It's previously been posted and discussed in this thread. If you knew anything at all about the industry or autonomous driving you'd know what this report is, because it's the biggest industry report on autonomous driving and it comes out every year. But you're a clown without a clue.

Okay, so you're not trying to hide the fact that silly chart dates back to 2016?

LOL

You're citing a 3+ year old chart as authority on the state of FSD today. Brilliant!

If I'm wrong, post the date and source. There are multiple iterations of that chart. They obviously change over time.

General tip when you're accessing your university's "business intelligence" databases: information is less useful if it's not current.

Lol Waymo drawing dead vs Tesla.

The chart has previously been posted and discussed, year after year in fact (it's done every year).

That's not "3+ years old". It's last year's report, you lying piece of ****. First you claim I made it up (proving you haven't the faintest clue about anything in the autonomous driving world) then you claim it's three years old - it's last year's.

And God, I love these softballs you serve me. There isn't a person alive who could do a better job of unintentionally making Tesla look awful than you. You just gave me the chance to point out that Tesla has fallen behind further since 3 years ago:

2017:



2018:



2019:



LOL bro, another self owning for the ages. You're a dream opponent: the dumbest piece of **** alive, knows nothing, and just serves up endless opportunities to show what an awful state Tesla is in.

Baidu making a solid run, must be stealing a little from everyone.

lol ToothSayer gives loads of useful info. My request was meant to limit verbosity because that's his tendency and he's incapable of letting any of your bullshit slide. Anyone's really. I would hope it's not beyond you to realize that sophistry and straw men don't exist in his posts, and that alone is far more valuable to me than posts exhibiting such.

I understand the need for egos to be fed and god forbid either of you concede a point or maybe just ignore a post or two? I can accept ToothSayer gobbling up a thread even if it's repetitive, because he pokes holes in everything, right or wrong, and that has value when I'm deciphering information for free

You, however, keep killing your credibility AND you argue endlessly. Honestly, it doesn't matter how much validity your posts contain because on closer examination it is glaringly obvious you make **** up and play games. You're also petty, and if I didn't know any better, Elon Musk himself. I can't trust your posts for honesty and I'm not spending even more time wasting on 22 to verify ****. A multitude of other posters have sort of done that anyway by open commentary directed at you. That should give you pause.

If nothing else, on that video, I suppose that Tesla is parking in the appropriate parking space. It drives like it's handicapped!

So after 5 posts you are 50% of the way to a basic citation, which you obv should have done without prompting.

The entire premise of Tesla's ranking in the Navigant study is invalid (even if it was accurate at the time, which is dubious). It's just laughably inaccurate.

As hard as despacito sucks, he at least provides value by getting toothsayer to write detailed posts

re: FSD
I watched the Autonomy Day video. They showed the architecture of the neural network they're hardwiring into every car. Critically, it is entirely a feed-forward network. There are no recursive loops that would allow it to store recent memories, and that is a fatal error. It means that the network is processing each moment completely de novo. It can't remember that a particular car just got off the on-ramp and is going more slowly than expected. It fails at basic tasks, like remembering that it detected a truck in the next lane which suddenly disappeared.

FSD in general will be years away, and TSLA will never succeed as long as they stick with this neural network architecture.

They are approaching each snapshot like a chess position essentially with no regard for previous moves.

That’s theoretically possible if your snapshot includes a lot more data (velocity at least, possibly even history of velocity) and get a **** ton more computing power and a lot of software engineering.

At some Tesla fanboys have to admit Tesla doesn’t actually KNOwlW the hardware required to make FSD work. It is however a pretty good bet what’s built into modern Tesla’s are not sufficient with technologies (including software algorithms) that will be available within the life span of Tesla’s currently on the road.

This is even if Tesla manages to make some pretty significant breakthroughs in real time image processing to replicate data you’d get from LIDARs (and other sensors) within the next few years.

Yeah I'd love to hear some other voices on autonomy but I really don't know what can be said. There's zero objective evidence outside of Musk claims that they're anywhere other than the kiddie pool in autonomous driving. No testing miles. No public demos. No viable strategy that anyone takes seriously. We basically have Musk's claims, deployed level 2 software that's buggy and not much else to go on.

