Follow

Tesla is once again hiking the price of the "full self-driving" feature (that does not provide full self-driving yet, of course), because "when it's done" it will be worth "much more". It was supposed to be done "in 2020".

Meanwhile Tesla cars can follow (unfaded) lines and slow down to avoid rear-ending slower cars. These features are common in just about any robotics tutorials you'd expect 8-year-olds to complete in a single CoderDojo session.

This is a 12,000 USD option. Musk is a scammer.

· · Web · 3 · 7 · 15

The only thing you have to know about the current state of the so-called "autopilot" in Tesla cars is this : in the Las Vegas "hyperloop" (which is actually a tunnel for cars, because Musk Reality Distortion Field), where all cars run in a single lane in the same direction, they still are driven by humans.

That's the kind of trust Elon has in his own AI tech.

There is a slight problem when it comes to fully autonomous vehicles on public roads: we have absolutely no idea how to achieve that, it's in the realm of science-fiction because actual science hasn't come even close to such feats. No actual AI researcher would ever tell anyone to expect full self-driving cars in the coming decade.

Of course that doesn't prevent crooks from selling it.

In related news, do you want to buy my portable fusion reactor? It's not completely done, but it will be soon

@jkb Rodney Brooks has some pretty good posts on its blog about the state of AI and of the features that were/are predicted. He's also a huge sceptic of full self driving cars.

rodneybrooks.com/blog/

@jkb there is a lesson to be learned here. A painful lesson. Not by some inane AI that is a glorified OLS algorithm but by the human collective:

We need mechanisms, persistent institutions, that survive the political cycle and which can bat the moles back into their holes rather than allowing them to expand into a vacuum of governance and common sense

@jkb Minor correction - Hyperloop is evacuated tube high-speed (and Musk's companies aren't even doing that), and the car tunnels are just "Loops".

And with that, time to mock the shit out of Loops: Morgantown PRT was doing basically the same thing as the Vegas Loop in the *1970s*, but *actually* fully automated, and with more than two stops (which means more routing complexity)... on a PDP-11.
@jkb Like, the problem space of a Loop was so simple that a PDP-11 could do it with a 100% safety record in the 1970s. (I think Morgantown PRT's first crash was in the 2010s, and that was after a control system retrofit to a Pentium-based PC in the 1990s.)

Yet they're using human drivers.

@bhtooefr Ah yes indeed, the commercial name for this particular kind of bullshit is Loop and not Hyperloop, thank you.

Also, Elon said it would be operational by the end of 2019 twitter.com/elonmusk/status/11

Also, the Las Vegas Convention Center's call for tender asked for a 4,400 passengers/hour capacity, this "loop" has only a fraction of that capacity at 800 passengers/hour.

That's what you get when you select the lowest bidder, who also happens to be a crook

@jkb @bhtooefr Who's shipped a full-self-driving feature when and for the amount they promised at this point? For that matter, who has ever shipped such a large technological leap on time and under budget?

It seems to me that, like with all technological advancements, even the people closest to it aren't able to do a good job of predicting.

You can quibble about whether Musk should be making such pronouncements when he can't possibly know, but it seems to me every company does that.

@freakazoid @bhtooefr Let me restate the facts in a simple way : Tesla has *sold* for years an option, for an ever-increasing price, that promised buyers their cars would have full autonomy by 2020.

They have promised, and taken money for, something that does not exist.

Now they are behind schedule, but are still selling that non-existing feature, for an increased price.

Actual AI researchers confirm there won't be full autonomy in the lab in the current decade, let alone in consumer products

@freakazoid @bhtooefr Also: this is not only Elon Musk the individual who makes these outlandish promises and sells vaporware, it's Tesla the company

@jkb @bhtooefr Yeah they act a lot like a Silicon Valley startup in that way. I had forgotten that they were actually selling it as a feature.

I think one major issue with both Tesla and SpaceX is that because so much of their funding has come from Musk himself they haven't had a lot of market discipline applied to them. This can be an issue with angel investors as well, since they may not actually have the experience the company needs.

@freakazoid @bhtooefr I think you're right, the lack of accountability is central to Musk's companies' problems. SpaceX is the one that's most on tracks because it's also the most closely scrutinized

@jkb @bhtooefr I'm biased in favor of SpaceX because I've been involved in "newspace" since 2000 and he's following exactly the playbook a lot of the people around me were pushing. But I think you might be underestimating the difficulty of actually shipping products, especially cars. Car startups with much more experience have failed, and at the time Tesla started the state of the art for electric cars was stuff like Corbin Motors, and cars the automakers were only making because they had to.

