I might be wrong, but surely it's a 4 SECOND ping time ?
Earth to Mars furthest distance 401 million km.
Speed of light 299 million km per second.
Round trip time 4 SECONDS.
Ok so at this distance the sun is smack bang in the way, but from all other viable viewing points the distance is less than the 401 million km used above.
Is this not what happens to a people under the stress of being told that they are constantly under attack from all sides ?
Between "terrorism", "cyber-terrorism", and the rising local fears and money troubles, is it any wonder that the signs of stress are showing ?
Not that I don't think that such drugs are over prescribed, far from it, rather than seeking the cause they are more likely to medicate the hell out of the symptom.
We'd not have so much of a problem with size up there, and the link from here to the moon would require smaller telescopes, with at least one at each pole.
Just how poor does one have to be not to benefit from the rich peoples technology?
That £40 - £60 per year is my entire entertainment budget for the year! If truth be told I can't really afford that much, a good 50% is gifts of vouchers at christmas and my birthday.
What monthly fees ? I am on pay as you go and *only* use free wifi for data thank you very much !
Cost of handset to me : FREE BIRTHDAY GIFT
I do not and will not purchase online, so your crack about 80 "cents" for a song is irrelevant.
I have over 150+ cd's and 40+ DVD's which were purchased at retail in my personal collection. I add one or two each quarter sometimes more if the bargain bin has something interesting.
Like I want to spend a weekend or two ripping that lot myself ?
Again it is NOT the tool, just the uses *some* put them to, no different from lock picks.
I think you were close with "own moral Geiger counter" but went wayward with "Maybe it's more honest to talk about right and wrong, and then think about the best way to define laws around that."
We already have.
Can you define the difference between a digital lock pick and a physical one ?
I can't so I can see no reason why they should be treated differently.
That is almost exactly what a gunman in America just did ! He couldn't afford heath care, so what did he do ?
He got himself a gun, held up a bank and stole $10. Yes TEN DOLLARS!
As for "does not, or at least should not, own a smart phone"
"should not ?" and why pray tell.. Is earning less than you some kind of a problem ? Do have less rights ?
As for "does not" : Ever hear of gifts?
You know - things people give you when they like you...
My friends and family chose to get me one birthday present instead of lots. The replaced my ageing Nokia 7110 with a refurbished HTC Hero (Android yay) on the "Three" network for £160.
If I choose to use it to acquire a backup of my legally purchased media then that should be my choice.
Once again the tools are not at fault, it is the uses that *some* put them too. This is no different than physical lock picks, only with them we learnt than banning or castigating the creators or owners or lawful users only serves to increase the illegal use. Why we cannot learn from history is beyond me.
Source : Toys R Us uk website.
Xbox 360 Call of Duty Modern Warfare 3 £42.89
Xbox 360 Disney Pixar Cars 2 £34.89
Amazon uk website:
Case Histories - Series 1 [DVD] £14.99
The Shadow Line [DVD] £14.99
I Am Number Four [DVD] £11.99
You may be rich enough for this to be loose change to you, but I can assure you this not not cheap. Particularly given the actual dvd media costs less than £1.
Breaking DRM and stripping CSS may be easy for you but it is not so much for mere mortal USERS.
I care about the difference between right and wrong, I just don't see how it is right to ban a digital lock pick but not a physical one.
With the exorbitant cost of physical media, the relative ease with which it can be damaged, and whatever DRM is embedded, I cannot blame people for wanting a cheap no hassle backup.
Anyone blaming the tools is nuts. We have had similarly disruptive tools before and should know that banning or making them illegal does nothing to stop their use.
Think if you will about lock picks. Legal in most of the world to own, to use on your own locks, to carry around in you car, but illegal to use to commit a crime.
What's so different with the digital equivalent ?
Why do governments and people world wide seem so scared of digital lock picks, but are content to have legally available physical ones ?
