For the post above with the dashcam video where the car was merging in the highway and the driver had to take control, that looks exactly how it was in 7.0...and 6.2 with TACC as well.
I had meant to watch that video earlier, but got distracted. Thanks for reminding me!
I've also had similar issues since 7.1 on the highway with traffic traveling at 65-70mph. If I ended up catching up to a slower car in my lane, my MS would not slow to keep it's distance. My TACC distance is always at 6/7. This never occurred while on 7.0 even for merging traffic like in the video link below from my Blackvue cam. TACC did not slow for the car. It's obvious in the video how much time I gave TACC to react before I intervened. The car was displayed on my dash LCD as a vehicle directly in my path. My distance was set at 7.
Watching that video, I actually think that may be normal, improved TACC behavior, and I agree with Todd, to a point. I'm just not sure TACC worked that well back in 6.2, and I know it didn't when it was first released, but at some point between when it was first released and now Tesla improved it as follows, and I think that is what we see in the video.
When TACC was first released it didn't seem to take into account the speed of the car in front of you when deciding if it needed to slow down or not. All that mattered was the distance, and the speed your car was travelling. So, as an example, if a car passed you, and then immediately cut into your lane, even if it was travelling much faster than you, TACC would slow you down briefly, until the distance between your car and the target car had increased. At some point Tesla adjusted this, such that if it was obvious that the car was travelling faster than you, and that the distance between the cars would be increasing, the Model S would not unnecessarily slow down briefly, simply to maintain some preset mandatory gap.
In the video above, the car entering the highway is accelerating. By the time the driver manually slowed the Model S down, the two cars were probably travelling at close to the same speed, as evidenced by how quickly the car in front pulls away. I think the TACC software was accounting for the other car's speed and possibly even rate of acceleration, and not slowing the Model S unnecessarily, allowing less of a gap than normal, because it was only going to be for a very short time.
I definitely see this as an improvement in TACC over the way the first iteration functioned.