I did. Not sure what document you're extracting this from, but it's not the one I'm looking at. Also, their definition of what disengagement is, is quite clear:
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
- Thread starter AudubonB
- Start date
I did. Not sure what document you're extracting this from, but it's not the one I'm looking at. Also, their definition of what disengagement is, is quite clear:
That report isn't for driverless Waymos... Notice the column that says "vehicle is capable of operating without a driver"? It says no for all of those. Apparently, those vehicles only have an ADAS system installed, not fully-autonomous.
Which begs to question, if a driver was in all of the driverless rides, how often would they disengage? Probably a lot more often than the "in-field retrievals". At a minimum likely every time remote assistance is required. This is where it seems like Waymo is "cheating".
This accounts 3rd post on the forum and a further continued barrage of posts that received disagreements from me due to their harshness of tone and disregard for meaningful discourse.
Is Baidu going to be a customer of Tesla to utilize FSD technology or are they just providing mapping data?
None. I don't think anyone is on the "cusp" (within 2 years) of solving autonomous driving globally. Arguably both Waymo and Tesla make money from their autonomous driving systems, but only Waymo actually makes money only when the system is autonomous. If I were to have to choose one, I'd choose Waymo. I think objectively is impossible for Tesla to be able to take the liability for the system in a diverse enough set of circumstances with their current sensor stack.
My thesis is a short thesis based on Tesla's stock price. Is not a short on the company's future, which I actually like (as I said before). And I think Elon (and the people around him) are amazing strategists. But that doesn't change how much FSD success is already priced into Tesla's stock price. It's an unbalanced risk. I also think they won't make any meaningful progress over the next two years on FSD (beyond setting up a limited L4 pilot programme akin Waymo's in some geography). Car sales are a losing battle without FSD due to compressing margins.
Krugerrand
Meow
Again with the obtuseness. We absolutely know the technology is catching up because we factually know what the first attempt involved and the results, what the second attempt involved and the results, what the third attempt involved and the results etc… because they told us and many of us have lived it.
It’s obvious to all of us who’ve been here from the start that this approach, this version is an entirely different animal. We also know Elon knows this because he’s suddenly gone balls to the walls. Perhaps he’s wrong. I’m betting not. We’ll find out soon enough.
2daMoon
Mostly Harmless
I wouldn't call it cheating, per se. There has been no standard definition for "disengagement" assigned for assessing this.
It could very well be that Waymo has a very narrow classification they use for this term, and, it does not include remote drivers "helping" by correcting the vehicle's path while in operation. They may call that an "intervention" rather than a disengagement. Or, some other classification of their choosing.
It makes a kind of sense for them to do it this way.
Tesla, on the other hand, apply the term "disengagement" to any time an FSD Supervisor intervenes, as well as when FSD might have to disengage on its own. When the supervisor does intervene, FSD is no longer engaged, by design. The supervisor must re-engage FSD.
If Waymo are not counting remote intervention as a "disengagement" the numbers associated with these terms cannot be compared between Waymo and Tesla.
Stretch2727
Engineer and Car Nut
Elon comment on on the Ron B. interview.
It's probably a reduced sensor stack for cost reasons, but it's the more valuable data in my opinion. As you pointed out, in such cases the driver would disengage more often, so the data is more pessimistic. If we go by those numbers, they have ~200 disengagements with a safety driver, while driving 3669962.4 miles. That's a disengagement every 18 350 miles.
On the link I've sent, there's similar data available for Waymo as well. Just use the data for rides that had a safety driver present. I got the number above.
Krugerrand
Meow
That’s not at all how Elon thinks. It’s not how he operates.
I understand that’s the narrative you want to hold on to. I think it’s bullocks.
Nope. Not buying that either.
And definitely not buying that.
Are you serious?
The message I quoted above was addressed directly to me? Is that not harshness of tone? As I can swear I can't see you disagreeing with those posts.
2daMoon
Mostly Harmless
How many lives has FSD, FSD Beta, and FSD Supervised saved, compared to Waymo?
This is the metric that will determine how effective either system is over time. Neither will be perfect. Caca occurs.
When a system is operational at a level that beneficially surpasses the injury, death, collision, etc. rates of human drivers, that will be the system that gets a green light. Regardless of "meaningful progress..." as you define it.
Insurance companies already place their bets based upon the statistics. If the rates of liability-related costs are reduced, then taking the risk for that liability will be less than it is for human drivers.
Whatever that cost is will be built into the cost of an FSD subscription, and it will be less than a human driver will pay for the same coverage. It will be as balanced a risk as any other insurance policy will be, based on actuarial data.
Who has the most data?
Krugerrand
Meow
It looks like I may get the ‘Try Better Next Time’ award.
Either the first 100m shares and fakies was enough to reposition for a bit, or the tail got away and is dragging the dog into the street. However, there’s still a chance one of those Waymos runs over the tail and changes the outcome of the day.
ThisStockGood
Still cruising my Model S 70 2015
I grapped an extra batch of 53 TSLA shares on the dip... feeling great! Cheers to the longsTM
Yet to be seen, but Baidu is also partnered with Hyundai and according to this clip, the partnership is deep into maps and AI compute.
"Hyundai & Baidu Join Forces
automaker expanding Partnerships with
BYU so is Hyundai the two companies
signed a deal to collaborate on
artificial intelligence connectivity
autonomous driving smart transportation
systems and cloud computing they say
they're aiming to provide consumers with
connected V icle Services by and Hyundai
have worked together since 2014
developing navigation systems voice
recognition and other connected services"
The combination of direct distribution, improving manufacturing efficiency and reduced component costs (notably in motors, battery costs and reduced warranty expense as a result of lower problem rates and improved OTA diagnosis, has, in your view, not a glimmer of value?
Thank you for pointing to that. It's a wrinkle that I never knew about a man that I didn't know.
For others who may not have done the search, it seems to go back to the days of Zip2. Remember that Elon is OG mapping and navigation. Robin Li was, among other things, a staff engineer at Infoseek.
Similar threads
- Locked
- Locked
- Locked
