Hi Sander. Thanks for the thoughts - I'll take them serially paragraph or sub topic at a time
...and let's all remember that I am using my preferred approach to this type of problem challenge which is;
1) Work from macro level evidence;
2) Based on what is observable from the record plus what is known of the structural aspects plus what can be reasoned; WHILST
3) Initially using 'ball park' guesstimates for as long as valid progress conclusions can be reached THEN
4) calling on more detailed evidence - mechanisms, numbers, structural analysis. whatever.
..and I am thereby (hopefully - tell me if I miss something

) avoiding the errors of losing track of context or applying detailed maths over wrong premises.
OK with that reminder lets look at this:
(I've had to cheat to put your picture there - attachments don't link in quotes.)
A good illustration - I would not hesitate to use it in explanation with two provisos:
1) The most important is that it shows 'integral blocks' and that is definitely what did not happen in at least two important ways:
(i) break up of both top and bottom 'blocks' started in the range of stages covered by the drawings; AND
(ii) the interface between top and bottom was complicated multi structural element interactions not 'integral rectangular block on ditto'.
I know you and I don't have that in mind so it does not detract from our debate.
If I was explaining 'initiation/transition to ROOSD' it would either be to a professional person with little 9/11 WTC knowledge who would understand the realities OR it would be to a less qualified lay person to whom I could explain the actual details as the explanation progressed. So a valuable drawing and useful. (Are there any royalty charges?

)
The second proviso is a positive one - the drawing shows the movement of the 'virtual fulcrum' - a point worth emphasising.
SanderO wrote:...In my little thought experiment I used FOS 2.. that meant that half the columns could support the actual load.. so half the columns would carry the load (and assumed strong enough bracing to support the cantilever....
It doesn't work out that simply. Cutting half the columns does not mean that the remaining columns carry double load. It depends on the physical layout of the remaining columns.
As first stage of explaining that take this simplified example:

So we have a simplified WTC'x' Twin Tower. Top 'block', total mass 400 whatevers and standing on three columns A, B and C. Scenario 1A we see the columns loaded A=100, B=200 and C=100. Reasonable for a typical building.
Now cut column 'C' - does the loss of 100 support spread evenly across the other two - No!. Does it spread proportionally across the other two i.e. 33 to A and 67 to B. No again.
What happens is either Case 2A OR Case 2B depending on whether the 'top block' is a rigid block (unlikely) or a flexible structure - the realistic option.
So point Number one - the loss of support at Column 'C' would see the load on 'A' reduce to zero; all the load carried by 'B' and the load on 'B' therefore doubling.
Do that again with Case 1B - uniform distributed loads of 133 to each of three columns. Same result - zero on Column 'A' and 400 on Column 'B' and, if the 'FOS' was 2 that puts Col 'B' into overload and probable failure. So removing one third of columns causes failure with FOS of 2.
Let's now look briefly at Case 2B - flexible top structure.
Initially it looks like the same result - Zero on column 'A', 400 on 'B' and zero on 'C'. BUT the flexible structure will have sagged downwards either side of the central 'B' column. The left side sagging over Col 'A' will try to put load on 'A' but overall cannot because nothing on 'C' to balance. So a bit of dynamic balancing will see left side and right side sagging BUT left side still in contact with Col 'A' net result double sag at 'C' and zero sag at 'A'. And that tilts the top with the consequence that the Centre of Mass moves outside of Col 'B' - the 'top block' would topple if it was flexible - and not held back by Column 'A' where the load goes negative - Col 'A' is in tension..
Both cases obviously simplified to illustrate the point. However it is analogous to the model of 'WTC2 Tilted Top' I have been discussing where:
-- Column 'B' approximates 'core';
--Column 'C' approximates 'low side perimeter'; AND
-- Column 'A' approximates 'high side' perimeter.
I have started an eleven columns model to take this explanation further - so I can support my response to your 10 column example. I said it would fail at 7 remaining if all the columns were in one row and removal from one end takes out three at that end.
Alternatively if you had an 11 long row of columns under a two dimensional building and took out 5 columns (2, 4, 6, 8, 10) leaving uniform doubled spacings the redistribution say 440 total mass distributed 40 per column, remove 5 leaves 440 distributed over 6
uniformly spaced columns or a load of 73.3 on each of the remaining columns which were designed for 40. So, with FOS 2, you survive - like this:

Therefore, as I said way back, "
It depends on the physical layout of the remaining columns."
I can complete the '11 columns model' if you wish, And I chose 11 over 10 because removing half of 10 ends up non symmetrical.
I'll post this much and come back with another post addressing your other paragraphs.