August 31, 2022
(rough, incomplete, flexible)
last update 9/2
For daring to actually challenge our work on the Ghouta chemical massacre, German open-source researcher Kostja Marschke is a relative giant. His efforts outstrip the combined total of all those useless hacks at Bellingcat (whom he emulates - he's not a member), and their weakling allies in and out of Western intelligence agencies. Almost all of these I know of have me block on Twitter, many after zero interaction, and the rest with just a short engagement marked by unsustainably dismissive attitudes on their part. Even as the work has attained some relevance, thanks to Aaron Maté, Saar Wilf and Rootclaim, William VanWagenen and others, the pros like Eliot Higgins have no kind words for our report(s), no specific words, and also as few unkind words as possible. Just a wall of quiet, mainly.
So off the bat kudos to Kostja for not getting that memo, or for having the balls or the cluelessness to even pretend he can debunk our work in its specifics. This allows for a rare challenge our findings always benefit from. So I'll try to give Kostja the benefit of the doubt, mainly, and let the record of flinging and sliding down the wall speak for itself.
See as needed Monitor on Massacre Marketing: 2021 Ghouta Sarin Attack Reports to explain the subject and its relevance, for the main report and other reference.
Previous Arguments Recap
I noted some of his first arguments in the post Ghouta Reports: First Debunk Efforts. He started with the usual snide, dismissive tone, employed in passing as he announced his refusal to say anything past some insults. The ballistic and video findings so stunned him into saying "I guess we have left the field of evidence now. This is where I leave, too ... There's no real point in engaging much beyond that. Using this video to try and defend Assad is just wishful thinking."
But for good measure, he did try the standard shortcut of declaring the sarin used was specific to the Syrian government's program, and must have been used by them, whoever fired it from wherever. He pointed to the hexamine link, which I challenged, pointing to my report, Then he complained instead "The more important issue, which is hydrogen fluoride, is not addressed. This is an important bias. This "analysis" was clearly done starting from a conclusion." Maybe I do have such a bias, picked up from J.P. Zanders listing hexamine under "sarin" w/no basis and repeatedly calling it the "principal telltale sign" of regime guilt, Dan Kaszeta calling it a smoking gun for a few random reasons, Eliot Higgins and the NYT et al. repeating that, Kostja Marschke himself leaping straight to it a few tweets back, with no mention of any other chemical clues ...
A complete answer on this point is still not for me, but I discuss the hexamine issue and the other impurities in the posts A Sarin Blame Shell Game and Whose Hexamine? Report sponsors Rootclaim may know more about that I, but all considered, as they put it: "We addressed sarin composition in our analysis. It indeed increases the probability that the SAA is responsible, but that's of course negligible compared to an actual video showing the opposition carrying [out] the attack right at the intersection of all rocket trajectories."
Put another way, it fails as a shortcut. Islamists DID it, ON VIDEO. That's much stronger than Assad musta diddit, somewhere, because we think it's his sarin. That failed and Kostja eventually decided that if our site match was wrong, there should be plenty of differences to discover, and some specifics were in order to counteract our dangerous findings. The sources were open, and he dove in, and came up with fistfuls.
The Moon is not Visible: In a Twitter thread on our errors, Kostja argued how even if this were our field, "The videos were demonstrably not shot the night of the attack| because "it was a full moon that night, with the moon being clearly visible when looking into southern directions (such as the south-east trajectory the rockets were fired into according to you). ... we know at which direction the camera had to point if your "analysis" was correct. We also know in which direction the moon was visible. But we don't see a full moon, nor any light it shines on the scene. That's because the video does not show the attack."
In the earlier post, I shared others' work on moon position - to the southwest - and considered views of the rockets fired - these are facing pretty much northeast - OPPOSITE of where the moon would be visible. But there is one point where the camera faces anywhere close to the southwest - at the D30 cannon in video 1, 0:30 forward - seeing 2 street lights (were those ever placed?) - probably not looking at quite the right angle or quite high enough to say, except maybe as the shell is fired and the scene is full of smoke.
