This was a "dumb" rocket Eagl. As one of the engineers said the test would be successful even if it just cleared the tower. None of the advanced flight avionics are implemented yet; would be silly to put lots of $$$ into the first test launch of a rocket you don't know will even fly.
Absolutely not true that it was "dumb", even if it used developmental flight control software. If you look closely, you can see the second stage rocket nozzle gymballing very dramatically in an effort to control the roll oscillation, but it is either overshooting or undershooting the required control inputs, or responding too slowly. I froze a few frames to measure it and the nozzle is "chasing" the roll, indicating that the control loop is either too slow or the gains are set really badly and the control inputs are too large. It isn't something unexpected in a flight test program, but NASA (and Boeing and Airbus) seems to do a lot better job predicting flight control laws via software simulation during the development program.
In any case, the rocket was clearly controlled and "aimed" for a particular orbit. Heck, you can see right after launch the rocket executed a pitch maneuver and settled down in a stable direction. The camera shows some disturbances in the rocket flight path, but during first stage burn the control laws appeared to be very effective so the rocket would bobble just a bit and immediately correct back to the correct path. It was the second stage that didn't dampen out the oscillations properly.
And that is why I point the finger at the guy writing the flight control laws, because the previous falcon rockets had a very similar issue near the same phase of flight. With the previous rockets, the oscillation didn't force a shutdown. With this rocket, the oscillations appeared to be increasing right before the video is cut short, and a news report stated that the motor shut down when the roll appeared to exceed the motor's ability to keep it going straight. It must have worked long enough though, because the orbital parameters seemed to match their expectations within a percent or so, again according to the news reports from actual space experts (not CNN).
I did notice that the first stage separation was violent, and I wonder if there may have been some damage during the separation. SpaceX has had problems with stage separation in previous launches, so again maybe they are reinventing the wheel for procedures that NASA, Boeing, etc. nailed down decades ago. To be cute, "there is an app for that" (software simulation) and they need to do it better if they don't want to throw away money on flight tests that are hampered by issues that ought to have been tweaked out in the simulations.
Heck, my Mom's work is all about simulation. Their clients describe a physical process, and then her group's egghead scientist types will create complex mathematical models describing the process. They run a few million test cases, varying their assumptions with each test run, and then compare the simulation results with a sample of real life test data. That lets them pick the computer model that best represents reality, and then they can run any simulation their customer wants, with a very high probability that the simulation results will match reality, without having to run more expensive or dangerous real world tests. Her company figured out why half of the early polaris missile launches blew up (they got a DoD commendation for that) and they did a ton of work on protecting vehicles from mines, by shaping the vehicle underbody and determining how much underbody armor was required for each shape. They probably saved their customers hundreds of millions of dollars in development work.
So my observation is that SpaceX needs to do better simulation work, because their videos are showing issues that ought to never have been discovered on flight hardware. Flight control laws in particular should be hammered out in simulations and they should not have seen those undamped oscillations on multiple consecutive launches.