Might have something to do with an anticipated fear reaction from the locations populace.
Or a "London" is too big a target to put a new nuclear in fear of "terrorism".
Me - I'd have no problem with a nuclear power station as per the newer and even safer Japanese designs as a neighbour. Not so keen on the French designs we're likely to end up with though.
The Japanese have already done the disaster tests after all, and have a better long term safety record than the French.
I would expect a discount on my electricity bill though.
The UK is an island with little in the way of seismic activity or tsunami like events.
The UK already falls short in power generation, requiring imports at peak times through lengthy vulnerable and costly to lay and maintain under sea cables and pipes.
By 2015 it is estimated that the UK will no longer produce sufficient energy for off peak, requiring imports 24 hours a day 7 days a week.
We can ill afford the risks these vulnerabilities pose let alone the 10+ years of money flooding out of our economy for what should be a matter of national security.
Randomly assigned IP addresses can be static or dynamic. You assign one static to each machine and let it generate dynamic addresses on its own. For incoming connections you use the static IP of the machine. For outgoing connections you use one of the dynamic IP addresses of the machine.
Thank you for this, it forced me to re-read the faq (http://www.faqs.org/rfc/rfc3041.txt). I must admit that had been focusing on it's primary declared relationship to "Stateless address autoconfiguration [ADDRCONF]", and failing entirely to grasp the "may also apply to interfaces with other types of globally unique and/or persistent identifiers" part.
Too many people with that attitude is the reason for the mess we have now.
Some of us are either more cautious, or less well informed. I was both, now I am merely cautious. I will gladly and with thanks, move on to basic connectivity testing rather than waiting.
If you have information regarding implementing Security Enhanced Neighbour Discovery please link it as this is now the final hurdle for me.
I did, and from it I headed down the path that you are on. That was until I also wanted a firewall as well as randomisation. If you implement a default deny firewall and are running randomised addresses, just how do you open a port ? Or otherwise grant access for inbound connections ?
All the flaws of NAT but without any of the benefits.
I am sure that there is a solution to this problem, it just has yet to be released.
I am just willing to wait for that or until ipv6 reaches critical mass and I am forced.
My thought is that running an open wifi does not provide plausible deniability. It's more likely that someone will do something malicious behind your gateway and you'll take the blame than vice-versa. *Especially* if you seem technically capable, the fact that you explicitly left your wifi open would be taken as a sign you were *trying* for plausible deniability. Face it, for the residential case, *there is no plausible deniability*, at least with respect to traffic that originates from your residence, *unless* you have a trusted proxy shared with others out there that you *know* won't retain enough data to trace your identity. The only way to have plausible deniability is to find an open-wifi somewhere and hope there's no security camera. If it is some poor sap's house, then they will probably get blamed, if a business, that business may be required to discontinue open wifi under legal pressure.
Here I think we will have to agree to disagree. Particularly when you consider some of the advantages to the privacy extensions. My point is that at present, there is no happy medium. You have a choice between a centralised traditional firewall, and a decentralised randomised more privacy friendly solution.
I think we can agree that ipv6 could be far better than it is with what we know today verses when it was designed 15 years ago. I'm just willing to wait a little longer for my feature set than you are for yours.
How is that different from your NAT today? If you want to accept incoming connections, you must tell your NAT box a port to DNAT map from your external thing to something internal, defined by, surprise surprise, a static entry.
The differences are:
1) A single static ip address in ipv4 can be either a single device or a NAT gateway. In ipv6 it is guaranteed to be a single device.
2) The perception that since a static ipv6 address is just one of the possibilities out of a 64bit subnet, that this renders address scanning useless. This perception is blatantly false, as without address randomisation you leave "footprints" everywhere you go hence the privacy extensions. Who needs to scan for your address when you leave it wherever you go ?
The current implementations of ipv6 leaves you the choice between security and privacy - you cannot have both.
If you choose security you cannot even have plausible deniability by running an open wifi as all ipv6 addresses are unique.
