Elon Musk is a nice chap

[nctnico]

@eti: you are missing the big picture of my remark and that is that new technology improves slowly but steadily. If you look at a modern car you'll see it has all kinds of electronic safety related features and no consumer cares about whether they are electronic and what kind of decissions are build into them.

[Zero999]

Yes driver assist will gradually become more compehensive and common, but I doubt fully self-driving will ever take of, due to the liability reasons mentioned above. Self driving cars will be more submissive and less agressive than human drivers, which will cause other road users to take the piss. If a cyclist knows the self-driving car will stop, if they pull out infront of them, without beeping the horn and swearing at them, they'll be more likely to do it. I can imagine a situation with a fully self-driving car stopped for ages, whist pedestrians and cyclists keep crossing the road in front of it, because they know it'll wait for them.

Not only that but how many teenagers will turn trolling self driving cars into a game? That's something I could see my youthful self doing, standing in a place that makes the car stop or choose a path that ends in a dead end, or putting up something the car will mistake for a stop sign along a busy arterial. There aren't enough (any) self driving cars around yet for this to be a problem but if they become mainstream so will these sort of incidents.

If you ever want to have a perfect example of a strawman argument, then you can use what you typed above. With the same line of reasoning you could say EVs never become mainstream because youngsters disconnect the charging cables and shove potatos in the charger outlets at night.

Fair enough he's taking it to the extreme, but why do you think the fact that most road users often break the rules, but a self-driving cars will follow the rules all the time, won't cause any problems? It will be fine if there are places where only self-driving cars are allowed, but I can see it being a huge problem, when they're mixed with unruly road users.

I guess you are referring to keeping to the speed limit. First of all it seems Waymo's algorithms allow to go over the speed limit because driving slower than the other traffic -even though at the speed limit- is very dangerous. Secondly the EU has plans mandate cars to have active speed limiting on all cars sold within a few years. In the end self driving car algorithms will need to deal with other road users. Waymo et all probably have collected tons of data for that purpose.

No, not just the speed limit. There are plenty of other occasions when other road users break the rules and one often has to do the same, otherwise the journey will take much longer than necessary. Other road users will also often let you pass, even if they have priority.

[nctnico]

I'm not convinced such situations are a such a big problem that self driving cars are useless. There are too many variables involved to make an accurate prediction of what will happen in the real world, what regulations will be needed, changes in driver's attitudes, etc. Time will tell.

[bson]

As an autonomous autopilot yes, but certainly not as a driver assist.  ....    It's also IMO a whole lot more immediately useful than cars that drive themselves, that's just a novelty.

I agree and so do some car companies, but you can't sell those systems for $10,000 and people won't show them off on social media, so no free advertising.

But the government can mandate that all new cars sold have them.  No need for expansive social media campaigns or placing media narratives - just the stroke of a pen.

[eti]

@eti: you are missing the big picture of my remark and that is that new technology improves slowly but steadily. If you look at a modern car you'll see it has all kinds of electronic safety related features and no consumer cares about whether they are electronic and what kind of decissions are build into them.

I think society has descended to a new dumb when people trust CODE over human judgement and experience. No bytes are gonna better a human driver, even a terrible one. This reminds me of the so-called “AI” which powers Siri, Amazon Echo, Google assistant etc … yeah, no way this junk is ANYWHERE NEAR even approximating even the stupidest person that ever existed.

I have all the aforementioned “assistants” in my house, and not one of them can go 24 hours without falling flat on their faces at one or other trivial request. Deep delusion and Silicon Valley’s OBSESSION with having to act out some childish space age sci-fi future, forcing it to come to pass, is what’s happening. Silicon Valley kiddies.

[Zero999]

@eti: you are missing the big picture of my remark and that is that new technology improves slowly but steadily. If you look at a modern car you'll see it has all kinds of electronic safety related features and no consumer cares about whether they are electronic and what kind of decissions are build into them.

