Flare is one thing, but PURPLE flare is rather obnoxious. Look at the examples linked from the story... iPhone4 gives a bit of flare (saturation) as expected, but iPhone 5 gives the startling purple flare.
Yes, but they also state that intent may be gauged by actions (& statements). In practice I'd guess the US doesn't often use this rule because they want to keep taxing you (as a US citizen), but anyone looking to do this would be well advised to reseach it thoroughly before just assuming they'll be able to retain their US citizenship without qualifications.
Saying that OS X killed Linux on the desktop is a misstatement at best since Linux on the desktop never became mainstream. You can't kill what's not yet alive.
However, client-side (but not "desktop") Linux is very much alive and kicking in the shape of Android tablets (Kindle, Nook, Nexus 7), and currently owns the small form factor tablet space. Apple is the one trying to play catch-up here with the rummoured upcoming iPad mini.
It's interesting to note that the success of Android is following the path led by Apple with OS X... first on the smart phone, then on the tablet. This was an easier path to mainstream adoption since it wasn't fighting the entrenched desktop ecosystems head-on, but rather building a brand new (smartphone) market. In this vein it might be better to view Nokia's incompetence as the real killer of pure (non-Android) Linux since they were the only ones targetting a Linux-based smartphone. If Nokia had moved faster and followed a release early, release often" path (as Android did), they could well have been successful.
Sure it (shorting against the box) can be used to defer/reduce taxes. That's one common use of it.
Say you have a stock that you'd like to sell, but you don't want to pay taxes this year. What you do is instead short it ("against the box" since you still keep your original shares), thereby locking in the current price. Since neither your long or short position is closed, you currently pay no taxes. Then, whever you want to take your profit (e.h. next tax year, or when you have offsetting losses), you cover your short position and sell your shares, thereby becoming liable for any taxes owed.
That the way to defer taxes...but you can also reduce taxes to ZERO by simply NEVER closing the positions.... You buy a stock (or get in in an IPO), then rather that sell it short againt the box, thereby getting your cash, and NEVER close either long/short position, thereby never incurring a tax liability. People actually do this - it's not just a theoretical tax hole.
In Linux it's simple enough to, say, mount your root (OS data) folder on an SSD and/home (user data) on a HDD, but Windows 7 isn't so flexible.
What most people (power users) end up doing under Windows 7 is to install the OS on an SSD, then use a "junction point" (cf Linux hard link) to redirect the/Users folder to a HDD (and reconfigure the Windows TEMP directory to be on the HDD to avoid killing the SSD with excessive temp file create/delete cycles). The trouble with this is that Windows 7 junction points don't play nice with restore points, as you'll find out when having to revert to a restore point and all your user data disappears requiring major hackery to restore.
So, for Windows 7, a HDD with built in flash cache is a MUCH more convenient solution than using a separate SDD - even for a power user.
It depends on what level of (artifical) intelligence you're talking about. If it's amoeba level intelligence, then maybe ML can achieve similar results, but if it's rat or human level intelligence then obviously not.
I think most people take AI to mean something that could minimally pass a Turing test, not a silicon slug.
Support Vector Machines are just a way of performing unsupervised data partitioning/clustering. i.e. you feed a bunch of data vectors into the algorithm and it determines how to split the data into a number of clusters where the members of each cluster are similar to each other and less similar to members of other clusters.
e.g. you feed it (number of wheels, weight) pairs of a lot of vehicles and it might automatically split the data into 3 clusters - light 2-wheeled vehicles, heavy 4-wheeled ones, and very heavy 4-wheeled ones. If you then labelled these clusters as "bikes", "cars" and "trucks" you could in the future use the clustering rules to determine the category a new data point falls into.
This isn't Artificial Intelligence - it's just a data mining/classification technique.
As I understand it, a Higgs Boson compatible with the standard model could have been found at a range of different masses, and the search for it has involved searching the possible mass range until it was either discovered or not.
Assuming that this new discovery is indeed the Higgs Boson as predicted and compatible with the standard model, what is the significance of the particular mass that it has been found to have? Are there any macro-scale predictions that depend on its mass?