As for the moon's ambient light, this probably existed, but it would be invisible on most 2013-era smart phone cameras, at least compared to the floodlight nearby. So this point failed, and it was down to challenging the site identification.
Yellow Vegetation: Soon after my "first debunk efforts" post, Kostja made a breakthrough regarding the vegetation aspect of our site match; even if every detail corresponded, this replica field must be somewhere else with a yellower version of this field's trees and grass. I thought I had at least started a blog post on or including this, but apparently not. Some review from scratch is past due.
Initial tweet, asking for Eliot Higgins' view: "In a still from the enhanced version of their video, the vegetation is yellow/brown-ish. There are small pieces of vegetation that are also green. Yet in their satellite images of the alleged location, the vegetation is green."
Next, he added "The satellite image is from 3 days after the attack, so the grass didn't turn green in the meantime." And he asked a good question: "Do you think it's the illumination that turns the grass yellow, or do you think the video was shot elsewhere?"
Higgins replied "I was kinda done debating this one 7 years ago, but that certainly looks yellow to me." He gave no answer to the good question, but he would not want the video filmed where we decided. Lazy fuck had to get back to some video game, I guess. Here's some partial record of where he was happy to leave that "debate."
So I'll answer Kostja's question. Indeed, the light seems key to what we see here - as it is anywhere. The foliage in the videos variously appears black, dark gray, orange and then dull magenta (as seen at right) during rocket ignition, and finally yellow. That's a good one to stop on, since it actually happens in nature. But even sampled later, the exhausts glow seems to have a clear magenta cast to it, I believe you would correct this by shifting light the opposite color - which is green.
But so far, my efforts to "correct" the grass to green require making everything greenish. And noting there is some green visible, it can't all be from a yellow or magenta cast. Some vegetation is yellower than others. But then notice that the green bits are closer up, farther from that rocket glow. The light here is dimmer, but a mix of exhaust glow, moonlight, and residual floodlight, which might yield a more balanced light than what floods over the grass and trees.
.png)
But also let's clarify what we'd be seeing, what span we're looking across in that brief moment. August 23 view on Google Earth, which we'll use heavily in this article - lines of sight from about where we estimate the camera was (may be a bit different), the main green field is under the ignition plume and beyond that, not visible. If we were on the main open platform, we'd look mostly north across bare earth or maybe yellow scrub, like the video shows. But if it's further west like this, some greener vegetation was closest, some big enough to cast shadows, and
then pat that we'd see some of that scrub and maybe some of the actual crop.
So accepting that the nearest plants should be green in real life, and noting the very nearest ones even APPEAR green, but seem very few ... I propose in the main view, the ground slopes down just past the platform, and pretty sharply at first, so only the tops of the shrubs at the very edge appear. The rest of that sizeable green stuff angles down out of view, flattens, yields to yellower grass that's soon visible again - some of it's quite tall, with heads illuminated in the magenta view above.
And I just noticed that by this line-up, there's some other thing past the platform - in red - roughly where there's a thing in the satellite view. See below regarding the size of this thing.
Recent Round of Efforts
When Aaron Maté bumped the subject on the Ghouta attack's anniversary, to rep. Ro Khanna, there were some who stepped in, Kostja being one - report co-author Michael Kobs engaged him and then he started me off with some general questions for Michael and I. Soon he had "some thoughts on the matter. Hope to write them down soon." Instead he presented them maybe prematurely, with others coming as they occurred to him in rapid succession. I won't be linking most of the tweets, but I think I have the right gist anyway.