I think society has descended to a new dumb when people trust CODE over human judgement and experience. No bytes are gonna better a human driver, even a terrible one. This reminds me of the so-called “AI” which powers Siri, Amazon Echo, Google assistant etc … yeah, no way this junk is ANYWHERE NEAR even approximating even the stupidest person that ever existed.

I have all the aforementioned “assistants” in my house, and not one of them can go 24 hours without falling flat on their faces at one or other trivial request. Deep delusion and Silicon Valley’s OBSESSION with having to act out some childish space age sci-fi future, forcing it to come to pass, is what’s happening. Silicon Valley kiddies.

Not much intelligence is required to drive a car, under most conditions. A computer doesn't get tired, so lots of accidents caused by people falling asleep at the wheel can be avoided.

[Marco]

Not much intelligence is still infinitely beyond AI.

[james_s]

Not much intelligence is required to drive a car, under most conditions. A computer doesn't get tired, so lots of accidents caused by people falling asleep at the wheel can be avoided.

"Under most conditions" is the key here. Unfortunately encountering a situation that falls outside of "most conditions" is not a rare occurrence and it is those edge cases which turn up all over that are the the challenge that self-driving proponents tend to overlook. I mean let's take a really easy example and consider snow. It doesn't happen often here, 99.9% of the time when I'm driving I don't even have to think about snow, yet I have been caught in it a few times when it started snowing and by the time I was nearing home the roads were dusted with enough that everything was white and starting to get slippery. A system like Tesla uses that relies entirely on cameras is not going to know what to do when the entire road is bright white, not to mention the slippery surface that makes driving a completely different experience.

If computers were any good at this sort of thing captchas would be useless. It took a decade or more before computers got to where they could even identify a spam email with any reasonable degree of accuracy and that's something that is usually trivial for even a below average intelligence human. Driving is similar, mechanically it is not a particularly difficult task, but there is a lot of nuance. I still remember when I was a teenager taking drivers ed, my instructor remarked that driving is primarily a social interaction, the physical part of controlling a car is easy but a large part of being a good driver is predicting what other drivers will do and responding accordingly. Anybody who has ever tried chatting with a bot knows that computers are notoriously awful at social interaction.

[Zero999]

Not much intelligence is required to drive a car, under most conditions. A computer doesn't get tired, so lots of accidents caused by people falling asleep at the wheel can be avoided.

"Under most conditions" is the key here. Unfortunately encountering a situation that falls outside of "most conditions" is not a rare occurrence and it is those edge cases which turn up all over that are the the challenge that self-driving proponents tend to overlook. I mean let's take a really easy example and consider snow. It doesn't happen often here, 99.9% of the time when I'm driving I don't even have to think about snow, yet I have been caught in it a few times when it started snowing and by the time I was nearing home the roads were dusted with enough that everything was white and starting to get slippery. A system like Tesla uses that relies entirely on cameras is not going to know what to do when the entire road is bright white, not to mention the slippery surface that makes driving a completely different experience.

If computers were any good at this sort of thing captchas would be useless. It took a decade or more before computers got to where they could even identify a spam email with any reasonable degree of accuracy and that's something that is usually trivial for even a below average intelligence human. Driving is similar, mechanically it is not a particularly difficult task, but there is a lot of nuance. I still remember when I was a teenager taking drivers ed, my instructor remarked that driving is primarily a social interaction, the physical part of controlling a car is easy but a large part of being a good driver is predicting what other drivers will do and responding accordingly. Anybody who has ever tried chatting with a bot knows that computers are notoriously awful at social interaction.

Snow is hardly rare. It's pretty common in some parts of the world, so I would expect any self-driving car would be able to cope with it.

I do agree though. Although I would like a fully self-driving car, as I said in my first reply there are plenty of reasons why it won't happen.

[Jan Audio]

The car looks to the white lines, snow is a big line for the car.
Dust can blind the car also.