I'm not sure they actually wanted to make a vapourware annoucement, but they were forced to by the Nokia turn of events and headlines (Nokia - Microsoft's main Windows phone parter in the process of imploding and just announcing 20% layoffs and reduced sales forcasts).
This event had "last minute" and desperation written all over it. Why would they time it on the day after the Greek elections when the stock market could have been plummeting. Why pre-announce without specifications, prices or a ship date? Why no real demo or applications? Why was presenter Steven Sinofsky (Windows head honcho) so nervous his hands were shaking?
I think they just decided to throw this out at the last minute as a desperate attempt to say "forget Nokia, our mobile strategy isn't dead yet".
It's a lot of money, but it's less than 10% of what Google is currently bringing in. I happen to think that Google will find a way to prosper as the internet changes (not sure how, but I have faith in their flexibility), but a 90% revenue cut certainly would be bad news!
Selling stock you don't have doesn't require options - it's called "naked short selling", and specifically is allowed by the SEC as a form of IPO stabilization.
I don't have any great source for how this went down, although you can google for the greenshoe (aka over allocation) option in the facebook IPO contract as well as the SEC sanction of naked short selling.
There is one site here claiming the same events, but it's really more a matter of assuming that the underwriters were smart enough to take advantage of the greenshoe provision rather than risking their own cash to support the stock. Seems like a fair bet to me.
Just wanted to make a correction to that last point... In fact it looks like, despite massive buying to support the stock, the underwriters don't in fact own the stock they bought back(!), and hence there won't be this large pending sale hanging over the market.
The reason for this is a "greenshoe" provision in the facebook IPO which would have allowed the underwriters to oversell the IPO by 15% if they chose to (i.e. to exceed the intended 420M share float by 15% - buying 15% more at IPO price from FB and selling it to the market).
The underwriters can (and appear to have) use this greenshoe provision to stabilize/support the IPO without any risk to themself.
The way this appears to have gone down is:
1) Underwriters sold 15% more stock than they bought from FB (they buy at 37.5, sell at 38). Specifically they sold 480M shares, having bought only 420M shares from FB (420M being the intended IPO size notwithstanding the greenshoe provision).
2) Underwriters having sold (i.e shorted) 60M (480-420) shares they don't own need to buy it back to cover this short. This could happen in two ways, depending on whether the IPO was successful or not (stock went up or down), but in either case it's a no-risk proposition to the underwriters!
2a) If FB stock had gone up above IPO price then the underwriters would exercise the greenshoe provision and cover their 60M short position buy buying an additional 60M shares (the 15% greenshoe provision) from FB at IPO price, therefore incurring zero loss.
2b) If FB stock had gone (or attempted to go) below IPO price, then the underwriters would cover their 60M short position buy buying in the open market at or below IPO price thereby stabilizing the stock price (purchase up to 60M shares). This is what actually happened. There was some underwriter support (buying) just before midday and then from 3:30-4:00, and the volume purchased seems very close to the 15% (60M shares) greenshoe position.
The net result is that the underwriters were able to artificially support the stock by massive buying, but did so at zero risk to themselves, and ended up owning zero stock! In theory they might have had to buy more than the greenshoe provision amount, and therefore been left actually holding some stock, but on the day the numbers seem to have worked perfectly for them.
So - it's a rigged game. The underwriters can't lose whether the stock goes up or down, but the public can. Facebook got greedy with the price such that there was no upside to the stock, so the suckers buying it AT THIS PRICE were the ones taking all the risk and almost guaranteed short term loss.
There were certainly NASDAQ software problems (acknowledged by NASDAQ), but AFAIK that was in order execution confirmations, not order execution itself (i.e. you submitted an order to buy/sell FB and the order executed in timely fashion, but your notification of the order execution was in some cases significantly delayed).