He tried for a better form this go-around. Aug 26 "Look, I clearly take your geolocation effort seriously," and he wouldn't call all of it laughable. But as soon as it seemed possible, he was back to saying things like " I think it's now quite clear that your geolocation project is debunked in its entirety." (Aug. 27). What mighty works allow total refutation of visual research of some actual quality? I've summarized his specific claims in this overview image - some points here aren't his arguments so much as implications of them, added for smart-ass effect. But the last 6 are his several new or standing lines of argument as I follow, which I'll briefly explain and rebut below. Note 9/2: I mixed up the colors on the last 2: object is dark red, canopy in pink, continuing off the platform.
First, his initial replies to an earlier version of this graphic, select stray points:
Red: he agreed with the implication. Green: "I think there's white object throwing a shadow," rather than a "burned area" as I said or, put better, a darker, likely burned or smoke-stained area covering some 1/3 of the field. I asked him what that meant, but he didn't answer. I'll disregard that as something confused. Yellow was still "Important point. Will elaborate further." I didn't notice anything to change the analysis above.
"White: I haven't subscribed to your trajectories." Of course not. Kostja adds his own zero to the total of zero serious refutation of this work the naysayers can provide. None of them can show it wrong because it's right, but for some reason none of them can agree with this best analysis. That's the part I know better, but what Kostja wanted to challenge happens to an area I'm less conversant with; the video - satellite match setting the apparent launch spot.
As I explained to Kostja, most of the work analyzing and modeling the launch spot was done in a long, vigorous process (by direct messages on Twitter) that I was not involved in, as I've said here due to a "social media snafu" that somehow blocked my invitation from arriving, I thought the work hadn't begun, while Michael Kobs and the others must have thought I was just flaking out on it - so I missed most of that work as I did my own report. I never even saw the main discussion, just the parts copied into a new PM thread Michael started for my benefit. Even now I'm not informed enough to speak for everyone in my responses to the other points - new and specific - covered one-by-one below.
Wall vs. Rise
Marschke's leading point so far is how our matching of fields is ruined by the presence of a wall in the satellite view where we have in the video a "ditch" next to a raised level to the east. Responding to my overview graphic: "Dark Blue: It's very clearly a wall. It casts a shadow with a [width] of almost 1 meter from space & can be clearly seen as a wall in 2012 images"
In the attack videos, a D30 cannon was situated at the northeast corner of the field , which in one video is approached from the east, through some mid-sized trees and over a line ... distinct, sharp-edged, possibly even concrete, like the lip of a wall, but at ground level - his headlamp light falls across, casting a shadow before illuminating ground further west, making a linear shadow. It doesn't seem wide, but it's foreshortened here. The sun would do the same, whenever it shone from the east. That could be because this ground level is higher than the other.
A floodlight on the truck in the other field provides the main light. Neither light falls in the ditch, the darkest strip here. Therefore: raised level and a ditch are in the video.
Most clearly, from the opposite view looking back at the trees - see any view of the rocket launch above - they have some centimeters of dirt beneath them, on a raised ridge (or "wall") as I first thought, or on a continuous raised level, as we now agree.
But Kostja thinks there's a shadow-casting wall there in the satellite version. He centers this on a joint impression of 2 walls meeting at a right angle, and then tends to conflate the two a lot. Here I trace one in red and one in pink, Red: he and I are agreed on a wall or narrow earthen berm here with 2-3 shrubs planted along its crest (see Feb. 2012) - it and its shrubs variously cast clear shadows to the north, - but the pink "wall" running south is what matters, and it's debatable at best.
Like the north mound, the west edge has a brighter-lit apparent crest at the northern start, suggesting it's elevated. All traced in red above right does look like 2 walls. But this seems to fade away quickly for the western wall, as if the mound slopes down to a flatter profile. It would be this flatter part seen in the video - the edge might rise to the north or right, if it were like in the satellite view. And I think it might. Here - about 0:20 in
video 1 - the weeds may just be taller, or the ground might rise a bit towards that mighty "wall" - towering atop a RAISED LEVEL, next to a trench or ditch.
