8 ms·

Rockets may be easier to develop than a product for the mass market. Car manufacturers have had many decades of optimizations of their processes. It’s very hard

Rockets may be easier to develop than a product for the mass market. Car manufacturers have had many decades of optimizations of their processes. It’s very hard to catch up.

Causality1 8y ago
It would be easier if he'd stop making grandiose claims like "self driving is only two years away!"
- FireBeyond 8y ago
  Two years away? Just in mid-February he said "You'll be in FSD, coast to coast, in of our cars, _this year_".
  - maxxxxx 8y ago
    These comments are weird. Everybody is stretching the truth to some degree but these claims are pretty bold.
    - natch 8y ago
      The claims sound bold because he never made them. The poster is misquoting. Elon said that Tesla would have a version of the software that could do full self driving by the end of this year. Then he very clearly spelled out that it would not be available to consumers until regulators approve it at some later date (possibly years).
      - tim333 8y ago
        I tried googling and found "Elon Musk promises fully autonomous, coast-to-coast Tesla road trip in the next 7 months" in Feb 2018, not this year. Obv didn't happen. https://www.bizjournals.com/sanjose/news/2018/02/08/tesla-fully-autonomous-roadtrip-musk-tsla.html https://www.bizjournals.com/sanjose/news/2018/02/08/tesla-fu...
        natch 8y ago
        He also stated that advanced summon would be rolled out last week, but then it didn’t happen. Why? We (the general public) don’t know exactly but rumors are that some issues were uncovered and Tesla decided to slow down the rollout of this feature until the issues are fixed. That’s good that they slowed down the rollout. I’d rather wait for a safer, better version of the feature than get a dangerous version, if that’s part of the issue. There is absolutely nothing wrong with this picture except for people interpreting forward-looking statements as 100% locked-in promises. They are not promises. In technology, people predict stuff all the time and sometimes the predictions come true on time; other times they take longer. Perfectly normal stuff but outsiders don’t always get this. Elon has correctly predicted plenty of stuff that has come to pass. Not always on time, but it is a given that schedules can slip in technology.
  - Dylan16807 8y ago
    On the other hand "coast to coast" probably just means "freeway, under normal conditions" which is the lowest of bars to clear. So if they can make it capable of detecting stationary objects they're basically already there. Still amazingly far from "full" self driving.
  - natch 8y ago
    He said no such thing. If you believe otherwise, please provide a link to back up your claim. He was saying only that the internal development versions of the software would be able to do that by then.
bobsil1 8y ago
Cars-as-software demand software-centric co's hiring ML people. Legacy won't catch up, startups will.
- maxxxxx 8y ago
  Dream on. Cars aren’t software.
  - bobsil1 8y ago
    Nope. Car differentiators are going to be tablet screen + special-purpose AI. Dumb cars = vinyl.
    - maxxxxx 8y ago
      A car is a thing to get from A to B in a reliable manner. Screens and AI are secondary.
      - bobsil1 8y ago
        “A phone is for voice calls.”
        maxxxxx 8y ago
        Current smartphones are an evolution of the PDA idea that started with the Apple Newton, Palm Pilot and others. People always dreamed of a handheld device that would enable them to communicate and get information. What do you expect a car to do other than carting people around? Once it can go autonomously from A to B what else will people use a car for?
- phkahler 8y ago
  There is a lot of software in cars, but cars are not software. Making cars is a lot harder from a business perspective than doing AI. In the end software is just a file that gets loaded into a module - one step in a very long chain.
cperciva 8y ago
The bar for success is different too. A rocket which safely delivers its cargo 99.99% of the time would be a huge step forward. A car which safely delivers its cargo 99.99% of the time would be taken off the roads.
- stellar678 8y ago
  It’s funny you mention 99.99% - because 40,000 people are killed by car crashes in a year in the United States, which is actually .012% of the population. So cars actually do a bit worse than your stat...
  - maxxxxx 8y ago
    If a majority of the population used a rocket daily then the numbers would be different. The number of people killed per space flight is probably a little higher than the number if people killed per drive in a car.
  - cperciva 8y ago
    Per trip, not per year.
  - iamaelephant 8y ago
    Yes this makes perfect sense if every person in USA took one car journey in a year.
the_duke 8y ago
Agreed. Can't find it now, but Musk said something like "For a rocket, manufacturing is 10 times harder than building the actual rocket. For cars, it's 100 times harder". This is not that surprising, the incumbent manufacturers have decades of experience, vast amounts of know how and optimized supply chains. There's a reason why no new manufacturers have emerged.