You can't say 100% (vs 99.99%) for sure that the massive buy orders at $38 were the underwriters, but:
1) That's the definition/job of underwriting - to be the buyers of last resort if there's insufficient demand for the stock, either: a) Pre-IPO (underwriter holds stock it can't sell), or b) Post-IPO to support IPO price, else massive loss of face due to having mispriced the offering
2) The price action can't realistically be anything other than due to underwriters. A normal stock buyer trys to buy at the lowest price possible, not to support a given price. If you're going to put in a limit order, you put it in below market, not at market. Those "unlimited" buy orders at $38 at 3:30-4:00pm when the stock was trying to go lower were therefore not from someone trying to buy for profit - they were from someone trying to prop up the price at that level), and the only people who do that are underwriters.
You may see more of that same artificial flatlining in the next couple of weeks to some extent, but the underwriters certainly won't support it for too long.
Who knows how long the underwriters will hold that stock before trying to sell it. Quite likely they may wait until FB is included in various indexes (QQQ, S&P 500) which creates artificial demand (index tracking funds MUST buy the stocks in the index)... but in any case that's a lot of stock waiting to get sold...
Total volume for the day was 570M shares traded - the most for an IPO ever. Volume in the last half hour (when all the buying was by the underwriters) was 50M+ (maybe 70M - I think it started around the 500M mark).
The only thing that prevented the stock from going below the IPO price was MASSIVE underwriter support (i.e. the banks that got paid to do the IPO buying the stock in large volume whenever it hit the IPO price of $38).
See the flatline at $38 from 3:30-4:00pm? That's due to the underwriters buying the stock in massive volume providing support at $38. If you monitored the detailed stock quote during this time (as I did), you'd have seen that the bid volume, normally single or double digit lots spiked to a continuous 99999+ lots (i.e. 10,000,000+ shares) during this time - i.e. the underwriters were essentially buying unlimited volume of the stock at $38 (to artificially support it). From the trade volume during that final half hour, I reckon the underwriters bought well over 50M shares (10%+ of the 420M shares floated).
So... stumbled out of the gate is being kind. It really slithered out like a giant wet turd.
Maybe, but I'd guess that prints have a better chance of surviving generations rather than getting lost due some "mishap" than digital copies. Sure prints *could* get lost due to say fire or flood, but the family photo album doesn't need any more maintainance than stuffing in the bottom drawer.
The current problem with digital photos is that they need ongoing ACTIVE maintainance to not be lost - you need to copy them to new media every few years to avoid media failure and have an adequate backup system for when a failure does occur anyway.
Maybe for geeks digital photo preservation isn't a problem, but who's to say that your (maybe non-technical) heirs will be up to the task, or that you'll remember to put the online back up account details into your will, or ensure the bill gets paid after your death, etc, etc.
I'd say that the optimal strategy is a traditional printed photo album (only containing best photos) in addition to an attempt to preserve everything digitally.
Yes, it's ranked #1 & #2 in the "Cell phones with service plans" category, but the (lower) top 100 rankings in the broader "Cell Phones and Accessories" category is more telling because there it's competing for sales rank with some very cheap high volume accessories.
Either way, it appears (if we can take it at face value) to be selling very well so far.
Actually, the Lumia 900 black and cyan models are both in the Amazon "Cell phones & accessories" top 100 best sellers (accessories includes cheap plastic skinns, etc), which from previous phone releases is indicative of 1M/mo sales rate, assuming it's for real (not just Microsoft/Nokia buying).
It's odd that Nokia is *still* pushing this QML stuff, although it did make sense (to Nokia at least, if not to most Qt developers) while they has been pursuing a Qt-based smartphone strategy.
So the car is able to evaluate the age of a human figure and guess it's predicted motion based on that (i.e. middle age folk arn't likely to step/run in front of a moving car, but young children and old folk may do).
How about recognizing a person in dark clothing sitting in the edge of the road at night time (I've seen it)? Humans excel at visual recognition tasks; computers not so much.
Nope. The problem isn't reaction time, but rather predicting human behavior. The automomous car would likely only break if it saw the child running out into the road, by which time it may already be too late to brake.