If on the other hand you choose privacy, then you cannot implement a default deny firewall as this would require a whitelist listing all of the allowed ipv6 addresses - something that you cannot provide if you are randomising your ip address as per the privacy rfc.
I will wait until someone figures out how to do both before I consider going live with ipv6.
But why do you care if they're known outside your network? You have a stateful firewall that protects them from the world. Here's my printer's IPv6 address: 2001:453:da65:1:94ab:7c00:8cba:beb5. Go ahead, have fun trying to connect to it.
You have far more confidence in your firewall than I have. One slip in the coding, one unchecked buffer is all that it takes for it to be breached.
Then post your password here and/or SSH private key here. "Security through obscurity" is not remotely close to what you think it means.
Those are secrets that have no existence outside of my network. Unlike IP addresses. I believe you are mistaken in equating them.
Set your firewall policy to "default deny" and whitelist connections you specifically want to allow. This has been the correct way of building firewalls since the idea was first invented.
Why? You don't have a firewall on your router? Again, "default deny": don't open up a rule that allows random Internet hosts to connect to your toaster.
And what prey tell should I do for my PC ? Set a static ipv6 address to be entered into the whitelist ?
Address randomisation does not even begin to solve the problem, in fact it makes it worse. How can my firewall be expected to know the difference between an address generated by my network printer that should not be seen from outside my network and one from a pc that should ?
So now even my network printer (toaster, fridge, whatever) needs a built in firewall with guaranteed bug fixes.
When was the last time you saw a printer or other device manufacturer fixing such security flaws in a timely manner ?
And this is progress ????
Auto configuration is a nightmare. I want to be alerted to the addition of any kit to my network and be given the choice to allow or disallow access to my resources before whatever it is starts to use the limited data allocation that is my internet connection, starts to print a copy of wikipedia or otherwise use resources that cost me time or money.
Before anyone chimes in with "Security Enhanced Neighbour Discovery" - find me a howto that shows the proper configuration of "SEND" that creates a secure network of Windows and Linux machines..... Go on... I'm not holding my breath......
There is only one fully autonomous train system in the world today. The Copenhagen Metro.
I have no experience of this system and would readily defer to someone who has or does use it.
Justice is not a matter of money.
Particularly when applied to the corporate world. No jail time no worries, kill someone, get caught, pay a fine that you probably have specific business insurance to cover.
For as long as the company's "death" rate does not cause the insurance to cost too much, then it's business as usual.
Proving liability in the first place is likely to be troublesome, as these are likely to be closed systems, can a private individual gain access to the source, schematic or datalog to verify the cause of an accident ? Or do we just trust without verifying the makers ?
Asking for a system that can not only track, but identify correctly at speed from a moving platform everything that could cause it to alter it's maneuvering is a bit much. Particularly when you need to take into account variable levels of light, weather (rain / fog / mist).
"That is not to say that the computer couldn't easily have a decent amount of situational awareness"
Predefined "caution areas" are so far from being situationally aware it is not funny. For an autonomous system, it has to be "aware" at all times. In order to be aware it needs to know not only "Where are things in relation to me ?" and "How are they moving in relation to me ?" you also need to know "What are these things ?" and knowing these answers gives you input into the "How are these things likely to behave in relation to me ?" question.
For our current systems we can know in realtime faster than human reflexes, the where and how things are moving questions. What we have difficulty with is the "what" - is it a manikin, a human a statue or a large dog on it's hind legs ? This is a question that we cannot currently compute in realtime. Hell we can't even reliably discern a dog from a cat with our current technology.
Even an over cautious system is worse than no system as every hesitation or slow down due to over cautious behavior has huge ramifications in the overall traffic flow.
And if it is so possible - why have we not got autonomous trams or trains? Surely as there are no "turns" and fewer human interaction areas these should be the first to be experimented with, before moving into areas with higher degrees of mobility, and greater risks to the public should a vehicle "crash" in any sense of the word.
