NIST blew WTC7 Stage 1 analysis

[DGM]

Not without the appropriate building data, no.

.

I suppose the building drawings just wouldn't do.

 

[DC]

I suppose the building drawings just wouldn't do.

well those drawings do not show where to place the thermite

 

[Starving for Truth]

I don't watch TV very often, but when I do, I make sure to record the program I'd like to watch onto a video cassette rather than watching it directly.

I then digitalize the video with a sampling rate of 44.1 kHz and run it through a Savitzky-Golay filter to smooth the data to the 3rd polynomail degree and interpolate the derivatives with a Chybyshev low-pass filter. The video will still look the same and I will have wasted lots of time, but the smoothing methods are absolutely VITAL for a proper understanding of that particular Simpsons episode.

 

[tsig]

I don't watch TV very often, but when I do, I make sure to record the program I'd like to watch onto a video cassette rather than watching it directly.

I then digitalize the video with a sampling rate of 44.1 kHz and run it through a Savitzky-Golay filter to smooth the data to the 3rd polynomail degree and interpolate the derivatives with a Chybyshev low-pass filter. The video will still look the same and I will have wasted lots of time, but the smoothing methods are absolutely VITAL for a proper understanding of that particular Simpsons episode.

 

[W.D.Clinger]

If you are going to make these kind of posts, make sure you "have a point".

Without that, it's more of that impotent nonsense I've mentioned above.

My, my, look what we have here...

Accuracy of my raw Dan Rather data has been clarified many times, including "praise" from your hero WDC.

As I have said on many occasions, I regard femr2's data as superior to NIST's. I have also explained why I do not hold femr2's analyses of that data in high regard.

As if to remind me of what's wrong with his analysis, femr2 wrote the following:

WDC deliberately used data outside the ROI in order to present a false picture. He was told many times about his "error" and failed to correct it. Almost 1s of data before the ROI was included, skewing WDC's "results".

femr2 likes to think he's the world's foremost (only?) authority on what constitutes the region of interest. As I have pointed out, someone who's actually interested in the mechanism of collapse (which, to be fair, probably excludes femr2) would be at least as interested in the one second of time that precedes the instant femr2 identifies as collapse initiation as in the seconds following.

femr2 doesn't understand why intelligent people would be interested in what happens just before collapse, and he's still upset that I took the trouble to show how poorly femr2's high-order polynomial approximations fit his own data in the neighborhood of femr2's very own designated T0.

("Neighborhood", by the way, is a mathematical term. By mathematical definition, any neighborhood of T0 must include points prior to T0. That upsets femr2.)

Irrelevant to this thread. Quoting WDC's failed "hit-piece" does you no favours, at all.

Although femr2 would have you believe my posts have failed, femr2 has been unable to answer many of them beyond the "incorrect/ROFL/failed/hit-piece" level of his usual rhetoric.

ROFL. The issue, which you fail to recognise due to your incompetence, is that WDC chose to use very narrow symmetric differencing to generate the velocity data, a method he knew was misleading.

Actually, the point of that narrow symmetric differencing was to reveal the noisiness of the data, and to demonstrate the need for some kind of smoothing when calculating an acceleration curve.

In other contexts, femr2 has been happy to cite my demonstration as an endorsement of his smoothing. Here, however, he says I was being misleading.

If femr2 truly believes my demonstration was misleading, then his many citations of that demonstration as endorsement of his methods must have been spectacularly dishonest.

Note also that you have ALSO made the same error as WDC, though in your case I suspect due to lack of understanding, by using near-ajdacent samples to perform your differentiation. That is plain WRONG. It is stupid to do such when treating high frequency data containing noise. Plain dumb.

Which was pretty much the point of my demonstration. Yet femr2 says I made an error.

If femr2 truly believes my demonstration was erroneous, then why has he been using my demonstration to make the same point I was making? It's as though I had said the mid-day sky is blue, and he had responded by saying no, you're wrong, the mid-day sky is blue!

NISTs datapoints are much further apart so, of course, amplification of noise is lower than by the inappropriate method used by WDC.

Has femr2 actually forgotten the purpose of my demonstration?

Ansolute nonsense. Even your revered WDC has stated that my raw data is significantly more accurate than NIST's, a fact demonstrated time and time again.

Yes, I believe femr2's raw data is significantly more accurate than NIST's.

As I've already told you, the poly fits are not models, but simply a means to smooth the data. A compression method if you will. There are no meaningful "parameters" at all. You listen to WDC far too much. Suggest you use your own, er, noodle. He can at least "try" and talk his way out of his mistakes (deliberate distortions). You can't.

Oh, dear.