The human has the edge because they can also predict when a dangerous situation is going to happen and react before it does. An autonomous car whose intelligence amounts to an expert system rather than a advanced general intelligenve (AGI - yet to be achieved) is always going to have situations where it fails because the situation didn't fall into it's playbook.
It's not a matter of the situations one can easily anticipate, such as pedestrians in the road, but rather the ones that havn't been anticipated. You may not have noticed, but human behavior is rather varied and can't simply be enumerated as a checklist.
For example, if a human driver sees a ball rolling out into the street, you know to slow down immediately since there's a good chance there's a child about to come running out after it, but the computer -if it hadn't been pre-programmed for that scenario (or had human level AI) - wouldn't react until it saw the child, when it may be too late to brake.
So you program that one in... but forget to program the fact that an oncoming driver may swerve to avoid an animal, etc, etc. In general it's the cases where you need to predict human behavior (not react to current conditions) that are going to be a safety nightmare.
The trouble is the real world is messy and unpredictable and you can't simply enumerate and pre-program optimal responses for all situations - you ultimately need human-level AI to fallback on if you're not going to fail in cases where a human would not.
Slashdot Mirror
Flare is one thing, but PURPLE flare is rather obnoxious. Look at the examples linked from the story... iPhone4 gives a bit of flare (saturation) as expected, but iPhone 5 gives the startling purple flare.
5. Do nothing to fix existing phones. Quietly fix the problem and roll it out as iPhone 5.1
Yes, but they also state that intent may be gauged by actions (& statements). In practice I'd guess the US doesn't often use this rule because they want to keep taxing you (as a US citizen), but anyone looking to do this would be well advised to reseach it thoroughly before just assuming they'll be able to retain their US citizenship without qualifications.
Saying that OS X killed Linux on the desktop is a misstatement at best since Linux on the desktop never became mainstream. You can't kill what's not yet alive.
However, client-side (but not "desktop") Linux is very much alive and kicking in the shape of Android tablets (Kindle, Nook, Nexus 7), and currently owns the small form factor tablet space. Apple is the one trying to play catch-up here with the rummoured upcoming iPad mini.
It's interesting to note that the success of Android is following the path led by Apple with OS X ... first on the smart phone, then on the tablet. This was an easier path to mainstream adoption since it wasn't fighting the entrenched desktop ecosystems head-on, but rather building a brand new (smartphone) market. In this vein it might be better to view Nokia's incompetence as the real killer of pure (non-Android) Linux since they were the only ones targetting a Linux-based smartphone. If Nokia had moved faster and followed a release early, release often" path (as Android did), they could well have been successful.
Sure it (shorting against the box) can be used to defer/reduce taxes. That's one common use of it.
Say you have a stock that you'd like to sell, but you don't want to pay taxes this year. What you do is instead short it ("against the box" since you still keep your original shares), thereby locking in the current price. Since neither your long or short position is closed, you currently pay no taxes. Then, whever you want to take your profit (e.h. next tax year, or when you have offsetting losses), you cover your short position and sell your shares, thereby becoming liable for any taxes owed.
That the way to defer taxes...but you can also reduce taxes to ZERO by simply NEVER closing the positions.... You buy a stock (or get in in an IPO), then rather that sell it short againt the box, thereby getting your cash, and NEVER close either long/short position, thereby never incurring a tax liability. People actually do this - it's not just a theoretical tax hole.
A giant mech warrior that throws chairs and can rip cars in half would be epic. A stress indicator butterfly? Not so much.
OTOH it might help protect the microsofties from Ballmer if they could get him to wear one.
Not just for power users ...
In Linux it's simple enough to, say, mount your root (OS data) folder on an SSD and /home (user data) on a HDD, but Windows 7 isn't so flexible.
What most people (power users) end up doing under Windows 7 is to install the OS on an SSD, then use a "junction point" (cf Linux hard link) to redirect the /Users folder to a HDD (and reconfigure the Windows TEMP directory to be on the HDD to avoid killing the SSD with excessive temp file create/delete cycles). The trouble with this is that Windows 7 junction points don't play nice with restore points, as you'll find out when having to revert to a restore point and all your user data disappears requiring major hackery to restore.