@freakazoid @bhtooefr Of course shipping new products is difficult, especially when you want to be on the bleeding edge of technology. I'm not blaming Tesla for not developing tech fast enough, I'm blaming them for selling tech that not only is not developed yet but also cannot exist at all given the current state of science.

It's like selling a teleportation device, or as I tooted earlier a portable fusion reactors. These are theoretically possible, but will not be feasible for decades to come

@jkb @bhtooefr Agreed. I was just responding to what you said about SpaceX being farther along. Their videos are extremely impressive, but in many ways the problem they are solving is easier. In fact, for the most part all of SpaceX's problems have been solved before, which makes SpaceX a lot more of an engineering company than a research company. It's weird to think of a car company as making more of a foray into unexplored territory than a space company, but I think that's really the case.

@jkb @bhtooefr Of course, the "unexplored territory" that Tesla has entered is market territory, and SpaceX's market is a substantially more speculative one than Tesla's. Tesla really started out as a luxury car company more than an electric car company, and the luxury car market was well established. Nobody has any idea if the space launch market can support the flight rate SpaceX needs to survive at their price point. I hope and believe it can, but over the next 2 years? No idea.

@jkb @freakazoid Honestly, I think one of the big problems is that Tesla “achieved the impossible”, a few times, through sheer force of will.

The problem is that the “impossible” that they achieved wasn’t impossible at all, it was just against conventional wisdom. And, that conventional wisdom had been poisoned by fossil fuel interests.

It also didn’t help that Musk’s management style - which he openly admits - is to publicly promise an impossible target for a future development, to motivate engineers to work harder to try to meet the target - even if they fail, the idea is that they still get it done much faster than they would have with a realistic target. This is, of course, incredibly abusive and burns people out rapidly… and contributes to the specific problems around FSD.

Basically, Tesla’s earlier history I think taught Musk and the other Tesla management two incorrect lessons: that the conventional wisdom is always wrong, and that you can achieve any goal through sheer force of will. That’s a dangerous combo.

I really do think that for quite a while, Musk and Karpathy actually did think that they were right, the conventional wisdom was wrong, and everyone just needed to work harder to get FSD. I also do think that now they have to know that they’re (along with everyone else, to be fair, that’s hawking self-driving) way further away than they thought they were, and now they just raise the price and recognize revenue every time they release a feature that was sold under the “full-self driving” banner, even if it doesn’t actually work right in the real world.

@bhtooefr @jkb That sounds like a pretty spot-on assessment to me. I watched video of a talk Howard Schultz gave at a Facebook sales conference. He said one thing that really stuck with me: in the early days, when Starbucks was growing rapidly, it seemed like they could do nothing wrong. And then as their growth leveled off, it seemed like they could do nothing right. He said "Neither of those things was true."

...

@bhtooefr @jkb Early success can be a curse, because as humans we tend to over-attribute both success and failure to our own intrinsic qualities, when a huge fraction of both is just happenstance. Even if Tesla was following Musk's original plan all along, it didn't work out because he's particularly brilliant but mostly by accident. There's only so much a human can even know.

@bhtooefr @jkb At any rate, I think Tesla has probably pushed us at least 5-10 years ahead of where we would have been down the road toward getting gasoline and diesel cars off the road. I think they've also advanced renewables just by making batteries cheaper, though not by as much since I think the main driver there has been photovoltaics.

With SpaceX, it could end up being anywhere from 10 to 50 years worth of advancement, just from embarrassing the government space programs.

@bhtooefr @jkb You know, I wonder if Musk really is interested in self-driving cars, or if he went down that path for financial reasons? It could well be that he realized that Tesla had a good chance of going under or at best getting acquired once the other car companies got their act together w.r.t. electrics? It seems like the "Loop" makes a lot more sense if that's the case.

@freakazoid @bhtooefr I think Musk is only interested in getting away from the masses by going to Mars with a select clique and an army of indentured servants, and every one of his current business ventures is actually prototyping technology to achieve this one goal (with public funding whenever possible, because libertarians are a strange breed)

@jkb @freakazoid @bhtooefr "Soon, the largest obstacle to reign in Musk will be the distance to Mars. Will it really be worth launching a billion-dollar interplanetary mission to make an arrest? Mars is several months away at its closest. It will be prohibitively expensive to reign Musk in after the fact." moderndiplomacy.eu/2021/05/15/

@angdraug @jkb @bhtooefr I had somehow missed the sovereignty aspect. But it's sort of irrelevant, because Mars will take a long time to become self-sufficient. Without regular resupply missions everyone there will simply die. It's impossible to be politically independent from someone you depend on for your very survival.

