MOL discussion

[Blackstar]

https://twitter.com/NatReconOfc/status/1625279930790445059

[Blackstar]

I looked at Lt. General Lew Allen's notes leading into this again:

Robert Perry “Recce Satellite R&D: Capabilities in Readout, Crisis Reconnaissance and Very High Resolution"

Very interesting. I had forgotten some of the things he wrote there. One of the things he pointed out was that quite often in satellite reconnaissance, it was "the technological imperative" that led development of new systems, not specific requirements.

For those who never studied the history of technology or science and technology policy (I did that many years ago in grad school), there is a theory about technology developing either as the result of "technology push" or "demand pull." You can substitute "requirements" for "demand" here. Simply put, Allen (citing Merton Davies' RAND buddy Amrom Katz) suggests that systems were developed by engineers pushing the technology, rather than as a response to a clear requirement.

Now one could argue with that. There were requirements:

-CORONA was required because large swaths of the Soviet Union were inaccessible to the U-2, and the U-2 was a "melting asset" anyway that would soon disappear.
-GAMBIT was required because once the U-2 was vulnerable to missiles, they needed a system with U-2 resolution.
-HEXAGON was the result of requirements established by a study (was it the Drell Panel or the Purcell Panel? I forget.).

But I think that Allen is talking a bit more generally here, and about other systems, mainly the MOL and the VHR (or "Hexador") and near-real-time. In particular, when he starts talking about the decisions that led to the KENNEN, Allen says that there was no real requirement for it. The requirement was for strategic systems (counting tanks, missiles, planes, and subs), and not for timeliness. He makes the fascinating argument that once the technology fell into place, Din Land and the CIA pushed for that system with a vengeance, and that there was not actually much of a requirement for it.

That argument puts the other programs like FROG and the systems that Joe Page II has been writing about (CORONA 6-Pack, PINTO, etc.) into a new light. What General Allen basically says is that because they were not really required, they didn't have a lot of support. He even says that he was wrong to push for FROG, because it was not that good. He suggests that these were only proposed because people thought that they should advance an alternative to the expensive KENNEN system, not because they thought that the intelligence community really needed a near-real-time system.

(more in later posts)

[Blackstar]

There's also some great subtlety in what Allen wrote. He suggests that there was not really a requirement for very high resolution (VHR) either. He says that they sort of backed into that: they approved MOL because they were canceling Dyna-Soar and there was a desire to have some new manned system to continue on from Dyna-Soar. Then they didn't know what to do with MOL, so they stuck a powerful camera on it. Then the Land Panel declared that this should also include an unmanned option. And then they ended up with a really expensive MOL program that was not ideal for anything.

There are a bunch of important points in all that. One is that Allen is saying that VHR was not required, and that what they should have pursued was an upgrade of GAMBIT, which is what they eventually did all along. That raises an interesting question about whether if MOL did not exist, would they have sought out faster improvements to GAMBIT instead (in the 1966-1970 period).

He also provides some dollar figures. VHR was "not worth" its $500 million cost. Manned MOL would have cost $1 billion as a lab, $2 billion as MOL/DORIAN, and $3 billion for the version that included both astronauts and the robotic option. (Interestingly, that final figure was deleted in the 2018 version but revealed in the 2022 version.)

Allen suggests that if it was not for MOL, then $500 million for VHR might have seemed reasonable. But apparently by the time VHR was rejected, people had been tired of spending so much money on MOL.

[Blackstar]

So...

Continuing my train of thought here, and my earlier posts about VHR and whether there was ever a study done about the need for VHR, I still think that there could have been a study done on VHR. It seems like it just makes sense: if you are studying a very high resolution reconnaissance satellite in terms of technology, orbits, etc., then an obvious question to ask is what it can do. And the related question to that is what is the value of doing that?

Allen may be right that there was no requirement/demand for VHR. But that does not mean that nobody studied what could be done with that kind of quality imagery.

[LittleBird]

I looked at Lt. General Lew Allen's notes leading into this again:

Robert Perry “Recce Satellite R&D: Capabilities in Readout, Crisis Reconnaissance and Very High Resolution"

Very interesting. 

It is, and bears many rereadings. Leaves me wanting to read more of his writing in fact.

One thing I wondered is when it was written ? As far as I can tell it is after the final EOI decision but well before the debut of the KH-11. In this respect it's interesting to that he even compares the KH-11 to the SST (cancelled in spring 1971) at one point, as something that may also not really be needed. 

What are also interesting to me is i) the way he talks about manned spaceflight:

many felt it was essential that DOD have some kind of man in space effort if the national commitment to Apollo was really going to generate a new era of manned spaceflight.

which is  particularly intriguing comment if made around the time of the Shuttle decisions, and ii) the bald statement that

the DORIAN camera specifications were actually optimised to make maximum use of man's contribution and  were substantially off-optimum for an unmanned application (e.g. focal length too long).