Normally the russians always had manual control in any system, while the american stuff you cant control manually.
I wonder if russians still use this rule.

Just take a look at the boeing 737 max, it dont drive in any road, it only has to stay in the air, simpler cant be possible.
While a russian airplane would not have crashed.

Simple sayd : its stupidity, humans are test subjects for technology ?, like the several corona vaccins they are testing.

[james_s]

Just take a look at the boeing 737 max, it dont drive in any road, it only has to stay in the air, simpler cant be possible.
While a russian airplane would not have crashed.

I'm not sure if that's a comparison I'd be using. Things may be different now, but for a while Russian airplanes did not have good safety records at all. It's true that much of this was down to maintenance but there were multiple fatal crashes caused by engines assembled at the factory with bearings that did not have a full complement of rollers. There have also been numerous other crashes due to design flaws. Given the choice of flying on a randomly chosen Boeing aircraft over say a randomly chosen Tupolev aircraft I'd choose the Boeing every time.

[eti]

Not much intelligence is still infinitely beyond AI.

I regularly question HUMAN intelligence, never mind AI 

[Jan Audio]

Ofcourse the russian plane would have crashed for another reason.
What i mean is, you should know, if it was manually flyable, it would not have crashed.
You know what i mean.

We are talking about A.I., not airplanes.
Please dont act stupid.

[james_s]

The Boeing plane was manually flyable too, MCAS was easily switched off. The issue is that pilots were not taught that it even existed.

[wizard69]

The problem with these things is that they are massive projects with no financial benefit while costing a lot to even run, let alone build. The reason most people are not driving electric cars (including me) is because an internal combustion car is cheaper over the cars whole lifecycle.

This is why I don't have an electric car, well truck, right now.   The entry price for a manufactured car is just very high even if the operational costs are lower.   It isn't like I wouldn't want to have one as I've been interested since the 1980's.    I even considered doing a DIY truck back in the 1980's.    The problem back then was more about the batteries than anything making a lead acid powered truck basically a summer car.

However I really see the equations turning in favor of electric products pretty rapidly.   If Tesla is successful with its "pickup" (really an SUV in my opinion) it will be very cost competitive.    In part that is due to Detroit grossly over charging for the pickup line.    But Tesla has certainly attacked the maneuverability problem from many directions which should change the electric cost structure over the next 2-3 years.

Well at least that is my hope; Tesla selling every car they make, months ahead of time, doesn't give them a lot of incentive to lower prices.   In any event an affordable electric automobile has been a life long dream.    The surprise is with retirement not far away at all it looks like this could be real and cost effective.

Musk has already made SpaceX turn profits on this whole rocketry shenanigans. There are costumers that pay big money to get there commercial satellite up there into orbit, or to deliver stuff to the space station. The reason they pick SpaceX to do it is because they developed the latest in rocket technology to make it more cost effective. They also use this new capability by launching a better satellite internet service called Starlink.

Musk certainly has upset the world with just about every company he has his fingers in.

So they keep developing there rocket tech to also become the first commercial provider of interplanetary transport, and the most attractive first step is Mars.

Musk's hope is for humanity to escape an extinction event on earth.   It is actually a good idea considering there is nothing we could do to survive some of the potential extinction events that could happen.    In this regard he is no delusional at all.

However establishing humanity on Mars will not be anywhere near as simple as he thinks.   does that mean he is delusional?   I can't really say but I do know that having all your eggs in one basket is never a good thing.    Sadly a lot of people will likely die on Mars before we have a self sustaining settlement.   From what I've seen he is not even delusional about that.

It's only governments that can launch and run a non profitable project that benefits humankind as a whole. Commercial companies can't do that because they would eventually run out of money, at what point the banks stop giving paychecks to the people that work there and those people stop working for you.