Comparing what we see (left) and what Kostja seen (right) - Seen from space and with sun from the right (east), the shadows of these two scenarios should look about the same, especially if the wall were a bit lower. He specified at one point "I don't think it's a high wall, but if it's high enough to throw a shadow on a satellite image," But see, according to us, the rise would cast about the same shadow. So ... what?
The test would be how a wall would also cast a shadow in afternoon sun from the west, while our raised level would not. Unfortunately, between all views in Google Earth, most views have the sun from the southeast or once from straight above, making for shadows running mostly along the possible wall, and a bit to the west... only one view has sun barely from the west and might allow for this test. Kostja located this view, from December, 2017, enhanced view below. "Here, in this image from December 2017, the sunlight comes more from the west - and the shadow is thrown to the east. This is a wall. A ditch doesn't throw shadows." But the "wall" that matters (app. at the pink line here) is long gone by then, although that northern ridge we agree on remains (red).
KM: "this image posted by Adam really puts the final nail into the coffin of your "elevated area/old foundation"-theories. The sunlight is from the south-west in that view, yet there is still a shadow." East of the pink line? No. This is years after the area was all torn up and rearranged. The ridge marked red remains, but now nothing in particular corresponds with the pink line, and THAT is the line we have
video of an Islamist cameraman stepping across.It's not the red line, or the blue area marked above either. That came up because of an error, I now think, in our main report (PDF) p. 41. We indeed identified that line as being the ditch under discussion. All agree it's a trench (crude topographical cross section in green FWIW). It's not the same as the ditch in question, being in a different spot and not on the same line as it, further from the trees, or about where there had been a "road." - plus it was only dug sometime between the images of August 2013 and January, 2014, where it has a markedly vaginal character - in this 2017 view and in 24 images between it looks much the same - starting at that pin, it's not at the spot we're considering, again marked with a pink line. The only trace of that might remain: what looks like a residual span of ... some kind of ditch.
Kostja, Aug 26 "I don't think mistaking a wall for a ditch on satellite photos is "laughable". I don't think that getting the spot of that wall wrong by 5 meters is "laughable", either. Yet it's still crucial because it implies a mismatch." Indeed, the match shown on page 41 is a "mis" kind. But there's also the
separate spot where he sees a "wall" and we see a raised level next to a ditch, some meters north of this brain fart. By the satellite views, just by visuals, seeing a wall or a rise makes enough sense, and I'm not sure either can be ruled out. But to see something so linear that could go either way - right where the video has that instead of regular ground - all in all,
that's a match, not a mismatch.
No Paved Lot
Our analysis includes a paved platform just south of the field, or the main, uncovered part of it - smooth gray, no detail, with a straight, sharp edge that casts a mild shadow in spots, and may have some bushes growing just at its edge, but nowhere on it - here as seen on satellite 2 days after the attack vs. in the video taken, we think, of the attack itself.
Note the exact span of pavement is probably different in each image, with the video taken further west in the "canopy" area discussed separately below. For now we're discussing this main, rectangular area.
Responding to my overview graphic, Kostja said: "Orange: Plants don't grow on concrete. Plants grew there." In another tweet: "If you go on Gogle Earth Pro and check some historical images of the area (specifically 2012 and 2014), vegetation in that area consistently happens. So either it was [paved] only in 2013, or it wasn't paved at all." Later, to his credit, he would concede ..."I agree that it looks paved in August 2013, but that might just be an optical illusion. It's really hard to tell from satellite images." And for 3 images in a row it has this appearance - Feb. 2013, Aug. 2013, and Jan. 2014, for a span of 11-16 months. But before and after, he says, it was earth with plants on it, so Occam's razor therefore requires it was probably the same all along.
As I first argued it ... 2012 and before, whatever, but in January 2013 the same area LOOKS concave as if excavated and maybe rimmed with a timber frame in advance of concrete pouring. With a basement? Noting the dig looks rather deep. Hm. In February, it looks filled or covered with concrete, and stays that way to at least January, 2014. After this, by June, it's clearly NOT concrete, with plants growing all over.