@angdraug @jkb @bhtooefr Musk may think he can play the US and China off against one another, but that would be a mistake; China is going to be too busy trying to keep its own people from rioting in the streets to even be thinking about the Moon, much less Mars. They are pretty close to running out of the low-hanging fruit they've been living off of from what I can tell.

Show newer

@freakazoid @jkb It is an awkward disconnect where he sells Teslas as enthusiast cars on driving experience, and then turns around and pushes self-driving.

Then again, I do actually think he’s personally offended by public transit, and that is what lead to the Loops. Pretty sure he’s even admitted to personally wanting the car tunnels to make his commutes shorter. (And, induced demand is the newest form of “conventional wisdom” (tbh it isn’t even accepted as conventional wisdom yet - conventional wisdom in North America is still that you can build lanes to alleviate congestion) that explains why his idea won’t work, so he disregards it.)

@freakazoid @jkb Also one thing about the claimed economics of FSD with huge numbers thrown around about the value of the cars: they only work if nobody else figures it out.

Basically, let’s say FSD actually does work, and Tesla actually does robotaxis: they can charge just under what Uber, Lyft, and local taxi services charge.

Let’s say Uber or Google actually get their own robotaxi services working: the moment they do that, there’s a price war and all of the profit margin goes away.

@bhtooefr @jkb My guess is that they'll just make a back room deal to divide up the market. Like Google's deal with Apple and Facebook (and I think some others, maybe Oracle?) not to poach employees from one another.

@bhtooefr @jkb Morgantown PRT sounds interesting.

I do additionally suspect two fat guys can climb a lamp-post? I mean, you could put little suspension bridges on something like that. Ideally pods could somehow hang from it use software to run only one pod at the same time. Even with that distance between them, they don't need to go very fast to carry a lot of cargo.

Other than the hanging part got the idea from lowtechmagazine.com/2008/02/a- (ok also the hanging part lowtechmagazine.com/2011/01/ae )

@bhtooefr @jkb suppose that's a very visual a but also a little dumb way to argue you could suspend a system from poles.. (i am not sure of the actual practicality of that..)

@jkb they had to focus on the most important features of the software first, like ensuring that you could buy the upgrade from your dashboard screen and it would just debit you twelve thousand dollars with no confirmation prompt

@robotcarsley Let's not forget the ability to use a fart sound or a goat call as your horn, which is extremely important and will not at all introduce confusion and chaos in situations where no one can afford any confusion or chaos

@jkb that's a good feature to have, assuming that its purpose is to report to the government, who will then immediately suspend your license as you've proven that you are too irresponsible to drive a car

@jkb @vfrmedia Some might call that “marketing” and I partially blame the public for falling for it. People let themselves be thick.

@jkb @swansinflight

you get those options (following the road lines and avoiding potential collisions) in normal cars here in Europe, for a much lower price - I could even have them as an option on my VW Polo!

Of course you still have to drive the car, but I have absolutely 0 interest in self driving cars - it took a lot of spoons for me to learn to drive which I did fairly recently but its something I enjoy doing and a skill I want to always improve...

@vfrmedia @swansinflight

I have active collision avoidance (it starts by an alert and then actually activates the brakes), but I stick to lane alerts instead of automatic lane following as these sensors get confused by repaired cracks in the road surface.

As a professional software developer I am terrified by the notion of roads full of cars driven by software, because I know how software quality degrades when there is a profit motive. More so when it depends on tech that has yet to appear

@jkb @swansinflight

I don't think I've ever seen anyone here order a car with active lane following; the state of the markings and general road surface in Britain is far too poor for this to ever work (also we have spiral roundabouts which are sure to confuse such a system).

I don't think any traffic authority in whole EU+UK is going to allow 100% self driving any time soon, even if they are allowing tests of the tech in limited areas..

@vfrmedia @swansinflight Same situation here in Belgium, especially in rural areas such as where I live where many roads don't have markings at all and when they have they're covered in mud half of the year.

As far as I know most self-driving tests ongoing in the EU are about convoys for bulk transportation or public transportation where the leading vehicle actually has a human driver. If I remember correctly Mercedes was talking about this kind of stuff ten years ago already

Sign in to participate in the conversation
Octodon

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!