I doubt that we are within 50 years of having the combination of the stability of our existing embedded systems, with the computing power and complexity required.
We are even further away from the programming paradigms and AI models that will allow us to build systems such as these that need to be able to cope safely with even the most unlikely events.
Dealing with pedestrians of all shapes, sizes, ethnicities appearing from the side of the road, stepping out, possibly from between parked cars, will be one of the tougher sets of tests.
Having to identify not only the potential for danger, but also recognizing when it is safe as the individual is loading the boot of their car and not about to cross the street.
"Seeing" is but one part of an incredibly complicated system, understanding - comprehending so as to properly adjust for potential future events is something else. "See" a ball bounce into the road in front of you - think child.... As an autonomous system, just how long should it wait for the ball to be collected before it considers it "safe" to proceed ? It can't exactly get out and move the ball itself...
Slashdot Mirror
Aww shucks got my decimals in the wrong place :-(
You are correct my apologies.
I might be wrong, but surely it's a 4 SECOND ping time ?
Earth to Mars furthest distance 401 million km.
Speed of light 299 million km per second.
Round trip time 4 SECONDS.
Ok so at this distance the sun is smack bang in the way, but from all other viable viewing points the distance is less than the 401 million km used above.
Is this not what happens to a people under the stress of being told that they are constantly under attack from all sides ?
Between "terrorism", "cyber-terrorism", and the rising local fears and money troubles, is it any wonder that the signs of stress are showing ?
Not that I don't think that such drugs are over prescribed, far from it, rather than seeking the cause they are more likely to medicate the hell out of the symptom.
Is a relay via the moon not possible ?
We'd not have so much of a problem with size up there, and the link from here to the moon would require smaller telescopes, with at least one at each pole.
Significant illegal use makes no negative difference to my argument. In fact it bolsters my comparison with lock picks.
As with almost all of our tools there are multiple uses.
If you are going to ban digital tools, why not start with physical ones first like lock picks?
Ohh wait we've been there and done that already and we know how that turned out.
Why should the law abiding suffer through the loss of useful tools thanks to the those breaking the law ?
Quit blaming the tools when it's the USERS who are at fault *if* they abuse them.
Just how poor does one have to be not to benefit from the rich peoples technology?
That £40 - £60 per year is my entire entertainment budget for the year! If truth be told I can't really afford that much, a good 50% is gifts of vouchers at christmas and my birthday.
Completely beside the point. The point, like lock picks, is that there are legitimate uses for these technologies.
Your view that since there are illegal uses that the technologies should be banned.
BUT with lock picks, where there are significant illegal uses, we still acknowledge the individuals right to own and use them responsibly.
There in lies the key if you will pardon the pun.
Individual responsibility.
You advocate for a course of repressive action that is completely unwarranted as has already been proven.
What monthly fees ? I am on pay as you go and *only* use free wifi for data thank you very much !
Cost of handset to me : FREE BIRTHDAY GIFT
I do not and will not purchase online, so your crack about 80 "cents" for a song is irrelevant.
I have over 150+ cd's and 40+ DVD's which were purchased at retail in my personal collection. I add one or two each quarter sometimes more if the bargain bin has something interesting.
Like I want to spend a weekend or two ripping that lot myself ?
Again it is NOT the tool, just the uses *some* put them to, no different from lock picks.
Ok so it was $1... Not my currency sorry for the error.
I think you were close with "own moral Geiger counter" but went wayward with "Maybe it's more honest to talk about right and wrong, and then think about the best way to define laws around that."
We already have.
Can you define the difference between a digital lock pick and a physical one ?
I can't so I can see no reason why they should be treated differently.
"gonna buy a gun and rob a store" rotflma
That is almost exactly what a gunman in America just did ! He couldn't afford heath care, so what did he do ?
He got himself a gun, held up a bank and stole $10. Yes TEN DOLLARS!