But in fact, the vegetation before is pretty debatable, and the smooth lot or platform existed in a very similar form for years, from first usable view in Google Earth in June 2009, and in the 4 sample images from 2012 compiled below. It's worth considering that earth packed like concrete could look the same from space, and maybe fit the video almost as well, except for the audio of heels on pavement (video 2, 0:45).
The two dark spots along the edge have never been clarified - we've taken them for bushes, but they might be something more solid and permanent - perhaps a permanent absence: two drainage pits? Whatever these are, the suggested green seems to center around these, especially the one on the right. From space, it's hard to say if this is some vegetation that comes and goes or something else like a puddle that forms after heavy rain, etc. In May, it looks suddenly overgrown. The same happened in May, 2011. It's a seasonal thing. This is an issue with little to no no sign in the 2013 pavement, even as these dots remain. But then there's no view for May.
.png)
All this might suggest a dirt or gravel lot in 2012 and before, or a concrete lot with problems that might explain why it was briefly re-paved. Why that was again undone in 2014 is a remaining mystery. It seems possible to me that clues of the chemical attack remained, and this all represents some kind of cleanup. Otherwise, any of a few more mundane reasons could account for the change. As everything else stands, this question is no basis to rule a mismatch.
No Tree 2
Next, we turn to the trees east of the "wall" or on the raised level east of the launch field, Marschke disputed two of these trees, then just what we called tree 2, the northernmost one. Responding to my overview graphic: Purple: "Trees" don't disappear in winter & "Trees" have shadows. This doesn't."
Actually, deciduous trees can seem to disappear in the winter, as viewed from outer space. With the leaves all gone, its branches can easily be invisible, and the trunk might vanish too, or might appear as a faint dot, etc. Slender shadows not appearing is even less strange.
Here I circled tree 2 in yellow, in February (top) and August, (below), showing how it might cast a shadow (white line) despite being invisible (green), maybe due to its light-colored bark.
In Google Earth views, every winter it vanished and then comes back in the spring. He proposed instead this was an oval-shaped plot of crops planted every spring, or - since crops are usually grown in rectangular plots - some low-level deciduous shrub. If it were tall enough, you'd call that a tree, but he rules out tree height because it casts no shadow across the "wall" just north of it. There is some shadow right up TO that mound, and maybe a little up its side, seen here foreshortened, and maybe even at its crest, but washed out due to the higher reflectivity here. But at the crest of the "wall" there's no sign.
But where there's no wall to cut off the shadow, it runs wider, on scale with other apparent trees, as shown in the 2 views below. Top view: sun from the usual southeast, wall cut-off (red) - reflective and raised sidewalk cuts the shadow of the building's roofline (blue), as I suggest that wall does for the tree's shadow - bottom view: sun from straight above, seen from the south - the wall doesn't matter here - fully tree-sized shadow beneath tree 2 - here called tree 1, oops - which is clearly a tree, not some oval of tall crops. It's one of those our Islamist cameraman walked past as he came to film the launching of the Ghouta chemical attack.
Missing Canopy
What we've been calling a "canopy" over the western part of the concrete platform - it seems to narrow and turn a grayer color, while more beige concrete is seen in the video - I don't think the point was agreed - not explained in the report that I saw, but labeled on page 40, and included in several of Michael Kobs' 3D models, like on pages 45 and 48. This was decided on just from satellite views, with questions I'm still sick of hashing over - what casts a shadow on what, what's a building, a canopy, a platform, a ramp, what's shadow or vegetation, or plants on a downslope or what ...
I tended to agree some cover is how the platform continues here, but looks different from space ... But the more I compare views today, the more I doubt this and suspect it's always been pavement, just a different color and shape, and the video's beige color is - surprise! - a bit misleading. So far, I still can't explain everything with one 3D answer, but I'm all right with that. So to start, I'm not wedded to our prior conclusions on this point.