The archives of this forum show not only that I was among those who characterized femr2's polynomial curves as having more to do with compression than modeling, I also calculated how poorly they perform at that compression task.

femr2 now agrees with me about his polynomial curve fits not being serious models, rather a "compression method if you will", with "no meaningful "parameters" at all." Although he now agrees with me about that, he wants you to think my characterization of his polynomials as having more to do with compression than with modeling had something to do with "mistakes (deliberate distortions)".

I'm sorry, folks, but femr2 is the one who's doing the deliberate distorting here.

As I have explained to you above, it was WDC's CHOICE to use narrow symmetric differencing to derive the velocity data which amplified inherent noise. That's the wrong way to treat the data, as WDC knew well.

As femr2 knew well, I was demonstrating the noisiness of the data and the need for smoothing.

femr2 has been citing my demonstration as endorsement of his methods. If femr2 were honest, he wouldn't be suggesting I had mistreated the data.

I utilised poly fits simply as a means of compression/smoothig....all of which was supersceded by the Savitzky-Golay based derivations.

femr2's Savitzky-Golay curve certainly supersedes his absurdly high-degree polynomial curve fitting.

Savitzky-Golay wasn't a particularly good choice of method either, but it's the best femr2 has come up with. It's too bad femr2 didn't know the Savitzky-Golay algorithm provides both first and second derivatives with a single curve fit.

Pathetic.

I don't think femr2's data analysis is quite so bad as to deserve that adjective, but the dishonesty of femr2's arguments in recent posts does deserve that adjective.

I've told him several times about the problems with WDC's old graphs and methods, but he's still reposting the same images and failed arguments.

I no longer pay much attention to femr2's pathetically dishonest arguments, but all readers of this thread and this subforum should take note:

femr2 is trying to have it both ways. He's been citing my images and arguments with approval when he wants you to think I've endorsed his methods, but he refers to exactly those same images and arguments as "failed" when he wants you to forget my criticisms of his methods.​

 

[beachnut]

...
femr2's Savitzky-Golay curve certainly supersedes his absurdly high-degree polynomial curve fitting.

Savitzky-Golay wasn't a particularly good choice of method either, but it's the best femr2 has come up with. It's too bad femr2 didn't know the Savitzky-Golay algorithm provides both first and second derivatives with a single curve fit.
...

Why are you so darn polite?
He never was able to give reasons he used SG.

Ironically he was using you to defend his methods, many posts ago. I did not have the heart to tell him you were telling him this. He missed it the first time.

Lucky femr2 was not working for NIST, the report would be stuck on a super-minor part of the report, or they would replace femr2.
I find his work reflective of his idea what a technical paper is.

 

[femr2]

femr2 likes to think he's the world's foremost (only?) authority on what constitutes the region of interest.

You're making it up as you go along. No, I think no such thing, however, as the poly fits were simply utilised as a form of smoothing/compression and were generated with data with a specific start time and specific end time, with NO data included outside of that region of interest, the behaviour of the poly fit outside of those bounds is innapplicable. The poly fits are totally unconstrained outside of that region of interest. You know this but choose to ignore it in order to make false and misleading conclusions.

As I have pointed out, someone who's actually interested in the mechanism of collapse (which, to be fair, probably excludes femr2) would be at least as interested in the one second of time that precedes the instant femr2 identifies as collapse initiation as in the seconds following.

Which would simply require regeneration of derivative data with a different region of interest, a point I have raised with you previously.

femr2 doesn't understand why intelligent people would be interested in what happens just before collapse

You are making it up as you go along. I have presented much data about early motion. You are simply trying to justify your errant actions of yesteryear.

and he's still upset that I took the trouble to show how poorly femr2's high-order polynomial approximations fit his own data in the neighborhood of femr2's very own designated T0.

You are making it up as you go along. Your "analysis" was flawed as I have described. I pointed out the ROI issue to you repeatedly, but surprisingly enough you chose not to address it.

Although femr2 would have you believe my posts have failed, femr2 has been unable to answer many of them beyond the "incorrect/ROFL/failed/hit-piece" level of his usual rhetoric.

You are making it up as you go along. I have replied to your posts in detail. This is not the place to begin requoting them all, but I suggest you justify your accusation by posting links to what you think is "unanswered". Given the likely text volume, it may be prudent to do so in the appropriate thread(s).

Actually, the point of that narrow symmetric differencing was to reveal the noisiness of the data, and to demonstrate the need for some kind of smoothing when calculating an acceleration curve.

Yet you and others have cited the graphs on numerous occasions within discussion of how much more accurate you think the NIST data is. Which is nonsense. The information is being misused to flummox the unwary and uninformed.

In other contexts, femr2 has been happy to cite my demonstration as an endorsement of his smoothing. Here, however, he says I was being misleading.

Where ? Links please.

If femr2 truly believes my demonstration was misleading, then his many citations of that demonstration as endorsement of his methods must have been spectacularly dishonest.

