The Lemon Fool

odysseus2000 wrote:

Lootman wrote:

servodude wrote:Even if your system is 200x better than humans driving... every mess up falls at your feet. That PR will be hard to wear.

Yes, and in fact it doesn't really matter if driverless cars are 200 times or even 2,000 times safer. The first time a driverless car mows down a schoolchild on a zebra crossing, it will set back driverless cars by years.

The people who don't die because of driverless cars won't know it. It is the families of the ones who did die that will kick off.

I don’t think it will work this way at least in the US.

There have been an horrendous number of gun mass murders. Most could be stopped if guns were banned, but despite all the rhetoric it never happens.

Once fsd launches the number of car accidents plummets & the odd disaster becomes like guns, too popular with too many to ban it.

The difference is that guns are already in use. It is getting rid of them that would be the change.

Whereas we do not currently use driverless cars. The change would be people using them.

Inertia always operates against change. And each "accident" that happens, like the Waymo (Google) car that hit a pedestrian on a pedestrian crossing in San Francisco a few weeks ago, will add to the resistance to change.

Those car accidents that you claim will "plummet" cannot do that prior to adoption and legalisation. Until then it is just a claim.

odysseus2000 wrote:

Midsmartin wrote:This is interesting. Someone on Twitter has been counting how many self driving disengagements they get with each version of their Tesla fsd. It's getting better quite quickly.

https://twitter.com/eliasmrtnz1/status/ ... 2Yj3g&s=19

Yes this is the way all the Monte Carlo codes I have written have behaved. All kinds of failures at the start, then slowly the number of failures reduces & then plummets.

The more I watch of US Tesla fsd videos the more I swing to thinking it is going to work.

If fsd works I can easily see stock 5x from where ever it was the day before.

Regards,

Ody, you were forecasting it would be "working soon" years ago. And convinced driverless taxis would be on the streets in 2020.

In the real world, in cities, it's making slow progress.

regards

Howard

Lootman wrote:

odysseus2000 wrote:

Lootman wrote:

servodude wrote:Even if your system is 200x better than humans driving... every mess up falls at your feet. That PR will be hard to wear.

Yes, and in fact it doesn't really matter if driverless cars are 200 times or even 2,000 times safer. The first time a driverless car mows down a schoolchild on a zebra crossing, it will set back driverless cars by years.

The people who don't die because of driverless cars won't know it. It is the families of the ones who did die that will kick off.

I don’t think it will work this way at least in the US.

There have been an horrendous number of gun mass murders. Most could be stopped if guns were banned, but despite all the rhetoric it never happens.

Once fsd launches the number of car accidents plummets & the odd disaster becomes like guns, too popular with too many to ban it.

The difference is that guns are already in use. It is getting rid of them that would be the change.

Whereas we do not currently use driverless cars. The change would be people using them.

Inertia always operates against change. And each "accident" that happens, like the Waymo (Google) car that hit a pedestrian on a pedestrian crossing in San Francisco a few weeks ago, will add to the resistance to change.

Those car accidents that you claim will "plummet" cannot do that prior to adoption and legalisation. Until then it is just a claim.

Yes, but in the UK there are 3000 deaths & serious injuries on the roads. You think a politician will stand in the way of reducing this carnage & financial cost?

Regards,

Howard wrote:

odysseus2000 wrote:

Midsmartin wrote:This is interesting. Someone on Twitter has been counting how many self driving disengagements they get with each version of their Tesla fsd. It's getting better quite quickly.

https://twitter.com/eliasmrtnz1/status/ ... 2Yj3g&s=19

Yes this is the way all the Monte Carlo codes I have written have behaved. All kinds of failures at the start, then slowly the number of failures reduces & then plummets.

The more I watch of US Tesla fsd videos the more I swing to thinking it is going to work.

If fsd works I can easily see stock 5x from where ever it was the day before.