I would also say it is almost impossible for government to find a non profitable project that they can invest in an not screw up.   About the only one we can really say was such a project was the Apollo program.    I'd also have to say the Manhattan project similarly benefited humans as it made contemplation of nuclear war pretty horrible for most of the big actors.   Sadly we now have far too many sub standard governments, run by the mentally ill, that don't see nuclear war as evil.

[wizard69]

How do you make advanced things like chips or solar cells from scratch for example.

You can't even make basic stuff required to keep a civilisatiosn going, let alone advanced stuff.
People extoling this self sustaining colony stuff have not given one through to the insanely complex supply chain structure we have on earth to produce even basic stuff, let alone advanced materials and engineering.
I've put a figure on this before on The Amp Hour, and it was something like 500 years before we have a colony even cabale of surviving on it's own should something happen to earth or transport.
i.e. a figure so large that it basically means you can't predict when such a thing would exist.

all you really need on Mars, to quickly establish a colony, is to find a large iron deposit with the materials to smelt it and a nuclear power solution.    Once you can create steel and cast iron you can quickly build the infrastructure to make just about every thing else possible.   Things will be pretty rough before that happens.

Those reactors will also be important for food production so any settlement needs to be a close to thorium, uranium or other element suitable for a reactor.   In a nut shell a settlement needs two things to get beyond temporary, that is energy and steel.

[james_s]

all you really need on Mars, to quickly establish a colony, is to find a large iron deposit with the materials to smelt it and a nuclear power solution.    Once you can create steel and cast iron you can quickly build the infrastructure to make just about every thing else possible.   Things will be pretty rough before that happens.

Those reactors will also be important for food production so any settlement needs to be a close to thorium, uranium or other element suitable for a reactor.   In a nut shell a settlement needs two things to get beyond temporary, that is energy and steel.

You make it sound simple, but in practice I suspect it will be fraught with problems, and what can go wrong will, Murphy will get you even on another planet. What happens when the reactor itself malfunctions or becomes damaged? I'm confident that a self sufficient colony on Mars is monumentally more difficult than a lot of people assume. Build a colony somewhere on the bottom of the ocean or in the middle of the Siberian tundra as a proof of concept. The most inhospitable places on Earth are vastly more livable than anywhere on Mars.

[james_s]

I would also say it is almost impossible for government to find a non profitable project that they can invest in an not screw up.   About the only one we can really say was such a project was the Apollo program.    I'd also have to say the Manhattan project similarly benefited humans as it made contemplation of nuclear war pretty horrible for most of the big actors.   Sadly we now have far too many sub standard governments, run by the mentally ill, that don't see nuclear war as evil.

It seems a lot more difficult now than it used to be, but there have been many others. The freeways are a good example, originally billed as "defense highways" they built vast highways criss-crossing the nation in an astonishingly short period of time. The hydroelectric dams were another government project that has been very successful, originally built to supply the vast amounts of electricity needed to refine aluminum during WWII they have continued to generate clean renewable energy for the Northwest region. Those are just the projects I can think of, I know there are others.

IMO the reason so many government projects become such messes is because the way the system works you always end up with "design by committee", nobody agrees on the direction a project should take so it is compromise after compromise trying to please everyone and it ends up pleasing no one. While not something I'm generally a fan of, there is something to be said for the efficiency of having one person or organization with the power to unilaterally make decisions and get things done without having to compromise to try to please everyone.

[sandalcandal]

Semi-technical/recruitment presentation going over Tesla's Autonomous Driving Technology Strategy at a computer vision (virtual) conference just yesterday. Presented by Tesla's AI director

https://youtu.be/eOL_rCK59ZI?t=28292
Lots of other talks by other industry players if you're interested too.

[EEVblog]

In a nut shell a settlement needs two things to get beyond temporary, that is energy and steel.

Steel doesn't help you eat.
The worlds largest biodome experiment could barely even feed a handful of people using 40 acres.
https://en.wikipedia.org/wiki/Biosphere_2

[sandalcandal]

In a nut shell a settlement needs two things to get beyond temporary, that is energy and steel.