As for their strategy, machine learning images is doomed to fail with current and medium future machine learning for 10+ different reasons:

1: A 3d point map is stable and highly reliable while images are not;

2. A 3d point map can be simulated for wayfinding; images cannot be. This simulation ability alone allows orders of magnitude more learning and testing.

3. Cameras aren't up to the task of autonomous driving just yet; they are too slow to adjust and affected by serious bugs like lens flares. For example, this is the only camera covering 40 degrees of vision on the side of the car:



4. Images are unreliable; while point maps are high speed and precise, and no inferring about object movement or 3d position is required, images require a highly tuned machine intelligence to discern depth and movement information from a truly vast array of shadings, lighting, situations, etc. No such thing exists and likely won't for a while; even static robots working with a limited set of images have trouble with reliable image recognition, which is one reason why car manufacturing lines can't be fully automated.

5. Boundary cases are a massive problem. Blanket an area in a light dusting of snow and all of your images look completely different. Changing light patterns are confusing. Etc. None of these issues even exist with Lidar.

6. Deep learning is fundamentally not a secure algorithm; it has unknowable reliability. Deep learning of images is 100x worse. We don't know what will make it fail or misread the situation, what boundary cases it will fail on (run right over someone in black against a black road in poor light?). And because these are images and not precise point maps, there's no hard stop that can be coded; depth information and object placement and trajectory is guessed rather than certain.

I could go on but this is getting verbose as you say. Anyone with a functioning mind and some software experience can tell that this is going to fail completely. Just like anyone with a functioning mind and a little manufacturing experience knew that Musk could not robotize his entire factory to outdo the majors; it wasn't possible with the current state of robot image recognition. But Musk pushed forward anyway and nearly lost the company with his stupidity; he doesn't understand detail when it comes to hard problems, and FSD is way harder than robots.

There are just no arguments you can make that Tesla has any chance whatsoever of FSD by 2020 or even 2022. Everything is wrong. I love playing devil's advocate but there's just nothing to latch onto; it's all bullshit spewing from one conman's mind, a bit like Elizabeth's Holmes "one drop" fraud.

Haven't posted in a while itt, just read up.

Just a few points for despacito:

1/ Please give us your reaction on the advanced summon videos posted ITT. They are made by Tesla fans and show a horrible product not capable of reading lane markings, handicapped signs, parking lines,...
If your theory of FSD being solvable by Tesla in a short time frame is accurate, what the hell are those videos?
They highlight basicly empty parking lot situations, a place where people literally take their kids when learning to drive, and the car is doing a horrible job. Like it's the first 5 minutes of someone learning to drive.

2/ You keep attacking TS and I agree with some posters that it's entertaining. But if you are not a paid shill, you should really take a deep breath and revaluate wtf you are doing. If you don't see anything wrong, you should probably seek help (I'm not joking).
I'm sure that a lot of people agree with you that TS probably is a little bit extreme wrt his views on Tesla, but he has rarely posted anything that was wrong or stupid. Meanwhile, you keep posting stupid and irrelevant ****. I would lay good odds that there is no one with >50 posts itt that thinks you have 50% of the credibility of TS when it comes to Tesla. Anyone who disagrees can quote me .

3/ Can't resist: I LOLed really hard when you posted that Moody's report as relevant news. Going from B3 / Negative to B3 / Stable is great but if you knew what B3 / Stable implicated, you would not be posting it.

Other thoughts on that topic:
I was actually surprised to find out that Tesla was B3 / Negative, you'd think a company like Tesla would be able to get rating agencies to give them a better rating (not incredibly hard to build a bull case, and if a company does the right things they can deff get rating agencies to move a little). It makes me wonder whether Moody's knows Tesla's accounting is problematic.

Re Moody's: Tesla has a quick ratio not far off bankruptcy and most of their metrics on financial stability and soundness are awful. It's a horrible balance sheet backed by an unviable business that needs >$50 billion in inflowing capital before it gets to profitability at scale. They're in the business of selling stock and bonds to fund their massively money losing operations, with no end in sight, constantly knowingly lying about timeframes and future profitability. Musk's ability to lie about 200K 2020 robotaxis earning $100K/year each for their owners and then get billions in bonds and secondaries is the only that's keeping them afloat. Why do you think he does it?

agree with all of this.