Regards,

Ody, you were forecasting it would be "working soon" years ago. And convinced driverless taxis would be on the streets in 2020.

In the real world, in cities, it's making slow progress.

regards

Howard

It is making very fast progress, incredibly fast.

Not fast enough for me, but still incredibly fast.

Optimus will come very quickly, the potential for this product dwarfs all others in the world.

Regards,

Lootman wrote:Inertia always operates against change. And each "accident" that happens, like the Waymo (Google) car that hit a pedestrian on a pedestrian crossing in San Francisco a few weeks ago, will add to the resistance to change.

You mean the Waymo which was being driven in manual by a driver at the time of the incident?

odysseus2000 wrote:Yes, but in the UK there are 3000 deaths & serious injuries on the roads. You think a politician will stand in the way of reducing this carnage & financial cost?

Regards,

Does a politician want to swap a situation where 3000 people a year die and they receive no blame at all, for a situation where 2000 people a year die and their vote in favour of legalising the technology responsible is mentioned in every article about every death?

Hint:: The answer is no.

Autonomous driving will make a metaphorical impact on British roads, but the requirements to play will include an auditable decision making process, and not being run by a 13 year old whose favourite hobby is taunting regulatory bodies both of which exclude Tesla.

Correct me if I'm wrong but I thought the right to bear arms was in the US constitution. If I'm right guns are irrelevant to these predictions.

CliffEdge wrote:Correct me if I'm wrong but I thought the right to bear arms was in the US constitution. If I'm right guns are irrelevant to these predictions.

It is also in the constitution about free right to assemble, but when covid came..,

The US constitution can be changed via the Supreme Court & they have already sanctioned various restrictions to gun ownership.

Regards,

BobbyD wrote:

odysseus2000 wrote:Yes, but in the UK there are 3000 deaths & serious injuries on the roads. You think a politician will stand in the way of reducing this carnage & financial cost?

Regards,

Does a politician want to swap a situation where 3000 people a year die and they receive no blame at all, for a situation where 2000 people a year die and their vote in favour of legalising the technology responsible is mentioned in every article about every death?

Hint:: The answer is no.

Autonomous driving will make a metaphorical impact on British roads, but the requirements to play will include an auditable decision making process, and not being run by a 13 year old whose favourite hobby is taunting regulatory bodies both of which exclude Tesla.

After the first few no one will care as happens now. Road traffic accidents do not lead the news.

Regards,

BobbyD wrote:

Lootman wrote:Inertia always operates against change. And each "accident" that happens, like the Waymo (Google) car that hit a pedestrian on a pedestrian crossing in San Francisco a few weeks ago, will add to the resistance to change.

You mean the Waymo which was being driven in manual by a driver at the time of the incident?

So they claimed, as they do every time. Because, well, they would, wouldn't they?

Lootman wrote:

BobbyD wrote:

Lootman wrote:Inertia always operates against change. And each "accident" that happens, like the Waymo (Google) car that hit a pedestrian on a pedestrian crossing in San Francisco a few weeks ago, will add to the resistance to change.

You mean the Waymo which was being driven in manual by a driver at the time of the incident?

So they claimed, as they do every time. Because, well, they would, wouldn't they?

Well you would say that wouldn't you?

BobbyD wrote:

Lootman wrote:

BobbyD wrote:

Lootman wrote:Inertia always operates against change. And each "accident" that happens, like the Waymo (Google) car that hit a pedestrian on a pedestrian crossing in San Francisco a few weeks ago, will add to the resistance to change.

You mean the Waymo which was being driven in manual by a driver at the time of the incident?

So they claimed, as they do every time. Because, well, they would, wouldn't they?

Well you would say that wouldn't you?

As would you.

Lootman wrote:

servodude wrote:Even if your system is 200x better than humans driving... every mess up falls at your feet. That PR will be hard to wear.

