|
Like you can look at this and just see very easily the action they would need to take to fix the traffic jam: https://www.youtube.com/watch?v=Suugn-p5C1M But that is basically never the action actual humans in cars on roads would take. Because of a quirk in human psychology that it doesn't seem like it'd fix things unless you look at it from outside.
|
#
?
Jan 17, 2017 21:35
|
|
|
|
| # ? May 8, 2024 22:58 |
|
|
i wonder what quirk in human psychology makes you completely impenetrable to the fact that congestion is simply more demand than supply and no amount of gadgets can change that
|
|
#
?
Jan 17, 2017 21:35
|
|
|
I wonder what quirk in human psychology causes threads like this one to go down the automated-car rabbit hole over and over instead of pursuing less circular topics.
|
|
#
?
Jan 17, 2017 21:40
|
|
|
Eletriarnation posted:I wonder what quirk in human psychology causes threads like this one to go down the automated-car rabbit hole over and over instead of pursuing less circular topics. instead of talking about things that will and do happen goons prefer to fantasize about what might happen this is why games is still the biggest subforum
|
|
#
?
Jan 17, 2017 21:41
|
|
|
boner confessor posted:i wonder what quirk in human psychology makes you completely impenetrable to the fact that congestion is simply more demand than supply and no amount of gadgets can change that Because the "supply" of a road is based on how people use it and there is certain things people will do forever in every traffic situation forever that are always a mistake every single time. Which causes congestion that would be avoidable if people did it right but can never be expected for people to just magically "do it right" after all this time across every culture showing people just do it wrong because that is what people do.
|
|
#
?
Jan 17, 2017 21:43
|
|
|
Owlofcreamcheese posted:Like you can look at this and just see very easily the action they would need to take to fix the traffic jam: This could easily be caused by AI cars with faulty sensors, differing precision, calibration issues and so on. Metrology still applies even if the car is automated. Solkanar512 fucked around with this message at 22:13 on Jan 17, 2017 |
|
#
?
Jan 17, 2017 21:51
|
|
|
Owlofcreamcheese posted:Like you can look at this and just see very easily the action they would need to take to fix the traffic jam:
|
|
#
?
Jan 17, 2017 21:58
|
|
|
A Buttery Pastry posted:If a pedestrian can step into the street at any point, then you're not going to be able to have the cars just tearing through the streets at high speeds, you'll have to settle for the cars distributing themselves in more efficient manner. we're pretty much talking about limited access high speed roadways here when OOCC is pretending it's possible to jam functionally infinite traffic on them so long as the ai drivers are maniacal enough
|
|
#
?
Jan 17, 2017 22:06
|
|
|
Owlofcreamcheese posted:Because the "supply" of a road is based on how people use it and there is certain things people will do forever in every traffic situation forever that are always a mistake every single time. Which causes congestion that would be avoidable if people did it right but can never be expected for people to just magically "do it right" after all this time across every culture showing people just do it wrong because that is what people do. No, the throughput of a road is based on various physical realities. If a car accident blocks half the lanes, traffic is going to slow to a crawl, because there is a physical bottleneck suddenly cutting the capacity of the road in half. Automation doesn't change that.
|
|
#
?
Jan 17, 2017 22:11
|
|
|
Main Paineframe posted:No, the throughput of a road is based on various physical realities. If a car accident blocks half the lanes, traffic is going to slow to a crawl, because there is a physical bottleneck suddenly cutting the capacity of the road in half. Automation doesn't change that. At least it can zipper merge!
|
|
#
?
Jan 17, 2017 22:16
|
|
|
Main Paineframe posted:No, the throughput of a road is based on various physical realities. If a car accident blocks half the lanes, traffic is going to slow to a crawl, because there is a physical bottleneck suddenly cutting the capacity of the road in half. Automation doesn't change that. Okay, but notice how that is literally not at all what happens? When one lane closes the traffic isn't cut by 50%, but instead the traffic all jams up and comes to a near standstill for miles for no apparent reason because the choices people make don't deal with things like merging very efficiently. Automation is what brings it to the "boop beep, 50% number of lanes means 50% throughput" that you are pretending is how it is now.
|
|
#
?
Jan 17, 2017 22:16
|
|
|
A Buttery Pastry posted:You keep calling it "a quirk of human psychology", but isn't it basically just down to sub-optimal communication? Which is where self-driving cars come in, since they can communicate in a manner that would allow them to act in unison to solve traffic jams or to prevent them from forming in the first place. Automated cars can theoretically drive closer together because they can all brake at the same time rather than serially braking when drivers see brake lights. Merging can be a lot more efficient, too. Owlofcreamcheese posted:Because the "supply" of a road is based on how people use it and there is certain things people will do forever in every traffic situation forever that are always a mistake every single time. Which causes congestion that would be avoidable if people did it right but can never be expected for people to just magically "do it right" after all this time across every culture showing people just do it wrong because that is what people do. Automated cars would also make it real easy to do demand-pricing for overcrowded roads. You could even charge cars for changing lanes.
|
|
#
?