It reminded me when the shill ValueAnalyst (hilarious name considering) was attacking bears about the point that Tesla was under reserving for warranties by coding all of their repairs to "Goodwill" (hence not charging to actual warranty expense, thus being able under reserve for warranties on cars sold and just expense as incurred), which is by every measure imagineable accounting fraud.

(S)he laid a very clever trap of, "If Tesla is coding all these expenses to goodwill, then why did Goodwill go DOWN on the balance sheet this quarter?"



I cant possibly comprehend the level of financial illiteracy it would require to make that statement, but this is a person who charges people for his financial analysis.

I just laughed at it again, it's so amazing

Good post. Pretty much nailed it.

This is the baffling thing. The rating agencies are telling everyone this company is in serious danger. The bond market is telling everyone this company is in serious danger. How is the equity market valuing the company at 60 50 40 38 Billion dollars?? I mean I get it, lots of dumb money out there. But rarely this much and this dumb.

I love that TSLA is such **** that it's prevented a similar thread from being started about Uber/Lyft/Wework. There are a lot of truly trash 'unicorns' out there right now.

Wework seems to be an outright money laundering scam.

WeWork is a giant REIT specializing in office rental masquerading as a tech startup and for inexplicable reasons not setting up new corporate entities for their new investments like every other REIT does to manage risk.

Its basically a 20-30 year bet on gentrifying/hipster/dense office neighborhoods’ office space. It’s not the first fund to buy basically cash neutral real estate banking on capital appreciation and it won’t be the last.

https://www.latimes.com/business/sto...-secrets-theft

Is it safe to assume no one really knows exactly who is winning the self driving race if people are getting indicted for stealing "secrets"?

thats the very nature of secrets, no?

Would like to hear more thoughts on we work. thread?

Haven’t read this thread, but figured best place to ask. How long we looking at until fully self driving cars (from someone, don’t care who) are getting routine use? Buying a house in a year and may be willing to live a little further out of town if in 5 years I was just going to chilling while the car drives me wherever.

I am guessing high density cities will soon experiment with geofenced self driving buses, possibly with dedicated lanes or just lanes where the buses get priority. What you end up with is basically a light rail system minus the expensive rails.

Yes. Exactly.

Hence it's a joke that anyone itt is confidently saying [x] or [y] company can or can't do this on a given timeline. They simply don't have the info to judge.

And trappy "business intel" subscription services are way off the mark too.

I agree with this. But what can we infer from it? It's a hyper competitive environment and stealing trade secrets is the norm. Secrecy is of the essence so the fact there's no evidence in the public domain does not mean they are not nailing it. Also doesn't mean they ARE nailing it. It is all proprietary, confidential, and very hush hush.

The correct answer, based on publicly available info, is that we don't know. Hence, your confident opinion about the current state of FSD is invalid.

No reaction. The feature is not complete, it's in beta, according to Tesla. Beta software isn't expected to be fully functional - it's not at all surprising or meaningful. Ask me again if it's trash after 1.0 ships.

Teslas can already do many of the things required for FSD discretely/separately but they're not yet woven into a cohesive end-to-end experience.

I also don't think we should over-infer from demo day what their ML/NN strategy is or was. Others have posted at more length about this but I'm not sure their premise was or still is valid.

I don't mean to attack TS personally. I like the guy for some reason. But he has a tendency to confidently state things which are either not true, or unknowable, as if they are facts, often without supporting evidence/sources.

That is not to discount that he is knowledgeable about some things, but he seems to have no off switch when he steps into territory that he does not or cannot (because no info is available) know about.

I've posted other negative things about Tesla itt. I don't have a purely bullish or bearish stance on the stock as I've said many times. Company failure is a possible future scenario. I don't recommend people to buy or not buy TSLA stock either.

A better analogy would be a talented young soccer player who hasn't perfected his game yet but is progressing rapidly and has a shot at the big leagues.

WeWork deserves its own thread. There's an epic implosion coming.

I could do a lot of things Messi does. I just haven’t woven those things into a cohesive package yet.