7/17

Today

Oh Captain, my Captain

https://www.profgalloway.com/wewtf

interesting write up on WeWork that touches on a little bit, and more.

That is not inconsistent with my last post. I still believe that. Tesla has openly said there are edge cases being worked out, is that controversial?

Interesting read. Softbank's preferential shares (and the fact the Vision Fund bailed on the majority stake, with Softbank taking a much smaller stake without the Vision Fund) are pretty telling. Already customer churn seems likely to be a huge problem for WeWork (people don't LOVE their product) but in a weaker economy I expect it will be catastrophically bad.

Stratechery take (includes a bull case): https://stratechery.com/2019/the-wework-ipo/

Should start a new thread for this imo.

How much property do they even own? I thought the business model was to lease office space for long terms and then sublet it out at higher rates for shorter terms.

Also I believe they do form different subsidiaries maybe not for every project but I think they can walk away from most of their leases.

The secret is the internals of Waymo's algorithms, not how far advanced they are. That's known. We know who's in the race and the relative position of the players. This isn't hard. You need to extensively test on public roads before you release, for obvious reasons. In Calfornia, where Tesla's team is based, it's a legal requirement to report all miles and disengagements. Tesla reported zero miles for last year. The last time Tesla reported testing miles, for their short preprogrammed demo a few years back, they reported 1 disengagement every 3 miles (compare with Waymo at 1 per 12,000 miles and dropping fast - Tesla was 4000x worse).

There's no way to hide a fleet of autonomous cars extensively testing in the city center - any city center. And you need to test with safety drivers and then without drivers for years/millions of miles before you release to the public. If they were doing this we would know - there's intense interest and it would be leaked. Not to mention Musk would be gleefully shouting it from the rooftops and showing demos. If it was a super secret program, why would he announce to the entire world that they'll be feature complete by the of the year? That's a guarantee to get your testing fleets outed. Yet no one has seen anything.

It's a transparent con like all of Musk's other cons.

Yeah man, everyone is off the mark, and the guy who's committed major security fraud, missed all of the targets he's set and put out clownishly incompetent software like this below, years later than promised:



is going to solve level 5 autonomy in 4 months or even 40 months? How much of a gullible loser are you?
Nah, you're like an 11 year old who just discovered philosophy and thinks he can use to get out of homework. Some months ago you were actually so absurd that you used the same reasoning as the above to excuse Musk for calling a stranger a pedo after the stranger criticized him ("we can't know for sure whether he is a pedo, therefore you can't say with certainty that Musk making up false claims to defame someone was invalid!" was your basic argument)

People who are older than 12 look at evidence. What evidence is there that Tesla are close to solving level 3, let alone 4, let alone 5? There is zero evidence. The only reason we're even considering the claim is because a serial liar and fraud claimed it, one week before going begging for $3 billion to prop up his massively money losing company (which he also promised would be profitable all quarters going forward just six months earlier).

So there's no evidence of Tesla being anywhere near level 3, let alone 5. Against that we have:

1. A history of lies and fraud by the person making the claim.
2. A gross lack of sufficient R&D to pull off an engineering feat of this complexity
3. Background priors that anyone can solve this next year with current hardware and software at a very low percentage
4. Background priors that it can be solved using just images (we know there are no Lidars on Tesla cars) orders of magnitude lower than the above
5. A software team shows rank incompetence where even their touted upcoming releases (advanced summon promised in 6 weeks 43 weeks ago) are a clown show - see the video above.
6. An absence of any of the kind of public testing that would be required if they were so close.

No, it's being released in 6 weeks ago 43 weeks according to Musk.

Your problem is that you have a very low comprehension of how the world works. (calling WeWork a subprime REIT shows this extends far past Tesla). Beta or even alpha or pre-alpha software still has the basic functionality in it. Tesla's "advanced summon" can't do the following:

- Wayfind even remotely competently through a near empty parking lot
- Drive on the correct side of the road
- Not block traffic or stop in the middle of the read.
- Obey lanes
- Recognize parking lines

It's highly incompetent. Beta software with a functioning core would not make such gross mistakes or display such incapabilities; it doesn't work like that. The very first task you build into a parking lot car retriever are:

- Detect the correct side of the road to drive on
- Detect parking lines
- Map out driving space vs parking space
- Wayfind by analyzing your surroundings
- Try to follow the roads rather than stupidly driving toward the user by any route possible if the way is clear.
- Teach it to pull into the parking space so it's not sitting in the main lane of traffic.