As for "does not, or at least should not, own a smart phone"
"should not ?" and why pray tell.. Is earning less than you some kind of a problem ? Do have less rights ?
As for "does not" : Ever hear of gifts?
You know - things people give you when they like you...
My friends and family chose to get me one birthday present instead of lots. The replaced my ageing Nokia 7110 with a refurbished HTC Hero (Android yay) on the "Three" network for £160.
If I choose to use it to acquire a backup of my legally purchased media then that should be my choice.
Once again the tools are not at fault, it is the uses that *some* put them too. This is no different than physical lock picks, only with them we learnt than banning or castigating the creators or owners or lawful users only serves to increase the illegal use. Why we cannot learn from history is beyond me.
Costs - a random selection :
Source : Toys R Us uk website.
Xbox 360 Call of Duty Modern Warfare 3 £42.89
Xbox 360 Disney Pixar Cars 2 £34.89
Amazon uk website:
Case Histories - Series 1 [DVD] £14.99
The Shadow Line [DVD] £14.99
I Am Number Four [DVD] £11.99
You may be rich enough for this to be loose change to you, but I can assure you this not not cheap. Particularly given the actual dvd media costs less than £1.
Breaking DRM and stripping CSS may be easy for you but it is not so much for mere mortal USERS.
I care about the difference between right and wrong, I just don't see how it is right to ban a digital lock pick but not a physical one.
With the exorbitant cost of physical media, the relative ease with which it can be damaged, and whatever DRM is embedded, I cannot blame people for wanting a cheap no hassle backup.
Anyone blaming the tools is nuts. We have had similarly disruptive tools before and should know that banning or making them illegal does nothing to stop their use.
Think if you will about lock picks. Legal in most of the world to own, to use on your own locks, to carry around in you car, but illegal to use to commit a crime.
What's so different with the digital equivalent ?
Why do governments and people world wide seem so scared of digital lock picks, but are content to have legally available physical ones ?
Might have something to do with an anticipated fear reaction from the locations populace.
Or a "London" is too big a target to put a new nuclear in fear of "terrorism".
Me - I'd have no problem with a nuclear power station as per the newer and even safer Japanese designs as a neighbour. Not so keen on the French designs we're likely to end up with though.
The Japanese have already done the disaster tests after all, and have a better long term safety record than the French.
I would expect a discount on my electricity bill though.
IF subsequent governments don't screw it up.
The UK is an island with little in the way of seismic activity or tsunami like events.
The UK already falls short in power generation, requiring imports at peak times through lengthy vulnerable and costly to lay and maintain under sea cables and pipes.
By 2015 it is estimated that the UK will no longer produce sufficient energy for off peak, requiring imports 24 hours a day 7 days a week.
We can ill afford the risks these vulnerabilities pose let alone the 10+ years of money flooding out of our economy for what should be a matter of national security.
Any delays will be costly in so many ways.
Randomly assigned IP addresses can be static or dynamic. You assign one static to each machine and let it generate dynamic addresses on its own. For incoming connections you use the static IP of the machine. For outgoing connections you use one of the dynamic IP addresses of the machine.
Thank you for this, it forced me to re-read the faq (http://www.faqs.org/rfc/rfc3041.txt). I must admit that had been focusing on it's primary declared relationship to "Stateless address autoconfiguration [ADDRCONF]", and failing entirely to grasp the "may also apply to interfaces with other types of globally unique and/or persistent identifiers" part.
Too many people with that attitude is the reason for the mess we have now.
Some of us are either more cautious, or less well informed. I was both, now I am merely cautious. I will gladly and with thanks, move on to basic connectivity testing rather than waiting.
If you have information regarding implementing Security Enhanced Neighbour Discovery please link it as this is now the final hurdle for me.
Did you even watch the video you linked to?