Again, where ? Links please.

Which was pretty much the point of my demonstration. Yet femr2 says I made an error.

The error is in subsequent use and recitation. Comparing NIST derivative date (with very low sample rate) to my data (highest sample rate possible) using the narrowest symmetric differencing is invalid and misleading. You even had the gall to justify that action by suggesting "[you] didn't want to run the risk that [your] choice of smoothing algorithm might favor one of the models being compared"

The archives of this forum show not only that I was among those who characterized femr2's polynomial curves as having more to do with compression than modeling, I also calculated how poorly they perform at that compression task.

I am not talking about your characterising of my poly curves in compression terms. I am talking about your (ab)use of data outside ROI and use of narrowest symmetric differencing.

femr2 now agrees with me about his polynomial curve fits not being serious models

You are making it up as you go along. I've never considered the poly fits as models in the slightest. Your self-serving self-admiration is quite funny in this instance.

rather a "compression method if you will", with "no meaningful "parameters" at all." Although he now agrees with me about that

You have a very flawed memory. See above.

he wants you to think my characterization of his polynomials as having more to do with compression than with modeling had something to do with "mistakes (deliberate distortions)".

Oh dear. No. Choosing what you say you think I am saying are your mistakes is very sad. I have been quite clear that your decision to base your "study" on data outside the ROI was wrong. Your decision not to correct it, same. Your decisions to continue to cite it, same.

I'm sorry, folks, but femr2 is the one who's doing the deliberate distorting here.

Clearly not. Quite willing to go into mind numbing detail with this if you like. However convoluted you might try and make it, it can always be straightened out.

As femr2 knew well, I was demonstrating the noisiness of the data and the need for smoothing.

In a comparison on the same graph with NIST data it was simply the wrong method to use, incomparable and no indicater of noise in any comparable way. In other words totally misleading. Subsequent use of that graph and surrounding information has been used to mislead.

femr2 has been citing my demonstration as endorsement of his methods.

Yet again, where. Links please.

If femr2 were honest, he wouldn't be suggesting I had mistreated the data.

I am always honest. You have utilised your flawed "analysis" in a misleading manner.

femr2's Savitzky-Golay curve certainly supersedes his absurdly high-degree polynomial curve fitting.

Absolutely...In 2010...

Savitzky-Golay wasn't a particularly good choice of method either, but it's the best femr2 has come up with.

Do you have a better ?

It's too bad femr2 didn't know the Savitzky-Golay algorithm provides both first and second derivatives with a single curve fit.

You're making it up as you go along. I'm fully aware of how the algorithm functions.

I don't think femr2's data analysis is quite so bad as to deserve that adjective, but the dishonesty of femr2's arguments in recent posts does deserve that adjective.

There's no dishonesty coming from me Will.

I no longer pay much attention to femr2's pathetically dishonest arguments

How rude, especially when there's no dishonesty coming from me. If we need to fine-tooth-comb, so be it.

femr2 is trying to have it both ways. He's been citing my images and arguments with approval when he wants you to think I've endorsed his methods

Yet, yet again, where ? Links please.

but he refers to exactly those same images and arguments as "failed" when he wants you to forget my criticisms of his methods.

You're making it up as you go along...

 

[Starving for Truth]

edit: he's not worth a post

 

[femr2]

As "some" folk seem to still struggle with this issue, I'll explain it again...

Within NCStar 1-9 Vol 2 12.5.3 information is provided about how the "instant of initial movement" was determined...

NIST T₀ Selection

By usage of the brightness profile in NIST Figure 12-75 the exact pixel and (interlaced) frame that NIST selected was determined...

[qimg]http://femr2.ucoz.com/_ph/7/2/378832692.jpg[/qimg]

Source Video | CBS-Net Dub7 47.avi (RAW NIST FOIA - 1Gb DV File)
Pixel | 304, 171
Frame | 5398

That point is the exact start of the NIST Stage 1. The NIST T₀.


Now, that value is set in stone as the start of their vertical descent trace.

The end point of their "trace" is defined as... "The lowest point on the north face of WTC 7 that was visible on the Camera 3 video, prior to any downward movement, was the tops of the windows on Floor 29"

That forces the end point to be within the region marked "B" in the following image...

[qimg]http://femr2.ucoz.com/_ph/7/2/516330913.png[/qimg]

Obviously, it doesn't take a genius to see that point "A" is at a different horizontal position to region "B".

The NIST "trace" consisted of a splice between data from point "A" (and possibly below) with region "B" and above.


It was impossible for the trace to be started above region "B" at T₀, as the Western rooftop features had not at that time descended...

[qimg]http://femr2.ucoz.com/_ph/7/393459964.png[/qimg]


Clear ?

Can I assume any remaining misunderstanding has been sorted ?