These problems are orders of magnitude easier (in both time and complexity) than level 5 autonomous driving. Yet even with these simple parameters, the software is clearly highly incompetent and can't perform basic functions required of version 0.1, let alone a beta version. At this rate they won't even have basically functional parking lot retrieval software by the end of 2020.

But we "can't know" if they're gonna solve level 5 FSD in four months.



Of course we can infer it, you clown. They have no lidar on their cars which means they're doing image recognition, and the only way to do that is machine learning, which is hopelessly inadequate for the task right now, and why far smarter people have opted to add Lidar. Tesla's head of development is a guy whose only expertise is in image-based machine learning.

I post large amounts of supporting evidence/sources, usually primary sources.

Your problem is that you're a raging dumbass when it comes to philosophy. You know a little but not a lot which ironically has made you dumber than someone that knows nothing.
This disingenuous bullshit doesn't fly. You're on record as saying you think they'll have full autonomy in 2020. But you won't make an escrowed bet on it. Says it all really. Even you know it's not coming in 2020 or any time soon.

Don't base any decisions on FSD being commercially viable anytime soon. This also depends on your city. Phoenix seems likely to get something first with their easy maps and predictable weather.
If I had to guess I would say there is ZERO percent chance at anything resembling a nationwide commercial offering inside of 10 years. Some version of this may be available on a limited basis in select areas (like phoenix or geofenced) before then.

Tesla has designed its own FSD hardware, including an application specific integrated circuit (the FSD chip) which has not only been manufactured with what looks like good price/performance, and redundancy, but is already on sale by Tesla in new vehicles or as an upgrade.

Your criticism of Tesla's FSD software team is both incorrect and incomplete. They have world class software engineers, with a ton of practical experience, including people who have been re-tasked from SpaceX (I don't know of any organization with more stringent operational intensity). Other car companies dream of having Tesla's software and software engineering talent. Yet you mock it and pass it off with a one liner about Karparthy supposedly being too narrow or academic.

Who the **** are you?

I do believe this will happen, but I frankly acknowledge there are multiple factors that could rate limit progress (for example, regulation, but there are many others). Therefore it is not something I am confident enough in to bet large amounts on, which is entirely reasonable.

That is also why I don't advise people to buy the stock. It is HIGH RISK.

How could I possibly be less disingenuous than this?

I stand by my prediction that Tesla will achieve FSD in 2020 though.

Are you saying you want to take the other side of that bet or just posturing? How much are you proposing?

I don't know enough to judge your credibility as a counter-party either, would need to look in that, but have seen some suggestions that you didn't fulfill the terms of a prop in another thread. Not expressing an opinion on that and would need to look into it further.

It's ridiculous to the point of being extremely stupid to infer anything about Tesla's FSD capability on the basis of a beta version of summon.

Why would they release 1.0 until it's thoroughly Q&A tested and approved by regulators? Doing so would be unforgivably negligent.

You're reaching and desperate. SAD! I expect more than this, even from you.

Long term leases become very valuable assets over time.

They do form subs. They just don’t do it as diligently as they should. Whereas a REIT would typically get the entities formed and use the entities to acquire, WeWork has sometimes simply signed and then contribute the stuff to new entities. It’s a pile of lawyer fees that they otherwise don’t need to pay.

This is actually not really a big deal by itself. I think of it more as a general sign that the firm lacks professionalism.

I'm the guy who correctly predicted (far more correctly than the CEO of Tesla!) the trajectory of their self driving starting several years ago. I simply looked extensively at the evidence of their software output and predicted that nothing Musk grandly claimed on autonomous driving would come to pass and that they were far behind and we would see no great breakthroughs in the next few years. I was correct.

By saying this:
And putting no money up, you're not standing by anything.

Confirmed a ******. On a gambling site no less!
Do you know what the bolded means?
Let's see. Musk promised in January 2016 that "summon" could drive from anywhere in the US to anywhere to anywhere in the US in ~2 years.



This is where "summon" - now called "advanced summon" - stands 3.6 years later, in August 2019:





Failing at about 10 basic long-solved tasks needed to navigate a traffic-free parking lot. It's comically inept. This is software that ~1.6 years in the past was suggested could drive from one of the US to another.

This same software was also promised 43 weeks in 6 weeks by Musk.

