Thing is, with
@AlanSubie4Life, is that with one notable exception, he's been banging on the drum of, "It's all BS! It'll never be anything
but BS!", for a long, long time now.
I kind of
vaguely understood his point when we were all meandering around with 10.x; clearly, 10.x was barely acceptable as a test load, with 10 interventions (or more!) per mile. But the point wasn't that the load was horrible: It's that we were collectively sending data back to Tesla so it could become
less horrible over time.
And, while there were fits and starts in the progression of updates with regressions from time to time, there's absolutely
no question that the software did improve. The intervention rate, as nebulous and probabilistic as it was, was trending downwards and the occasional 10 mile drive without an intervention would pop up from time to time.
The 11.x series of releases were an eye opener. First, the limited access highway (LAH) versions of this software got seriously good; in fact, people use this stuff
now and rack up serious interstate miles sans problems. The City Streets, with the occasional regression, also improved, to the point where a random drive somewhere had about a 50-50 chance of not requiring an intervention. Seriously better than the 10.x releases.
While
@AlanSubie4Life did properly criticize the failings of the 10.x and 11.x branches and, yup, there're many of those failings, his "It's all BS! It'll never be anything
but BS!" analysis of what FSD was doing
never changed. I had begun to get the impression that, no matter how good FSD got, if it didn't drive exactly the way Alan wanted it to drive, then it was back to full roar "It's All BS!"
And then we got 12.x releases. Something weird happened. For almost a week, Alan actually said
good things about the software. I just about fell off my chair.
I'm a-telling you: I use FSD(S)
all the blinking time. Yep, I do intervene - but, out of, I dunno, ten drives, I might only end up intervening twice, maybe three times. And I'm talking drives from three miles to fifty. I've seen 12.3.6 respond to
hand signals, politely let cars turn in front of it, and, when another car pauses to allow the car room to turn, FSD(S) has taken that turn. It's acting humanish, and
that's new.
There's this point I've made over time: "Is the software ready for Ma and Pa Sixpack?" On 10.anything or 11.anything, I'd've definitely said no. On 12.x..
Right now, there's two close friends, both of whom have bought Model Y's in the last eight months, one of whom is the type one has to lead down the garden path on anything technical, the other being somewhat self starting, but no expert of self driving cars.
Both of them love the software and use it. Um. One of them
is a Ma Sixpack, the other's not far from it, and they're using the driver assist software the way it should be used. One of them has definitely paid for the FSDS and plans to keep using it.
So.. what happened with Alan? Looking at his current set of comments and evaluations: It's as if nothing has changed in three years. The, "It's All BS!" rant.. sounds old. And has no sense of
proportion, or the changes that have been wrought in the last three years, and especially in the last six.
Frankly, while I'd like to see a newer version of FSDS, just like the rest of you, there are the repetitive statements from Tesla principals that they are Very Much going for Safety. I have the distinct impression that they
are going for a Ma and Pa Sixpack target. In that case, the lack of new-every-two-weeks releases that we'd all like to see isn't a
failure: It's a desire to Get It Right This Time.
We'll see. But I'm expecting great things from the company.