Is it known in detail what is he getting at with the latter comment ?

[hoku]

So...

Continuing my train of thought here, and my earlier posts about VHR and whether there was ever a study done about the need for VHR, I still think that there could have been a study done on VHR. It seems like it just makes sense: if you are studying a very high resolution reconnaissance satellite in terms of technology, orbits, etc., then an obvious question to ask is what it can do. And the related question to that is what is the value of doing that?

Allen may be right that there was no requirement/demand for VHR. But that does not mean that nobody studied what could be done with that kind of quality imagery.

The attached pages from a 1969 proposal for a study on "(...) the effect of photographic ground resolution on photointerpretation" briefly address the difficulty in quantifying the "dollar value" of intelligence.

The biggest challenge might have been to assign a strategic (rather than pure tactical) value to VHR.

[Blackstar]

and ii) the bald statement that

the DORIAN camera specifications were actually optimised to make maximum use of man's contribution and  were substantially off-optimum for an unmanned application (e.g. focal length too long).

Is it known in detail what is he getting at with the latter comment ?

It's not known, but I can guess.

This goes back to the crux of the issue of VHR--why do you need it? (Specifically, why did anybody in the late 1960s think that they needed it at that time? We can posit that maybe decades later there emerged reasons why it was needed.)

My guess is that what he is saying is that when they decided on that powerful camera (long focal length), it resulted in such a small viewing area and precise pointing requirements that only a human could effectively point the camera. We can guess that there were probably several reasons for that, including the fact that the precision of the spacecraft orbit and the precision of the geolocation of targets on the ground were not sufficient for any of that to be done automatically. So you really needed a human to do all of that and the system was optimized to have humans running the system in real time--telling the computer exactly what to do in real time. An automated system could not do it that precisely. And Gen. Allen may have been implying that the cost of trying to get the automated system to be capable of that was very expensive.

I think we have to think out the automated/unmanned MOL carefully. It was not simply a bunch of return vehicles and then a system for turning the camera on and off and pointing it at pre-set targets (like GAMBIT). It probably entailed a lot more sophistication in terms of high-precision determination of both the satellite's orbit and attitude, and information about the precise location of targets on the Earth. That probably required more computer processing power and other stuff like measuring the shape of the Earth and its gravity field and putting that all into the computer models that directed the camera operations. Those details are not revealed in the documents that have been released.

[Blackstar]

The attached pages from a 1969 proposal for a study on "(...) the effect of photographic ground resolution on photointerpretation" briefly address the difficulty in quantifying the "dollar value" of intelligence.

The biggest challenge might have been to assign a strategic (rather than pure tactical) value to VHR.

Another good find. It also strikes me that this is the kind of thing they should have revisited periodically, because technology would have changed on the ground processing side, affecting what the satellites could deliver. For instance, 1-foot ground resolution in 1970 might have been necessary, but maybe a new system on the ground that provided image enhancement might have made it possible to get the same results with 2-foot resolution. There was certainly a major push for developing better processing systems that could do things like change analysis. I suspect that much of that was only incrementally useful until recent decades when massive computer power made it possible. Today it would not surprise me if a lot of the initial analysis is done by a computer, so an imaging satellite takes an image and a computer on the ground compares it to previous images of the same location and determines if there has been any real change. If not, the image is classified as lower priority.

Certainly the advent of near-real-time imaging with the KENNEN probably forced a lot of reexamination of imagery interpretation. It was no longer simply resolution, it was how quickly you got it. Was 2-foot resolution "better" than 1-foot resolution if you got it in 30 minutes compared to 2 days?

One person who did know some of this stuff told me that GAMBIT and HEXAGON were still producing unique and better quality imagery into the 1980s. KENNEN was fast, but it wasn't as good, at least for certain applications.

[edzieba]

the DORIAN camera specifications were actually optimised to make maximum use of man's contribution and  were substantially off-optimum for an unmanned application (e.g. focal length too long).

Is it known in detail what is he getting at with the latter comment ?

Probably down to image circle: the human retina has cells packed only so close together in the fovea, so that puts an absolute maximum limit on effective lines/mm at the focal plane, as well as an absolute limit on image plane size (due to eyeball motion, this is more defined by the eye entrance pupil/optics exit pupil interaction, but for all intents and purposes has a maximum limit). If you want a higher ground resolution then you must reduce angle of view (increase focal length) to achieve it.
With film on the other hand, as long as you are capable of manufacturing the optics of sufficient size at sufficient optical quality (and do not run into limitations like the f/0.5 limit or optical element intersection) then you can fix focal plane lines/mm and magnification (and thus image plane resolution, which ideally would be the 6cm seeing limit) at your desired fidelity, and increase the image plane size (film strip width, or film square size) to achieve a wider coverage area at the same fidelity.
On top of that, film grain density can be higher than rod & cone density in the foveal region of the retina, again meaning that a shorter focal length for film-based camera will result in the same angular resolution as a human-viewed telescope, with the potential for a larger imaged area.