This same loser, who failed so horribly at his prior predictions and the same team that failed so horribly at achieving very simple tasks compared to level 5 autonomy, is now the only source we have for claims that Tesla will have 1 million autonomous robotaxis on the road in 2020. Anyone who believes him is a gullible clown. That's without even getting into his massive stock fraud, solar roofs fraud, dozens of missed timelines and vaporware, etc. That's just solely considering his autonomous driving track record.

You claim we shouldn't look at this evidence and infer anything about FSD from it. I can understand why you'd say that; it totally destroys Musk's claims.

Elon Musk is reaching and desperate as is amply demonstrated by his comical claims right before asking for billions of dollars in funding when the stock is tanking. I'm laying out the evidence, which you're trying your hardest to obfuscate away, but those tweets and those videos speak loudly. I love that you keep posting and letting a wider and wider audience see the reality.

Edge cases. I guess you could call it that.

Parking is an edge case apparently. So is not running into a highway divider. That’s some massive edges.

"Those are..."

I love how much of a self-aggrandizing ****ing moron you are.

Yes, of course.

But even if I wanted to bet in general, you'd be the last opponent I'd choose. Even with an escrow I'd expect it to be a ****ing **** show (see this thread to understand why).

It's close to being able to achieve this already.

Know many people first-hand who do 500 Mile+ journeys on Autopilot the vast majority of the way (90% or more).

If you summon a Tesla from NYC to LA and it arrives flawlessly, but then misses the lines on the carpark, do you really think robo-taxis are 10+ years away? Of course you don't.


I claim you're deliberately ignoring the vast majority of relevant evidence which points squarely to the fact that FSD is progressing SWIMMINGLY.

Instead you focus on less relevant or inaccurate sources and blow them wildly out of proportion.

Why would anyone raise any money to create a company that manufactures fully autonomous electrical vehicles? How dare they? Outrageous! It should be fast, FREE, and furious!

And far from promoting Tesla, he should be talking DOWN the company and its future prospects whilst fundraising. This is unheard of and totally unacceptable!

As for the stock, it has performed brilliantly since its IPO (at $17 per share).

I didn't want to dig this up but you forced my hand by persisting with the prop bet suggestion (after I previously asked you to not raise it again).

It's absurd that you criticize Musk's credibility and integrity, and yet you yourself act super shady over a measly $1000 prop bet.

Just imagine what you'd do if there was IPO money on the line.

It would be a flaming dumpster fire of such unprecedented proportions it would make Theranos look like a well managed business.

It's no coincidence that TS rhymes with BS. Wake up and smell the bull ****!!!

Do you know what you call a car that drives itself perfectly 90% of the time and unpredictably 10% of the time? An accident.

Do you know what you call a poster who deliberately misconstrues my posts 100% of the time and responds with trolly nonsense 100% of the time?

Didace.

I said they drove on Autopilot 90%+ of the time. At no point did I say they were driving on Autopilot and unpredictable **** happened 10% of the time.

Various reasons for not being on Autopilot always, including laws, an abundance of caution, and the current state of the software.

Not a single one of them has had even a scare, let alone an actual accident.

I personally wouldn't place as much faith in Autopilot (yet) as some of them do, but it has not turned out badly for any of them.

despacito having a full meltdown on me pointing out that he doesn't really believe they'll have robotaxis in 2020. Great stuff. I think the videos are triggering him too, how embarrassing is the reality after Musk's grandiose promises?
Yes, I think that might be a factor.

The poor bastard is caught between a rock and a hard place. He knows 2020 robotaxis is bullshit (he's a fool, but not that much of a fool) but he also wants to avoid having to lose money just so as not to lose any remnants of his reputation after claiming he "stands behind" 2020 robotaxis claim (with what?).

Tesla’s Autopilot was engaged when Model 3 crashed into truck, report states.
It is at least the fourth fatal crash involving Autopilot

Man killed in gruesome Tesla autopilot crash

Tesla Autopilot Malfunction Caused Crash That Killed Apple Engineer

There are many more that weren't so high profile. Both autopilot, automatic emergency braking and collision avoidance fail completely on Teslas quite often in situations any competent level 3 should be able to handle without a collision.

The relevant comparison (which I have never seen you even attempt to make) is how safe Autopilot is relative to a competent human driver.

Your modus operandi is: post sensationalist news articles, with zero attempt at a data-driven analysis.