I did, and from it I headed down the path that you are on. That was until I also wanted a firewall as well as randomisation. If you implement a default deny firewall and are running randomised addresses, just how do you open a port ? Or otherwise grant access for inbound connections ?
All the flaws of NAT but without any of the benefits.
I am sure that there is a solution to this problem, it just has yet to be released.
I am just willing to wait for that or until ipv6 reaches critical mass and I am forced.
1) Whether it is an IPv6 address or an IPv4address+DNAT port, the exposure is the same, the outside world has a door into a specific system.
Unless you are running the ipv6 privacy extensions :
http://playground.sun.com/ipv6/specs/ipv6-address-privacy.html
http://www.faqs.org/rfc/rfc3041.txt
My thought is that running an open wifi does not provide plausible deniability. It's more likely that someone will do something malicious behind your gateway and you'll take the blame than vice-versa. *Especially* if you seem technically capable, the fact that you explicitly left your wifi open would be taken as a sign you were *trying* for plausible deniability. Face it, for the residential case, *there is no plausible deniability*, at least with respect to traffic that originates from your residence, *unless* you have a trusted proxy shared with others out there that you *know* won't retain enough data to trace your identity. The only way to have plausible deniability is to find an open-wifi somewhere and hope there's no security camera. If it is some poor sap's house, then they will probably get blamed, if a business, that business may be required to discontinue open wifi under legal pressure.
Here I think we will have to agree to disagree. Particularly when you consider some of the advantages to the privacy extensions. My point is that at present, there is no happy medium. You have a choice between a centralised traditional firewall, and a decentralised randomised more privacy friendly solution.
I think we can agree that ipv6 could be far better than it is with what we know today verses when it was designed 15 years ago. I'm just willing to wait a little longer for my feature set than you are for yours.
How is that different from your NAT today? If you want to accept incoming connections, you must tell your NAT box a port to DNAT map from your external thing to something internal, defined by, surprise surprise, a static entry.
The differences are :
1) A single static ip address in ipv4 can be either a single device or a NAT gateway. In ipv6 it is guaranteed to be a single device.
2) The perception that since a static ipv6 address is just one of the possibilities out of a 64bit subnet, that this renders address scanning useless. This perception is blatantly false, as without address randomisation you leave "footprints" everywhere you go hence the privacy extensions. Who needs to scan for your address when you leave it wherever you go ?
The current implementations of ipv6 leaves you the choice between security and privacy - you cannot have both.
If you choose security you cannot even have plausible deniability by running an open wifi as all ipv6 addresses are unique.
If on the other hand you choose privacy, then you cannot implement a default deny firewall as this would require a whitelist listing all of the allowed ipv6 addresses - something that you cannot provide if you are randomising your ip address as per the privacy rfc.
I will wait until someone figures out how to do both before I consider going live with ipv6.
But why do you care if they're known outside your network? You have a stateful firewall that protects them from the world. Here's my printer's IPv6 address: 2001:453:da65:1:94ab:7c00:8cba:beb5. Go ahead, have fun trying to connect to it.
You have far more confidence in your firewall than I have. One slip in the coding, one unchecked buffer is all that it takes for it to be breached.
Yes, of course. Why wouldn't you?
Privacy.
http://playground.sun.com/ipv6/specs/ipv6-address-privacy.html
http://www.faqs.org/rfc/rfc3041.txt
Then post your password here and/or SSH private key here. "Security through obscurity" is not remotely close to what you think it means.
Those are secrets that have no existence outside of my network. Unlike IP addresses. I believe you are mistaken in equating them.
Set your firewall policy to "default deny" and whitelist connections you specifically want to allow. This has been the correct way of building firewalls since the idea was first invented.
Why? You don't have a firewall on your router? Again, "default deny": don't open up a rule that allows random Internet hosts to connect to your toaster.
And what prey tell should I do for my PC ? Set a static ipv6 address to be entered into the whitelist ?
Pull the other one it's got bells on.