[LittleBird]

Meanwhile, I was interested to see four inches resolution quoted for DORIAN in the long and very detailed mapping camera history that I know was talked about here last year: https://www.nro.gov/Portals/65/documents/foia/declass/ForAll/061422/F-2022-00041_C05099307.pdf  (page i-9, see grab below)

Was that figure public anywhere else ?

[Blackstar]

Meanwhile, I was interested to see four inches resolution quoted for DORIAN in the long and very detailed mapping camera history that I know was talked about here last year: https://www.nro.gov/Portals/65/documents/foia/declass/ForAll/061422/F-2022-00041_C05099307.pdf  (page i-9, see grab below)

Was that figure public anywhere else ?

Ooh, that's a great find. I've told the story about how I found out about the resolution goal back in the 1990s. A guy told me a story...

I may have that interview on cassette tape. I have a bunch of cassette interviews that I really should get converted to MP3 files and then transcribed.

[LittleBird]

Meanwhile, I was interested to see four inches resolution quoted for DORIAN in the long and very detailed mapping camera history that I know was talked about here last year: https://www.nro.gov/Portals/65/documents/foia/declass/ForAll/061422/F-2022-00041_C05099307.pdf  (page i-9, see grab below)

Was that figure public anywhere else ?

Ooh, that's a great find. I've told the story about how I found out about the resolution goal back in the 1990s. A guy told me a story...

This story ?

According to a recently declassified memo concerning the visit [to NPIC] written by Major General James T. Stewart, Vice Director of the MOL program, [Hubert] Humphrey arrived about 45 minutes late for the visit. The initial briefing, held in a secure room at NPIC, involved Stewart telling Humphrey about the increasingly expensive and controversial program. According to a former NPIC photo-interpreter who was present for the briefing, at one point during the meeting Dick Helms, who had not spoken at all, wrote a note on a piece of paper, folded it, and slipped it in front of Humphrey, who glanced at it but did not say anything. After the briefing was over and the senior officials left the room, the photo-interpreter lingered behind and grabbed the piece of paper. On it, Helms had written “Why four inches?”

From your piece at https://www.thespacereview.com/article/3490/1

 

[hoku]

The attached pages from a 1969 proposal for a study on "(...) the effect of photographic ground resolution on photointerpretation" briefly address the difficulty in quantifying the "dollar value" of intelligence.

The biggest challenge might have been to assign a strategic (rather than pure tactical) value to VHR.

Another good find. It also strikes me that this is the kind of thing they should have revisited periodically, because technology would have changed on the ground processing side, affecting what the satellites could deliver. For instance, 1-foot ground resolution in 1970 might have been necessary, but maybe a new system on the ground that provided image enhancement might have made it possible to get the same results with 2-foot resolution. There was certainly a major push for developing better processing systems that could do things like change analysis. I suspect that much of that was only incrementally useful until recent decades when massive computer power made it possible. Today it would not surprise me if a lot of the initial analysis is done by a computer, so an imaging satellite takes an image and a computer on the ground compares it to previous images of the same location and determines if there has been any real change. If not, the image is classified as lower priority.
<snip>

The "Final report" of the study, which had Autometric/Raytheon as one of the sponsors, and which is dated July 1970, describes that they obtained aerial overflight images of Aberdeen Proving Ground and artillery museum, and of Ft. Meade at 1.5" resolution. The negatives were processed by Eastman Kodak to create sets of images at a range of degraded resolutions, and then handed to "experienced photo-interpreters".  The outcome of the PI analysis, though, seems to be subject of yet another report - security compartmentalization in action 

https://www.cia.gov/readingroom/docs/CIA-RDP79B00873A001600040034-7.pdf

[hoku]

Attached are the (initial?) overflight request at 500' altitude for Aberdeen, with the NPIC reference NPIC/7SSG/RED-1774-69, and a criticism of the study on PI performance, with reference to NPIC Memorandum NPIC/TSG/RED-45/71

[Blackstar]

This story ?

According to a recently declassified memo concerning the visit [to NPIC] written by Major General James T. Stewart, Vice Director of the MOL program, [Hubert] Humphrey arrived about 45 minutes late for the visit. The initial briefing, held in a secure room at NPIC, involved Stewart telling Humphrey about the increasingly expensive and controversial program. According to a former NPIC photo-interpreter who was present for the briefing, at one point during the meeting Dick Helms, who had not spoken at all, wrote a note on a piece of paper, folded it, and slipped it in front of Humphrey, who glanced at it but did not say anything. After the briefing was over and the senior officials left the room, the photo-interpreter lingered behind and grabbed the piece of paper. On it, Helms had written “Why four inches?”