I wouldn't wipe my ass with those sources (register.co.uk? gizmodo.com? REALLY?)

You claim to post authoritative sources, but in the end you always fall back on your staples: Vanity ****ing Fair and trash newspapers with zero integrity.

If Autopilot has 4 crashes in [x] years but is 5000% safer than human drivers, it's a really good thing. Without the statistical context such figures are meaningless, and you clearly don't care.

Not at all embarrassing.

As I said, you can drive 500 miles+ on Autopilot without a problem.

You neglect to mention that in one of the videos the narrator explains that if there are cars in the parking spaces, it does recognize and avoid those. It's only when they are empty that it straddles multiple parking spots. That's actually a good thing. Of course, you're too biased and stupid to know why.

Who cares if Summon doesn't yet avoid lines in an EMPTY carpark (which it will, imminently, with a software update)?

It won't be long before people who are NOT using Autopilot are in the minority, and people like Toothsayer are the subject of clickbaity headlines, for causing a disproportionately large % of total accidents. Enjoy driving your non-autonomous hybrid you ridiculous douche.

The stories I posted are extremely well known.

Autopilot's decapitation of multiple people was widely reported. The pulverizing of the body of a Model 3 driver (an Apple engineer) which autopilot ran into a highway divider at full speed was also widely reported. Here is one example of someone actually verifying the flaw in Autopilot where it attempts to slam people into the barrier even after the accident, because it gets confused by lanes (showing a complete lack of basic safety features and proper 3D mapping), as reported on CBS news:



You either don't know this or are desperately obfuscating again as you often do.

That's a tiny fraction of autopilot-caused deaths. While we're making up total bullshit: If Autopilot had 4 crashes and solved world hunger, it's a really good thing!

Later model luxury cars are extremely safe, especially on highways where autopilot is engaged. We don't have enough data points to say whether it's more dangerous or not, but we know it fails repeatedly and in very stupid ways it shouldn't if the software was even slightly competent above level 2 (running someone at full speed into concrete dividers/fire trucks/crossing trucks and pulverizing/decapitating them is a pretty major flaw).

You're defending their performance on these videos? That's gold, Gerry.

I agree level 5 autonomy like Google and GM/Cruise are building will be amazing. Tesla are just a waste of space in this arena however, as confirmed by multiple reliable points of evidence.

Of course I'm not obfuscating.

Why don't you address my point: that you take a sensationalist, rather than statistical/analytical approach, because doing anything else would uncover your blatantly false arguments.

Is that your final answer?

You're saying: there are not enough data points to judge whether or not Autopilot is already safer than a competent human driver?

Please confirm.

As a counter-point, here is what Tesla claims.

On Tesla's data, driving with Autopilot engaged = less than 1/6 the chance of an accident than the national average.

*** I am aware that Tesla's methodology and data are not without critics.

I wasn't referring to the 90% you talked about. It's just a coincidence that we used the same number. Go ahead and reread my post using 99% and 1%. Same idea.

So how about that escrowed bet between Tooth and despacito on Tesla achieving FSD in 2020?

despacito, you say you're not confident enough to bet large amounts on this, but given the right odds and amount surely you're willing to put some money on this for you +EV situation?

Back up a second.

You give me a personal anecdote:
In response to your anecdote, I point out that it does indeed turn out badly for people, with sourced factual examples of deaths caused by autopilot. Your response is to claim my sources are bullshit (a highly dishonest claim that you know is dishonest). You then (after posting an anecdote!) demand I use statistics to make my point after you used a non-sourced anecdote to make yours?

There are enough datapoints to know that Teslas are less safe than a number of modern luxury cars. There aren't enough datapoints on autopilot safety because Tesla holds that data close and doesn't release the raw data for analysis (for good reasons as far as they're concerned). All we have are anecdotes of multiple Autopilot killings in highly incompetent ways (see the CBS news above for example), and thousands of forum posts by Tesla owners and on youtube documenting on the crazy stuff it does (swerves toward trucks, tries to drive into barriers, slams into parked objects, gets confused by lens flares and shadows).

Tesla are currently undergoing NHTSA investigation for their false safety claims as a result.

Tesla received a cease-and-desist letter from US agency over Model 3 safety claims

Tesla is currently under subpoena to release their data because they're not giving it willingly it seems. I guess we'll know for sure in a few months once the report comes out.