Steel doesn't help you eat.
The worlds largest biodome experiment could barely even feed a handful of people using 40 acres.
https://en.wikipedia.org/wiki/Biosphere_2

That experiment is from over 30 years ago, are there more current experiments? Particularly considering significant advances in solar energy and vertical farming.

Edit1: The approach of trying to recreate "natural" Earth environments is questionable too. Particularly if the goal is to be as productive as possible in limited space as your comment suggests is a barrier to self-sufficiency.

Edit2: These more recent projects seem to have none looking at self-sufficiency, just research labs. https://en.wikipedia.org/wiki/Mars_analog_habitat

Edit3: The "40 acres" also seems to refer to the total grounds area of the facility, not the area enclosed in the biodomes which totals 3.14 acres! Did you actually read this or are you sharing a point you heard from somewhere else?  If so can you give the source you got this info from?
Edit 4: The agricultural part of the experiment was apparently only 0.6 acres! (2500 m^2) according to the Wikipedia article you linked.

[vk6zgo]

In a nut shell a settlement needs two things to get beyond temporary, that is energy and steel.

Steel doesn't help you eat.
The worlds largest biodome experiment could barely even feed a handful of people using 40 acres.
https://en.wikipedia.org/wiki/Biosphere_2

On earth, it does------steel plowshares increase the productivity of people using manual or ox drawn
plows.

[Psi]

I'm confident that a self sufficient colony on Mars is monumentally more difficult than a lot of people assume. Build a colony somewhere on the bottom of the ocean or in the middle of the Siberian tundra as a proof of concept. The most inhospitable places on Earth are vastly more livable than anywhere on Mars.

I wouldn't say it's all that "difficult", just extremely time consuming.

True, there probably is a small amount of new science that needs to be discovered/invented to deal with the fact that it's another planet with different environmental processes going on.
But 99% of the stuff you need to know to create a self sufficient mars colony is known science, stuff we know from doing it on earth.

You simply need to do it again, all of it.
Given the current cost to get stuff to mars it will take an extremely long time.

That said, we don't know what new tech will appear in the next 100 years.   
If SpaceX (or some other company) does get space travel to the point where it's the same as current passenger jets in terms of reliability and reusability that is a shit ton of mass you can put on mars.

And if we got some sort of nano-assemblers, something where you can fabricate new items, including a new fabricator machine, and all you need is energy input + raw elemental matter. Then that would make it much much easier.
You would no longer need 10000 different specialized industries to make tools & spare parts on mars

[jonovid]

Elon Musk is a marketing genius 

just take one good look into the unproven transportation system of hyperloop, shows after 6 Years of research and development, this is about marketing an idea. not engineering a working proof of concept.
no feasible working hyperloop exists today . no track or tube longer than a mile or kilometer.
thow Boston Dynamics is working had to change this, in 2021 the age of autonomous reproducing robots is sill science fiction. so is the idea of artificial government intelligence.  thinking software .
government is still run by gullible humans that can be bribed or bought off, even the office of president is not immune,  by 1998 Beijing got givan the knowledge to send men into space or was it to send Lewinsky under the then president's table.   

[EEVblog]

Elon Musk is a marketing genius 

just take one good look into the unproven transportation system of hyperloop, shows after 6 Years of research and development, this is about marketing an idea. not engineering a working proof of concept.

And people still don't understnd that Elon Musk is not working on a Hyperloop, and he never really has. He has not put any serious engineering resources behind it apart from from that student pod competition. The last thing Musk did was the whitepaper in 2013.
Yet people still confuse the Boring company with Hyperloop, they ditched the Hyperloop reduced pressure tube idea years ago, which is the entire concept of hyperloop. And that's another mistake people make, once you take away the lo pressure tube, it's not Hyperloop any more.