Rockets may be easier to develop than a product for the mass market. Car manufacturers have had many decades of optimizations of their processes. It’s very hard to catch up.

Causality1 8y ago

It would be easier if he'd stop making grandiose claims like "self driving is only two years away!"

FireBeyond 8y ago
Two years away? Just in mid-February he said "You'll be in FSD, coast to coast, in of our cars, _this year_".
- maxxxxx 8y ago
  These comments are weird. Everybody is stretching the truth to some degree but these claims are pretty bold.
  - natch 8y ago
    The claims sound bold because he never made them. The poster is misquoting. Elon said that Tesla would have a version of the software that could do full self driving by the end of this year. Then he very clearly spelled out that it would not be available to consumers until regulators approve it at some later date (possibly years).
    - tim333 8y ago
      I tried googling and found "Elon Musk promises fully autonomous, coast-to-coast Tesla road trip in the next 7 months" in Feb 2018, not this year. Obv didn't happen. https://www.bizjournals.com/sanjose/news/2018/02/08/tesla-fully-autonomous-roadtrip-musk-tsla.html https://www.bizjournals.com/sanjose/news/2018/02/08/tesla-fu...
      - natch 8y ago
        He also stated that advanced summon would be rolled out last week, but then it didn’t happen. Why? We (the general public) don’t know exactly but rumors are that some issues were uncovered and Tesla decided to slow down the rollout of this feature until the issues are fixed. That’s good that they slowed down the rollout. I’d rather wait for a safer, better version of the feature than get a dangerous version, if that’s part of the issue. There is absolutely nothing wrong with this picture except for people interpreting forward-looking statements as 100% locked-in promises. They are not promises. In technology, people predict stuff all the time and sometimes the predictions come true on time; other times they take longer. Perfectly normal stuff but outsiders don’t always get this. Elon has correctly predicted plenty of stuff that has come to pass. Not always on time, but it is a given that schedules can slip in technology.
- Dylan16807 8y ago
  On the other hand "coast to coast" probably just means "freeway, under normal conditions" which is the lowest of bars to clear. So if they can make it capable of detecting stationary objects they're basically already there. Still amazingly far from "full" self driving.
- natch 8y ago
  He said no such thing. If you believe otherwise, please provide a link to back up your claim. He was saying only that the internal development versions of the software would be able to do that by then.

bobsil1 8y ago

Cars-as-software demand software-centric co's hiring ML people. Legacy won't catch up, startups will.

maxxxxx 8y ago
Dream on. Cars aren’t software.
- bobsil1 8y ago
  Nope. Car differentiators are going to be tablet screen + special-purpose AI. Dumb cars = vinyl.
  - maxxxxx 8y ago
    A car is a thing to get from A to B in a reliable manner. Screens and AI are secondary.
    - bobsil1 8y ago
      “A phone is for voice calls.”
      - maxxxxx 8y ago
        Current smartphones are an evolution of the PDA idea that started with the Apple Newton, Palm Pilot and others. People always dreamed of a handheld device that would enable them to communicate and get information. What do you expect a car to do other than carting people around? Once it can go autonomously from A to B what else will people use a car for?
phkahler 8y ago
There is a lot of software in cars, but cars are not software. Making cars is a lot harder from a business perspective than doing AI. In the end software is just a file that gets loaded into a module - one step in a very long chain.

cperciva 8y ago

The bar for success is different too. A rocket which safely delivers its cargo 99.99% of the time would be a huge step forward. A car which safely delivers its cargo 99.99% of the time would be taken off the roads.

stellar678 8y ago
It’s funny you mention 99.99% - because 40,000 people are killed by car crashes in a year in the United States, which is actually .012% of the population. So cars actually do a bit worse than your stat...
- maxxxxx 8y ago
  If a majority of the population used a rocket daily then the numbers would be different. The number of people killed per space flight is probably a little higher than the number if people killed per drive in a car.
- cperciva 8y ago
  Per trip, not per year.
- iamaelephant 8y ago
  Yes this makes perfect sense if every person in USA took one car journey in a year.

the_duke 8y ago

Agreed. Can't find it now, but Musk said something like "For a rocket, manufacturing is 10 times harder than building the actual rocket. For cars, it's 100 times harder". This is not that surprising, the incumbent manufacturers have decades of experience, vast amounts of know how and optimized supply chains. There's a reason why no new manufacturers have emerged.