Oh, I do. Go for it...

(The level of blind support for NISTs clearly crappy methods is astounding. Funny stuff.)

I'm not talking about blind support for NIST. I'm talking about your mistakes.

 

[Oystein]

This is probably the second biggest issue of confusion or conflict - so much criticism directed at femr2 can be attributed to both sides failing to define "relevance"

I think I supplied a few criteria in the OP. Let me boil it down to this:

Does his analysis provide reasons to come up with new conclusions about why the three towers fell, or change existing conclusions?
​

I am okay if different poster in this thread have a different definition of "relevant", and each is invited to provide his own. The above is mine. I just want to see any results beyond more graphs and the observations that some graphs look somewhat different than other graphs.

 

[NoahFence]

No way in hell does he come here and answer anything. This is why I didn't make the thread - but good luck to ya.

 

[Oystein]

No way in hell does he come here and answer anything. This is why I didn't make the thread - but good luck to ya.

I guess his non-participation could mean that he sees no relevance himself, and draws or changes no conclusions from his months and years of work. I could be wrong of course, and he just likes to keep secrets.

I see the possibility for valid reasons not to formulate conclusions yet. In that case, I'd be interested if there is a game plan, if it involves drawing conclusions later, and what premises would have to be met for this to happen.

 

[W.D.Clinger]

Nonsense. I have repeatedly highlighted that I was not building a model at all, simply containing the data in a form which could be differentiated.

You contradict yourself. Differentiating a lossy compression of your data makes no sense. When you differentiated your polynomials, you were using them as models.

As you have now admitted that differentiation was the purpose of your polynomial approximations, we know your analysis was inept.

Differentiating your Poly(10) approximation yields an upward acceleration of over 12 ft/s^2 at your t0=11.875s, which does not decline to zero until 12.2s. That sustained upward acceleration is not present in your data. It is a compression artifact created by your mindless use of high-degree polynomial curve fitting and exacerbated by your naive choice of an endpoint that's in the neighborhood of collapse initiation.

Now that you have declared "demolition" to be the relevance of your analysis, it is clear that your analysis is worse than useless for its avowed purpose. Your analysis has created the false impression that a significant and sustained upward acceleration occurred at beginning of the collapse. That is highly misleading.

NIST's nonlinear model yields far more accurate accelerations near the beginning of the collapse.

Note, by the way, that NIST's linear regression is limited to the interval defined as Stage 2 for the purposes of discussion. NIST's nonlinear model of acceleration is the only model NIST could have used for Stage 1. NIST used that nonlinear model of acceleration to define the endpoints of Stage 2. (NIST could not have used its linear regression to define those endpoints, because the interval had to be defined before the linear regression could be performed over that interval.)

Poly(10) is, AS YOU KNOW, meaningless before my stated T0.

Your Poly(10) approximation is meaningless, period.

ROFL. I never said the poly(10) curve was a model. That's your own false assertion. I had no problem with Myriad suggesting such, of course. It's interesting to watch you try and manipulate the discussion. Lots of useful prior discussion elements you are misrepresenting. Might be prudent ro remind you (and your avid readers) of the true nature of the discussion eh

The true nature of this discussion is that, whether deliberately or incompetently, you are promoting an analysis that interprets compression artifacts as upward accelerations near the beginning of the collapse. That misleads those who hope to find some scientific-sounding excuse to believe in deliberate demolition.

Since demolition is the relevance you claim for your analysis, and your analysis is highly misleading with respect to demolition, your analysis (as opposed to your data extraction) is worse than useless.

I did not present Poly(10) as a model.

If Poly(10) and Poly(50) are not models, why did you compute their derivatives?

 

[femr2]

In other words, NIST would have to revise some conclusion ("*debunks* ...theories")? Which?

Ooh, let me see...

• NIST WTC 1 initiation sequence is wrong. Implications galore.
• NIST WTC2 impact orientation and trajectory is wrong, leading to the impact damage assessment being wrong, leading to all subsequent study being based on incorrect damage assessment. Pretty darn serious that one.
• NIST WTC7 early motion behaviour wrong. No vertical kink, but primarily N-S twisting, leading to incorrect T0, incorrect *phase 2* metrics, incorrect published freefall metrics, etc. Much, much closer to global freefall that NIST state. ~20% slower than freefall, not 40%.
• NIST tracing process severely flawed as highlighted with the rather long list of issues I have presented.

etc...

On the other side of the fence, the trace data and early motion studies prove many *truther* theories which rely upon instant changes in rate of motion to be false, along with data to back up arguments against the oft bandied suggestion that 2.25s of freefall proves...whatever.

You see, there are issues on both sides

See, you apparently claim that this thread is about debunking theories, but you explicitly refuse to state which!

ROFL. See list above. See repeated assertion of same and additional throughout this thread.

