You seem to have gotten your tail in a twist over a bunch of things I never wrote. I was commenting on a post by a self-described "idiot" who was saying how amazingly wonderful FSD Beta had suddenly become for him. Not autopilot, not divided highways on long road trips: FSD Beta 10.69.2. And that it is now worth the ridiculous price Tesla is now charging, or even more, because wow!
And yeah, it's gotten a bit better, on the whole. Sadly it isn't anywhere close to autonomous, which means it isn't much use for all the use cases that would make it particularly valuable. That's undisputed. Sure, it's getting closer. And if you are willing to bet on the come, then maybe it's worth real money now. But so far, progress has been asymptomatically approaching better but not nearly good enough for autonomy, so I think not.
And while I'm sure you have "used it successfully on city streets at times", that doesn't mean it's anything but a novelty at this point. Successfully pretty much just means no interventions, not that you could pay no attention. You still had to remain attentive and you were responsible for any problems. So, it's fun if you like that sort of thing. But it's hardly worth real money.
What is worth something at this point is Navigate on Autopilot on highways. It consistently leads to less stress and better drives. This is (for me) mostly because I can pay much less detailed attention to what's going on, just monitor the big picture for unusual threats.
I originally thought that FSD was zero-or-one too (two?
). That is, it only had value if it required no oversight.
Now, I have a more nuanced view. It has considerable value even short of full autonomy.
As with Autopilot on freeways, I find that it takes less effort to drive and I feel I have a better overall situation awareness than if I have to focus more on keeping to the lane and maintaining a particular speed.
There was a time where I felt it demanded more than it contributed. Now, I think it contributes more even with the limitations that remain.
Over time, it’s learned—quite a bit actually—and I’ve also learned. I pay attention to where I am, the time of day, and what’s ahead and then I engage, disengage, or modulate FSD based on the situation.
For example, I use it in local school zones—where there are sidewalks, bike lanes, crosswalks all well marked—even though I still have to manually and quickly switch from the 30 mph speed limit to the 20 mph school hours limit every time I enter one here during school hours (because FSD doesn’t ‘get’ school hours yet). I find that my car keeps to the limit and the lane quite accurately. I expect that it will also avoid kids, bicycles, etc. I am able to keep a better watch for kids, bicycles, etc. myself than if I were driving without FSD. This is one reason FSD can make for vastly safer driving.
We ‘share the road’ where I live, meaning pedestrians, bicycles, and vehicles mix on many roads—without bike lanes, sidewalks, or much in the way of shoulders. Often there are tourists during the summer. We have a local tradition of giving pedestrians wide berth (remember pedestrians usually walk against traffic by law in the US) well before it is necessary just so folks walking don’t have to guess whether they are going to need to jump aside.
When will FSD learn this courtesy? Who knows? I don’t know if it’s learning from me. Does the FSD team study road segments where everybody turns it off?
Roundabouts I take on a case by case basis. I report issues, but usually not more than once per release.
Still FSD is a great tool now: I use FSD on almost every drive and most of the time during those drives. Even though I knew this, it was still a bit of a surprise how much I missed it when I didn’t have it when driving other cars recently.
Sure, you can complain that it isn’t yet fully autonomous everywhere and all the time. But the claim that it must be fully autonomous to have value is a specious one. I paid for it in part to help encourage the technology and I am encouraged by the progress.
My personal investment target for initial robotaxi deployment in the US has been 2024 (iirc, that’s been my target since I joined TMC). Seems to me we are track for that still.
