I talked to my coworker about how they operated when he was a Blackbird RSO. He said they could exceed 85,000 feet on a really cold day but they didn't do it very often because the controls became sloppy. As for the zoom climb technique for evading missiles, he said they never did it. They depended on pre-mission planning to avoid high altitude SAM sites to the greatest extent possible. After that, they depended on a low radar cross section (early stealth) to reduce the detection range to a point where the SAMs wouldn't have time to lock on, launch, and climb to their altitude. He said it wasn't a problem.

There were attempts to shoot down Blackbirds by firing SAMs blindly, but that would be about as likely as throwing a bunch of rocks into the air in the hope of hitting a diving falcon. Possible, but not very likely.

He was still end the program when the Blackbirds were grounded in 1990. He said that Lockheed had proposed a faster plane in the 1980s but it was never built. It would've only been a Mach number or so faster than the SR-71 which would've allowed it to fly somewhat higher, but it just wasn't deemed worth the money to build.

One thing to keep in mind was that we had a seriously stupid Chief of Staff of the Air Force back then. He was quoted as saying, "If it doesn't drop bombs or fire missiles, then I'm not interested in it."

Satellite reconnaissance is good but it has a lot of limitations, largely due to predictability. Unless a satellite maneuvers, you can easily predict when it will fly over you. This allows you to take measures to hide what you're doing. This is nothing new.* Maneuvering burns propellant which impacts the satellite's operational life, so they don't maneuver any more than absolutely necessary. Aerial reconnaissance can be much less predictable than satellites.

We both agreed we need a mix of reconnaissance capabilities to meet the military's and the national leadership's needs. Today, we have the U-2 manned aircraft for high altitude reconnaissance. We also have a series of larger reconnaissance and surveillance aircraft like the RC-135s. More interestingly, we have a series of unmanned aerial vehicles (UAVs) entering or now in service like the Predator and Global Hawk. They're a lot cheaper (at least to buy) than manned platforms and can stay on location for a long time. Unfortunately, they have a much higher accident rate than manned vehicles.

My coworker believes that Lockheed's later proposal for a Mach 8-10 aircraft could be built with today's technology. Unfortunately, neither of us sees it happening due to the expense. Perhaps a more affordable alternative is for the NRO to buy several of these satellites like QuickBird that offer better than 1 meter resolution. You could buy and launch several of them for the cost of a single of the current "national technical means" satellites and put them in different orbits. You'd end up with a much more robust constellation that would be a lot harder to avoid.

* I have a book about the Son Tay rescue attempt, where Green Berets raided a North Vietnamese POW camp late in the war to try and rescue the prisoners. They trained at Eglin AFB, FL on a full size replica of the camp. When they knew Soviet photoreconnaissance satellites were due overhead, they'd rearrange the camp so it wouldn't match any existing POW camp.