Jan 17, 2017 22:27
|
|
|
Owlofcreamcheese posted:Because the "supply" of a road is based on how people use it and there is certain things people will do forever in every traffic situation forever that are always a mistake every single time. Which causes congestion that would be avoidable if people did it right but can never be expected for people to just magically "do it right" after all this time across every culture showing people just do it wrong because that is what people do. High capacity roads like highways are limited by low capacity pedestrian accessible surface streets. Have you ever driven to a location in a city or to a stadium and spent half the trip driving the last five miles. Making highways more efficient and dumping more traffic onto city streets wont reduce travel times independent of latent demand.
|
|
#
?
Jan 17, 2017 22:31
|
|
|
Eletriarnation posted:I wonder what quirk in human psychology causes threads like this one to go down the automated-car rabbit hole over and over instead of pursuing less circular topics. As machines are adapted to take on roles once reserved for humans, humans move to corner the former machine role of getting stuck in a loop and doing the same dumb thing over and over and over and over and over and over and over and over and over and over and over and over and over
|
|
#
?
Jan 17, 2017 22:32
|
|
|
Owlofcreamcheese posted:Okay, but notice how that is literally not at all what happens? Do you not understand the word "bottleneck"? 50% of lanes means a lot less than 50% of throughput, because four lanes of cars becomes two lanes of cars all of a sudden. Squeezing four lanes of cars into two lanes all at once imposes significant extra burdens on throughput - not because humans are bad drivers or something, but rather because of significant physical limitations on merging in congested situations. It's a kind of traffic flow that isn't supposed to happen, and goes against major principles of road design.
|
|
#
?
Jan 17, 2017 22:45
|
|
|
Automated cars would be better at merging, some people are in fact really bad at merging.
|
|
#
?
Jan 17, 2017 22:54
|
|
|
Main Paineframe posted:Do you not understand the word "bottleneck"? 50% of lanes means a lot less than 50% of throughput, because four lanes of cars becomes two lanes of cars all of a sudden. Squeezing four lanes of cars into two lanes all at once imposes significant extra burdens on throughput - not because humans are bad drivers or something, but rather because of significant physical limitations on merging in congested situations. It's a kind of traffic flow that isn't supposed to happen, and goes against major principles of road design. Oh so you are saying that roads can't use 100% of their possible capacity because of inefficiency? I wonder what could reduce that?
|
|
#
?
Jan 17, 2017 23:32
|
|
|
"i believe that one day, thanks to the magic of cloud computing, we will be able to fit two gallons of water into a one gallon jug"
|
|
#
?
Jan 17, 2017 23:33
|
|
|
boner confessor posted:i wonder what quirk in human psychology makes you completely impenetrable to the fact that congestion is simply more demand than supply and no amount of gadgets can change that The efficency of of a mode of transportation can be changed. An example would be container follow through marine terminals. But usually those changes involve systems level changes and not just individual technologies. More can be done with existing capacity but usually it must come with changes to how the particular transportation system is organized.
|
|
#
?
Jan 17, 2017 23:36
|
|
|
boner confessor posted:this is something people often bring up when saying stuff like "oh well we wont need parking lots in the future the cars will just disappear down a rabbit hole" but this idea of the self driving fleet is contradictory to the idea of reducing congestion - cars which are parked are not causing congestion, cars which are driving are, and we have to accept that a much larger number of vehicles will be driving 'uselessly' as in waiting for someone to get in or headed to pick someone up who may be some ways away Okay, so what's you're opinion on something like this article on ride-sharing planning that I posted upthread? The point I was making wasn't that self driving cars would magically reduce congestion, but that self-driving cars could allow cities to adopt policies that reduce congestion while still allowing for taxi-style transportation. The claim here is that better route management would drastically reduce the total number of vehicles needed to serve the same number of people and self-driving vehicles will likely be better at following "perfect" routes planned out by a central authority. All you would need to do to leverage something like this into reduced congestion is to disincentivize single occupancy-only trips through added fees.
|
|
#
?
Jan 17, 2017 23:40
|
|
|
Owlofcreamcheese posted:Oh so you are saying that roads can't use 100% of their possible capacity because of inefficiency? I wonder what could reduce that? Well, for one thing, we could give cars the ability to drive sideways. Or, as long as we're fundamentally changing the design of a transportation system, we could just put jump-jets in every car so they could just jump right over the accident. Hell, why not put big armored cowcatchers on the front of cars so they can push disabled vehicles (and animals, and pedestrians) out of their way?
|
|
#
?