From your piece at https://www.thespacereview.com/article/3490/1

Yeah. I'm trying to remember the name of the guy who told me the story. But I loved the fact that he told me that around 1997 or so and then two decades later I discovered a document that confirmed the basics of his story.

[Blackstar]

The "Final report" of the study, which had Autometric/Raytheon as one of the sponsors, and which is dated July 1970, describes that they obtained aerial overflight images of Aberdeen Proving Ground and artillery museum, and of Ft. Meade at 1.5" resolution. The negatives were processed by Eastman Kodak to create sets of images at a range of degraded resolutions, and then handed to "experienced photo-interpreters".  The outcome of the PI analysis, though, seems to be subject of yet another report - security compartmentalization in action 

So that's interesting, but raises some questions. First, an aside: the 1963 study did essentially the same thing. (Was that the Drell Panel? It was either the Drell Panel or the Purcell Panel. I'm too lazy to look that up at the moment.) That's what led them to the conclusion that they needed a system with CORONA area coverage and GAMBIT resolution, and I think at that time GAMBIT resolution was 2-3 feet.

But why in 1970 did they start with 1.5 feet? Why not start with something higher, like 6 inches, and then work down from there? After all, they were already close to 1.5 feet, and they were discussing systems that could do better than that. This seems like they framed the question of "Assume you have the capability of the system planned for 1973. What can you do with it?"

In other words, the parameters of the study seem like they were trying to justify the planned capabilities of GAMBIT, rather than asking if something better was required.

[edzieba]

The "Final report" of the study, which had Autometric/Raytheon as one of the sponsors, and which is dated July 1970, describes that they obtained aerial overflight images of Aberdeen Proving Ground and artillery museum, and of Ft. Meade at 1.5" resolution. The negatives were processed by Eastman Kodak to create sets of images at a range of degraded resolutions, and then handed to "experienced photo-interpreters".  The outcome of the PI analysis, though, seems to be subject of yet another report - security compartmentalization in action 

So that's interesting, but raises some questions. First, an aside: the 1963 study did essentially the same thing. (Was that the Drell Panel? It was either the Drell Panel or the Purcell Panel. I'm too lazy to look that up at the moment.) That's what led them to the conclusion that they needed a system with CORONA area coverage and GAMBIT resolution, and I think at that time GAMBIT resolution was 2-3 feet.

But why in 1970 did they start with 1.5 feet? Why not start with something higher, like 6 inches, and then work down from there? After all, they were already close to 1.5 feet, and they were discussing systems that could do better than that. This seems like they framed the question of "Assume you have the capability of the system planned for 1973. What can you do with it?"

In other words, the parameters of the study seem like they were trying to justify the planned capabilities of GAMBIT, rather than asking if something better was required.

It could have been meant to study what level of degradation could be applied to existing imagery to obscure full capability, whilst it still remaining useful. That was only a year before the debate on how to more widely utilise satellite imagery (e.g. the "bring more people into TALENT-KEYHOLE"/"move some images out of TALENT-KEYHOLE"/"declassify images within TALENT-KEYHOLE" question studied by COMIREX https://www.nro.gov/Portals/65/documents/foia/declass/Archive/NARP/1971%20NARPs/SC-2021-00001_C05134214.pdf).

[hoku]

The "Final report" of the study, which had Autometric/Raytheon as one of the sponsors, and which is dated July 1970, describes that they obtained aerial overflight images of Aberdeen Proving Ground and artillery museum, and of Ft. Meade at 1.5" resolution.
<snip>

<snip>
But why in 1970 did they start with 1.5 feet?
<snip>

Starting ground resolution was 1.5 inch (''), not feet (')     

[LittleBird]

The "Final report" of the study, which had Autometric/Raytheon as one of the sponsors, and which is dated July 1970, describes that they obtained aerial overflight images of Aberdeen Proving Ground and artillery museum, and of Ft. Meade at 1.5" resolution.
<snip>

<snip>
But why in 1970 did they start with 1.5 feet?
<snip>

Starting ground resolution was 1.5 inch (''), not feet (')     

[Blackstar]

The "Final report" of the study, which had Autometric/Raytheon as one of the sponsors, and which is dated July 1970, describes that they obtained aerial overflight images of Aberdeen Proving Ground and artillery museum, and of Ft. Meade at 1.5" resolution.
<snip>

<snip>
But why in 1970 did they start with 1.5 feet?
<snip>

Starting ground resolution was 1.5 inch (''), not feet (')     

Touché.

In my defense, I did a little too much LDS in the 60s.