Yes, and in fact it doesn't really matter if driverless cars are 200 times or even 2,000 times safer. The first time a driverless car mows down a schoolchild on a zebra crossing, it will set back driverless cars by years.

It depends on the attitude of the government. If they are keen to promote driverless cars, and if the accident was seen as a rare, fixable problem, then they will make allowances. The point will be made that driverless cars are 200 times (or whatever) safer, and most members of the public could probably grasp this.

odysseus2000 wrote:If the testing routine is good it should catch any serious errors before the neural net is sent to the wild.

Yes, that is what should happen.

It isn't. Tesla is releasing beta software out to the wild and letting the real world be the test.

murraypaul wrote:

odysseus2000 wrote:If the testing routine is good it should catch any serious errors before the neural net is sent to the wild.

Yes, that is what should happen.

It isn't. Tesla is releasing beta software out to the wild and letting the real world be the test.

Which at some point makes sense as then the cars encounter real world situations rather than just the artificial situations dreamt up by your testers.

Suspicious?

"Waymo filed a lawsuit last week against the California Department of Motor Vehicles (DMV) arguing that it should keep the company's crash data private."

"Autonomous vehicle companies continue to argue that crash data is not a helpful way of assessing the overall safety of their vehicles and software."

https://sfist.com/2022/01/28/waymo-suin ... ta-secret/

The whole driverless car fiction stinks.

odysseus2000 wrote:

Howard wrote:

odysseus2000 wrote:The Telegraph article is rubbish.

Musk on the last Friedman podcast described how much better the camera system they now have is better than a human eye.

Regards,

Have you driven a Tesla? If you have, you'll understand its limitations.

I have driven a new Tesla and its cameras couldn't see a broken down car on the motorway partially blocking the slow lane approximately a mile ahead. When it got much closer it did see the obstacle but much later than me and, if the motorway had been crowded, it would have caused a problem taking avoiding action.

A friend with a Tesla has observed similar issues.

regards

Howard

Unless you are in the US with the latest Fsd, one can't appreciate how much the tech has improved in the last 6 months.

The Telegraph article was about traffic lights. Sure this is a non trivial problem but comparing a public road Tesla to a mining truck makes no allowance for the quality of the Tesla AI neural net training. I have not seen any traffic light mistakes in the US.

Fsd is a rapidly expanding field, changing by the week & getting better in leaps & bounds.

Regards,

Ody. Can you justify your statement? Source?

Sadly Tesla doesn’t agree with you. It seems to agree with my experience actually driving the car.

The very latest car’s spec has cameras which Tesla state have a maximum range of 250 metres.

Even the ultrasonic sensors have a range of less than 500 metres.

It’s fun to dream but do facts get in the way?

Worth reading what Tesla claim for autopilot for their US cars in 2022.

regards

Howard

"Eight surround cameras provide 360 degrees of visibility around the car at up to 250 meters of range. Twelve updated ultrasonic sensors complement this vision, allowing for detection of both hard and soft objects at nearly twice the distance of the prior system."

https://www.tesla.com/autopilot

ursaminortaur wrote:

murraypaul wrote:

odysseus2000 wrote:If the testing routine is good it should catch any serious errors before the neural net is sent to the wild.

Yes, that is what should happen.

It isn't. Tesla is releasing beta software out to the wild and letting the real world be the test.

Which at some point makes sense as then the cars encounter real world situations rather than just the artificial situations dreamt up by your testers.

As an ex-software engineer I can tell y'all that it is axiomatic that no in-house testing procedures will catch all the bugs that real users in the real world will uncover and suffer. That's what beta testing is about (and even then...)

I worry about car manufacturers taking on liability. Remember the Pinto with its badly positioned gas tank. The committee in charge of recalls looked at the stats of rear-end fires and decided it was cheaper to pay out the injury lawsuits than implement a different design. They'd need strong regulation to handle going beyond the financial implications that insurance companies currently face to take on the legal ones us drivers do. And few lobby better than car companies.