Jan 17, 2017 23:41
|
|
|
lol goons get so salty over the dumbest poo poo
|
|
#
?
Jan 17, 2017 23:44
|
|
|
Speaking of salt, automated cars would put themselves through a carwash as needed in winter weather, thus decreasing rust and corrosion on the bottom of the car.
|
|
#
?
Jan 17, 2017 23:47
|
|
|
Main Paineframe posted:Well, for one thing, we could give cars the ability to drive sideways. Or, as long as we're fundamentally changing the design of a transportation system, we could just put jump-jets in every car so they could just jump right over the accident. Hell, why not put big armored cowcatchers on the front of cars so they can push disabled vehicles (and animals, and pedestrians) out of their way? I just don't understand why this claim is even controversial. Human's are basically never good at shaping their individual behavior to optimize group behavior. People don't even think about that stuff when driving. People are barely even ABLE to think about that sort of thing. It takes researchers in labs running simulations to even be able to say what even is optimal traffic behavior. A system that is mostly or entirely automated cars can maneuver in the way that traffic models say is most efficient for traffic as a whole instead of "however you feel like" the way people do. Not even mentioning cars that potentially communicate with each other and with live traffic data.
|
|
#
?
Jan 18, 2017 00:04
|
|
|
Paradoxish posted:Okay, so what's you're opinion on something like this article on ride-sharing planning that I posted upthread? The point I was making wasn't that self driving cars would magically reduce congestion, but that self-driving cars could allow cities to adopt policies that reduce congestion while still allowing for taxi-style transportation. The claim here is that better route management would drastically reduce the total number of vehicles needed to serve the same number of people and self-driving vehicles will likely be better at following "perfect" routes planned out by a central authority. All you would need to do to leverage something like this into reduced congestion is to disincentivize single occupancy-only trips through added fees. the tradeoff here as a user of this system is time for money. there will still be exactly as many single occupant vehicles on the road as there are first class subscriptions to the network, people who would rather pay more to not participate in group rides. somewhere in the middle are people who might take the cheaper tier and do the auto carpool thing, so long as it doesn't take up too much time. and then at the low end of the spectrum, where you have a lot of time to go out of your way to ride a centrally-planned route with other people... buddy, that's mass transit Owlofcreamcheese posted:I just don't understand why this claim is even controversial. it's not controversial. you're just unable to accept that it wont solve anything because you're looking purely at technology and what it can maybe be capable of doing in the future, ignoring the context of how that technology exists in society. according to things we know right now, no level of self-driving car flocking behavior will be able to solve congestion. it's like saying you can design a spoon to end hunger, you're just not grasping the link between tool and tool use here boner confessor fucked around with this message at 00:57 on Jan 18, 2017 |
|
#
?
Jan 18, 2017 00:55
|
|
|
Owlofcreamcheese posted:I just don't understand why this claim is even controversial. Human's are basically never good at shaping their individual behavior to optimize group behavior. People don't even think about that stuff when driving. People are barely even ABLE to think about that sort of thing. It takes researchers in labs running simulations to even be able to say what even is optimal traffic behavior. A system that is mostly or entirely automated cars can maneuver in the way that traffic models say is most efficient for traffic as a whole instead of "however you feel like" the way people do. Not even mentioning cars that potentially communicate with each other and with live traffic data. Because the net effect of all of that is trivial compared to the impact of the physical limitations cars operate under, and you've absolutely refused to acknowledge that those limitations even exist - every time someone points one out to scuttle one of your hypotheticals, you immediately change the subject without acknowledging it and whine about how no one will accept the potential of technology. Even with literally instant reaction times and perfect group behavior, knocking out two lanes in a four-lane road is still going to cause serious slowdown, because no matter how smart the autodrive is, cars still obey the laws of physics, space, time, and road design.
|
|
#
?
Jan 18, 2017 01:17
|
|
|
What the future could look like in 100 years: https://www.youtube.com/watch?v=Vrxyr1CjiSM
|
|
#
?
Jan 18, 2017 01:31
|
|
|
Main Paineframe posted:Because the net effect of all of that is trivial compared to the impact of the physical limitations cars operate under, and you've absolutely refused to acknowledge that those limitations even exist - Where the hell are you driving where you are pushing against the physical limitations of your cars or biology? Haven't you seen any of the endless studies that show how for example it's totally safe for most roads to quite significantly raise speed limits because they have virtually no effect on death rates until extreme rises because no one is actually straining at their limits of reaction times or car physical abilities?
|
|
#
?
Jan 18, 2017 01:42
|
|
|
Owlofcreamcheese posted:Haven't you seen any of the endless studies that show how for example it's totally safe for most roads to quite significantly raise speed limits because they have virtually no effect on death rates until extreme rises because no one is actually straining at their limits of reaction times or car physical abilities? what do studies about speed limits being too low have to do with congestion on highways? absolutely nothing? this is just another impotent attempt for you to position yourself as having any idea what you're talking about? please stop embarassing yourself oocc and just drop the topic
|
|
#
?
