The streak climbed exactly where I never engaged FSD.
That is not what I thought it measured.
First recorded v14.3.8 trip: 64.6 miles between Aguadilla and Mayagüez, 93% on FSD and no intervention while it was driving. I left home with a 103.4-mile streak record and ended the day at 165.0, while driving the hard parts myself.
Watch the complete trip.
Fifteen minutes covering the wrong turn, both parking attempts, the scare, and why the streak goes up when you stop engaging FSD.
The numbers, untouched.
The streak does not measure what I thought.
I believed the streak measured the miles the car drove itself on a trip. It does not. The streak measures consecutive miles with FSD engaged without me having to intervene. Those are two different things, and the difference matters.
On this trip there were four stretches where I never engaged it: the potholes and the speed bump in my neighborhood, parking in Mayagüez, hunting for a spot before lunch, and the speed bump coming home. The hard parts, all driven by me. And the streak kept climbing: 103.4 leaving, 122.1 mid-afternoon, 131.5 later, 165.0 on arrival. 6:20 9:36 13:31
That means the number can be protected, or inflated, just by not engaging it where you know things get tricky. I say it because it changes how any streak record should be read, mine included. That same day the cumulative FSD share read 95.6%. That is the honest number. 14:42
What it did well and what went wrong.
What it did on its own
- It held the main road. Two cars wanted into its lane. It did not yield where it had no reason to and carried on correctly, without wavering. 3:16
- It parked itself, twice. The maneuver was its own. The problem was where it did it. 5:35 7:40
- It stopped on its own before the speed bump. I set the destination right before it so the car would arrive and stop, and from there I drove manually. 12:42
What went wrong
- The wrong turn. It should have gone straight and instead looped us around the whole area. 2:01
- The scare. A car was pulling out, FSD hesitated and braked hard, then carried on. It never became an intervention, but it did not feel natural. 11:26
- Three times it parked where it should not have. Wrong spot, a spot we were not sure was the right one, and a spot where we did not want to leave the car. All three had to be moved. 10:02
- I changed the profile myself. During the drive it went from Hurry to Mad Max. That is why SPD-02 and SPD-06 are marked "no": it did not hold the speed on its own.
A real test is not a lab.
This is one of two v14.3.8 trips. The version score comes from both together. The other is Aguadilla ↔ Hacienda Muñoz, on September 6, and it is the long one.
The left turn out of my house, the neighborhood speed bump and its potholes were not evaluated here because I never engaged FSD on that stretch. They are known failures, but with no observation they do not enter the score. 0:56