 

[femr2]

I'm not talking about blind support for NIST. I'm talking about your mistakes.

Yet you fail to highlight any. Cool

 

[NoahFence]

Yet you fail to highlight any. Cool

I've got a notepad document all ready for the copy and paste treatment as soon as you tell the class (in the other thread because that would be off-topic here apparently) how your data disproves NIST's conclusion that fire took down WTC 7.

 

[NoahFence]

NIST WTC 1 initiation sequence is wrong. Implications galore.

In my world, "galore" means "a lot".

Care to name say...two or three that change the outcome?*




*outcome being collapse - feel free to use the other thread.

 

[femr2]

You contradict yourself. Differentiating a lossy compression of your data makes no sense. When you differentiated your polynomials, you were using them as models.

Nonsense. Simply a container for the data. That's why I prefer the Poly(50) to Poly(10). Simply a means to smooth the data in a form that can be used to derive acceleration profile data.

Certainly not using them as models, so, no, you're utterly wrong I'm afraid. Simply containers for smoothed data.

Of course, the addition of the S-G curve to the pot has simply highlighted that excellent accuracy of the Poly curve results. It's all good

Differentiating your Poly(10) approximation yields an upward acceleration of over 12 ft/s^2 at your t0=11.875s, which does not decline to zero until 12.2s. That sustained upward acceleration is not present in your data.

Have you ever bothered to actually LOOK at the raw displacement data in any detail ?

It is a compression artifact created by your mindless use of high-degree polynomial curve fitting and exacerbated by your naive choice of an endpoint that's in the neighborhood of collapse initiation.

ROFL. I note you are avoiding the Savitzky-Golay profile like the plague, and focussing upon your pet choice of curve, the Poly(10), which of course has issues near the beginning and end of the ROI.

You are utterly transparent.

Do you have any issues with the Savitzky-Golay curve ?

Now that you have declared "demolition" to be the relevance of your analysis

This entire sub-thread is about the same topic, over and over

it is clear that your analysis is worse than useless for its avowed purpose.

ROFL. The trace data has provided another means to *debunk* many a *truther* theory, whilst highlighting many serious errors on NISTs part.

Your inept bias of personality is your own problem. You are a funny, funny man. Sad, but funny.

Your analysis has created the false impression that a significant and sustained upward acceleration occurred at beginning of the collapse. That is highly misleading.

What is misleading is your interpretation. Who said the upward acceleration at the start of the curve indicated real-world upward motion ? NO-ONE, except you kinda implying it here. You are misleading, and doing it deliberately. It's not like you are not aware of how you are deliberately adding inference in order to support your personal desire to present negative information.

The trends graphs are fine. They do contain some residual noise, and the S-G curve is better than the Poly curves. All show the acceleration profile in much greater detail and accuracy that the pathetic NIST curve when you derive it yourself (which, of course, they didn't do, as they didn't use their non-linear model for acceleration, and instead used the linear regression ONLY).

You can keep banging your head against the wall if you like though

NIST's nonlinear model yields far more accurate accelerations near the beginning of the collapse.

ROFL...

The NIST model has the acceleration at 10ft/s^2 at a time when the NW corner has not actually started sustained vertical descent

Note, by the way, that NIST's linear regression is limited to the interval defined as Stage 2 for the purposes of discussion.

Indeed, and the specific data points (determined via adjacent difference) they used are highlighted.

NIST's nonlinear model of acceleration is the only model NIST could have used for Stage 1.

Absolute nonsense. I suggest they used *eyeballs*.

NIST used that nonlinear model of acceleration to define the endpoints of Stage 2.

Nonsense. They used *eyeballs* and a ruler I imagine. Pathetic.

You have nothing to support your claim.

(NIST could not have used its linear regression to define those endpoints, because the interval had to be defined before the linear regression could be performed over that interval.)

Utter nonsense.

Here is what NIST actually state...

Now, tell me W.D.Clinger, where exactly are you pulling your assertions from ? I could make a highly accurate suggestion.

Your Poly(10) approximation is meaningless, period.

Nonsense. It is a container for smoothed displacement/time data and the derived acceleration profile is very similar to the S-G curve, confirming its accuracy.

The true nature of this discussion is that, whether deliberately or incompetently, you are promoting an analysis that interprets compression artifacts as upward accelerations near the beginning of the collapse.

No, YOU are promoting that interpretation. Painfully obvious and transparent as well Will, tsk. And under your own name too. Shame. Think I'll write it up in a manner similar (but better of course) than the one you did for MT

If Poly(10) and Poly(50) are not models, why did you compute their derivatives?

To determine acceleration. You MIGHT want to realise that the METHOD of differentiating the Poly(10) and Poly(50) curves was symmetric differencing. Mathematical derivation of the Poly(10) curve was only performed as a cross-check. Again, I'm not treating them as models. You are. They are simply containers for smoother displacement data.