Jan 18, 2017 01:56
|
|
|
boner confessor posted:what do studies about speed limits being too low have to do with congestion on highways? absolutely nothing? this is just another impotent attempt for you to position yourself as having any idea what you're talking about? please stop embarassing yourself oocc and just drop the topic How is it possible to raise speed limits if everyone is always pushed to their physical and mental limits at every moment of driving?
|
|
#
?
Jan 18, 2017 02:15
|
|
|
Owlofcreamcheese posted:How is it possible to raise speed limits if everyone is always pushed to their physical and mental limits at every moment of driving? how is it possible for you to be this lost in the conversation and so desperate to prove that you're not
|
|
#
?
Jan 18, 2017 02:26
|
|
|
Main Paineframe posted:Because the net effect of all of that is trivial compared to the impact of the physical limitations cars operate under, and you've absolutely refused to acknowledge that those limitations even exist - every time someone points one out to scuttle one of your hypotheticals, you immediately change the subject without acknowledging it and whine about how no one will accept the potential of technology. Even with literally instant reaction times and perfect group behavior, knocking out two lanes in a four-lane road is still going to cause serious slowdown, because no matter how smart the autodrive is, cars still obey the laws of physics, space, time, and road design. Better reaction times and group behavior can increase throughout on roads because the cars can go faster and drive closer together. The wide safety margins required for human drivers result in significant inefficiency.
|
|
#
?
Jan 18, 2017 02:40
|
|
|
JeffersonClay posted:Better reaction times and group behavior can increase throughout on roads because the cars can go faster and drive closer together. The wide safety margins required for human drivers result in significant inefficiency. The wide safety margins needed for ancient decrepit cars already keep interstate speed limits a lot lower than they need to be!
|
|
#
?
Jan 18, 2017 02:52
|
|
|
Owlofcreamcheese posted:Where the hell are you driving where you are pushing against the physical limitations of your cars or biology? No matter how fast your reaction time is, it takes your car a certain amount of time - and distance - to stop. The higher the speed, the longer the time - and distance. On top of that, various aspects of road design take expected speed into account. Posted speed limits are often set too low, but don't pretend that means there's no difference between driving 50mph and 150mph.
|
|
#
?
Jan 18, 2017 02:53
|
|
|
News just ran a story saying ⅔ of Australian businesses have or intend to install AI. The guy they interviewed talked about how they can do Jobs quicker and better than humans. Looking forwards to seeing the new jobs that will be made for the people who lost their jobs. Sorry for the derail back to yet more super interesting self driving car talk.
|
|
#
?
Jan 18, 2017 03:14
|
|
|
Owlofcreamcheese posted:Where the hell are you driving where you are pushing against the physical limitations of your cars or biology? Where the hell are you driving that the interstate isn't packed with retards following at distances well below the physical limits of their car's braking ability, let alone their reaction times? I swear to god I can't drive 5 minutes on 101 without seeing a chain of people going 80mph with less than 2 meters between them. At least if someone decides to quickly change into their lane it will be well signaled and timed!
|
|
#
?
Jan 18, 2017 03:21
|
|
|
Please stop talking about autonomous cars. Here, I made a thread just for you guys.
|
|
#
?
Jan 18, 2017 07:16
|
|
|
Tasmantor posted:News just ran a story saying ⅔ of Australian businesses have or intend to install AI. The guy they interviewed talked about how they can do Jobs quicker and better than humans. Looking forwards to seeing the new jobs that will be made for the people who lost their jobs. Wait, for the businesses that have actually installed AI, what was it that was actually installed? Also, the "plan to install" category sounds donors to me.
|
|
#
?
Jan 18, 2017 11:33
|
|
|
Some guy from Salesforce was bragging about their advanced new Einstein AI, turns out it's basically automated lead scoring  I mean it's absolutely valuable but this is a pretty low bar for AI so I wouldn't think too much about those figures.
|
|
#
?
Jan 18, 2017 11:44
|
|
|
|
| # ? May 8, 2024 22:58 |
|
|
mobby_6kl posted:Some guy from Salesforce was bragging about their advanced new Einstein AI, turns out it's basically automated lead scoring Some powerful AI algorithm are really simple. You can get natural behavior, like how birds fly in flocks with a few lines of code, applying 3 simple rules. That some AI algorithm is simple should not surprise us all that much, maybe the most complicate poo poo is artificial vision, but everything else can be made very simple. You can track back AI to the first steam engines pressure regulators, a simple mechanism that would take the decision to allow pressure to escape when over limit (and not below that limit).  (this is a image of the first AI system in steam engines)
|
|
#
?
Jan 18, 2017 14:44
|
|