Kostja thinks the video proves there is no such canopy, and I would be fine with that. But even here, the evidence is inconclusive. He assumes its corners must have vertical supports and these would appear in the video, but they don't appear so they weren't there. But as one of Michael's 3D models of the view shows, the video view at launch is approximately in the red box here, and excludes this canopy and any supports. The view tracking the rocket up also seems set to miss it (to say for sure is very tricky). Furthermore, as I show in yellow, there might have been horizontal braces instead or, depending on the material, other ways of keeping a canopy up. In fact Michael modeled it with horizontal braces.
Beneath that model view is another view Kostja noticed, from the actual videos, a different moment, where the truck cab is visible. According to the above model with its green lines, one vertical support should be visible here, but up to the frame's edge (red), it's not visible. This can have 4 explanations: the model is a bit off and the support is just off frame - there's no vertical support, but some other kind - there's no canopy, but our site match is correct anyway - there's no canopy and it means we got the whole site match wrong. Kostja takes option 4, and I'll take any other. As noted, I'm leaning to 3 right now, but that was different yesterday and might change again.
Rectangular Object Too Small
In the last image above, I traced a red box and possible platform edge to go under the canopy. Now we turn to that rectangular object. He can see it lining up with a wheel of the loading truck - or perhaps not the right one. "That object doesn't even have 20% of the wheels height. You think that is visible from space?" He might have meant the lower of 2 black rectangles shown here.
The other rectangle - if it starts at ground level - appears at least as big as the wheel, but it looks bigger in comparison because it's closer. We had noticed this larger one, and taken it as the boxy object along the platform's edge, helping confirm the location and set the line of sight. The smaller one may be the edge of a raised part of the platform, as I sketch it below in 2 possible spots, or maybe it's part of the object, as noted above, possibly glimpsed at rocket launch. See how its shadow seems to extend at the west end, nearly filling the space between blue lines. That's what the other rectangle does.
FWIW this object may have been movable and moved. It - or something about the same size, but perhaps whiter - first appeared in mid-2012 a bit out in the field (see above under "no paved lot") It's moved to this spot by January, and stays there in February and August. There's no sign it was moved in between views, but it could have been, and that might explain any difference in the view (if that comes up).
What is it? I don't get out much, but maybe an electric generator? to run whatever, like floodlights, or a pump to manage the seasonal flooding?
Review
Vegetation color: likely down to just the nature of the light
- Wall vs. rise: not testable, but that unclear linear feature appearing just where the video has one is a general match
- Unpaved lot: pre-2013 vegetation is unclear, maybe consistent with prior paving, and either way, there are signs of new concrete work in 2013 to go with the following appearance of concrete, regardless of how it looked in mid-2014 and later
- Tree 2: plain stupid. Tree shape and color, casting a tree-height shadow, seeming to "vanish" in the winter all says deciduous tree.
- Missing canopy: not terribly relevant, but some good points raised, and he might even be right - on the absence of a canopy, not on that mattering.
- Object too small: evidently wrong, and likely just momentary confusion he already abandoned of his own accord. The presence of such an object, on the other hand, is one of the many strong points in our matching of this field to the one in that video of the Ghouta false-flag chemical attack.
There were some other passing quibbles and maybe a few serious arguments I missed in all the noise. But these six alleged mismatches have all failed. A review of them has helped identify one relevant error, and to question another finding of middling relevance. Along the way, I - with Kostja's help - might have improved the site match in a few ways (depending how my new thoughts stand up to scrutiny). This might be why the true professionals refuse to engage our work seriously, and leave Marschke a relative giant among the turd-hurtlers. Nothing kills it, every challenge just makes it stronger, and any informed observer should know why that is. The "expert" response will continue to be ignore these findings, claiming they don't matter, it was a long time ago, etc.
b.png)
.png)
c.png)
.png)
.png)
b.png)
.png)
b.png)
.png)
.png)
.png)
.png)
.png)
b.png)
.png)
b.png)
b.png)
.png)
b.png)
.png)
b.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