Security though obscurity is no security at all.
For every website or service you encounter on the internet you have to provide an address to which replies can be sent.
Who needs to port scan ?
Port scanning is not even as difficult as was first believed : http://www.youtube.com/watch?v=c7hq2q4jQYw
Address randomisation does not even begin to solve the problem, in fact it makes it worse. How can my firewall be expected to know the difference between an address generated by my network printer that should not be seen from outside my network and one from a pc that should ?
So now even my network printer (toaster, fridge, whatever) needs a built in firewall with guaranteed bug fixes.
When was the last time you saw a printer or other device manufacturer fixing such security flaws in a timely manner ?
And this is progress ????
Auto configuration is a nightmare. I want to be alerted to the addition of any kit to my network and be given the choice to allow or disallow access to my resources before whatever it is starts to use the limited data allocation that is my internet connection, starts to print a copy of wikipedia or otherwise use resources that cost me time or money.
Before anyone chimes in with "Security Enhanced Neighbour Discovery" - find me a howto that shows the proper configuration of "SEND" that creates a secure network of Windows and Linux machines..... Go on... I'm not holding my breath......
As per the link : http://en.wikipedia.org/wiki/List_of_driverless_trains
There is only one fully autonomous train system in the world today. The Copenhagen Metro.
I have no experience of this system and would readily defer to someone who has or does use it.
Justice is not a matter of money.
Particularly when applied to the corporate world. No jail time no worries, kill someone, get caught, pay a fine that you probably have specific business insurance to cover.
For as long as the company's "death" rate does not cause the insurance to cost too much, then it's business as usual.
Proving liability in the first place is likely to be troublesome, as these are likely to be closed systems, can a private individual gain access to the source, schematic or datalog to verify the cause of an accident ? Or do we just trust without verifying the makers ?
Asking for a system that can not only track, but identify correctly at speed from a moving platform everything that could cause it to alter it's maneuvering is a bit much. Particularly when you need to take into account variable levels of light, weather (rain / fog / mist).
"That is not to say that the computer couldn't easily have a decent amount of situational awareness"
Predefined "caution areas" are so far from being situationally aware it is not funny. For an autonomous system, it has to be "aware" at all times. In order to be aware it needs to know not only "Where are things in relation to me ?" and "How are they moving in relation to me ?" you also need to know "What are these things ?" and knowing these answers gives you input into the "How are these things likely to behave in relation to me ?" question.
For our current systems we can know in realtime faster than human reflexes, the where and how things are moving questions. What we have difficulty with is the "what" - is it a manikin, a human a statue or a large dog on it's hind legs ? This is a question that we cannot currently compute in realtime. Hell we can't even reliably discern a dog from a cat with our current technology.
Even an over cautious system is worse than no system as every hesitation or slow down due to over cautious behavior has huge ramifications in the overall traffic flow.
And if it is so possible - why have we not got autonomous trams or trains? Surely as there are no "turns" and fewer human interaction areas these should be the first to be experimented with, before moving into areas with higher degrees of mobility, and greater risks to the public should a vehicle "crash" in any sense of the word.
I doubt that we are within 50 years of having the combination of the stability of our existing embedded systems, with the computing power and complexity required.
We are even further away from the programming paradigms and AI models that will allow us to build systems such as these that need to be able to cope safely with even the most unlikely events.
Dealing with pedestrians of all shapes, sizes, ethnicities appearing from the side of the road, stepping out, possibly from between parked cars, will be one of the tougher sets of tests.
Having to identify not only the potential for danger, but also recognizing when it is safe as the individual is loading the boot of their car and not about to cross the street.
"Seeing" is but one part of an incredibly complicated system, understanding - comprehending so as to properly adjust for potential future events is something else. "See" a ball bounce into the road in front of you - think child.... As an autonomous system, just how long should it wait for the ball to be collected before it considers it "safe" to proceed ? It can't exactly get out and move the ball itself...