So, for Windows 7, a HDD with built in flash cache is a MUCH more convenient solution than using a separate SDD - even for a power user.
It depends on what level of (artifical) intelligence you're talking about. If it's amoeba level intelligence, then maybe ML can achieve similar results, but if it's rat or human level intelligence then obviously not.
I think most people take AI to mean something that could minimally pass a Turing test, not a silicon slug.
Support Vector Machines are just a way of performing unsupervised data partitioning/clustering. i.e. you feed a bunch of data vectors into the algorithm and it determines how to split the data into a number of clusters where the members of each cluster are similar to each other and less similar to members of other clusters.
e.g. you feed it (number of wheels, weight) pairs of a lot of vehicles and it might automatically split the data into 3 clusters - light 2-wheeled vehicles, heavy 4-wheeled ones, and very heavy 4-wheeled ones. If you then labelled these clusters as "bikes", "cars" and "trucks" you could in the future use the clustering rules to determine the category a new data point falls into.
This isn't Artificial Intelligence - it's just a data mining/classification technique.
As I understand it, a Higgs Boson compatible with the standard model could have been found at a range of different masses, and the search for it has involved searching the possible mass range until it was either discovered or not.
Assuming that this new discovery is indeed the Higgs Boson as predicted and compatible with the standard model, what is the significance of the particular mass that it has been found to have? Are there any macro-scale predictions that depend on its mass?
I'm not sure they actually wanted to make a vapourware annoucement, but they were forced to by the Nokia turn of events and headlines (Nokia - Microsoft's main Windows phone parter in the process of imploding and just announcing 20% layoffs and reduced sales forcasts).
This event had "last minute" and desperation written all over it. Why would they time it on the day after the Greek elections when the stock market could have been plummeting. Why pre-announce without specifications, prices or a ship date? Why no real demo or applications? Why was presenter Steven Sinofsky (Windows head honcho) so nervous his hands were shaking?
I think they just decided to throw this out at the last minute as a desperate attempt to say "forget Nokia, our mobile strategy isn't dead yet".
It's a lot of money, but it's less than 10% of what Google is currently bringing in. I happen to think that Google will find a way to prosper as the internet changes (not sure how, but I have faith in their flexibility), but a 90% revenue cut certainly would be bad news!
Selling stock you don't have doesn't require options - it's called "naked short selling", and specifically is allowed by the SEC as a form of IPO stabilization.
I don't have any great source for how this went down, although you can google for the greenshoe (aka over allocation) option in the facebook IPO contract as well as the SEC sanction of naked short selling.
There is one site here claiming the same events, but it's really more a matter of assuming that the underwriters were smart enough to take advantage of the greenshoe provision rather than risking their own cash to support the stock. Seems like a fair bet to me.
http://dealbreaker.com/2012/05/facebook-ipo-goes-nowhere-in-exciting-fashion/
Just wanted to make a correction to that last point ... In fact it looks like, despite massive buying to support the stock, the underwriters don't in fact own the stock they bought back(!), and hence there won't be this large pending sale hanging over the market.
The reason for this is a "greenshoe" provision in the facebook IPO which would have allowed the underwriters to oversell the IPO by 15% if they chose to (i.e. to exceed the intended 420M share float by 15% - buying 15% more at IPO price from FB and selling it to the market).
The underwriters can (and appear to have) use this greenshoe provision to stabilize/support the IPO without any risk to themself.
The way this appears to have gone down is:
1) Underwriters sold 15% more stock than they bought from FB (they buy at 37.5, sell at 38). Specifically they sold 480M shares, having bought only 420M shares from FB (420M being the intended IPO size notwithstanding the greenshoe provision).
2) Underwriters having sold (i.e shorted) 60M (480-420) shares they don't own need to buy it back to cover this short. This could happen in two ways, depending on whether the IPO was successful or not (stock went up or down), but in either case it's a no-risk proposition to the underwriters!
2a) If FB stock had gone up above IPO price then the underwriters would exercise the greenshoe provision and cover their 60M short position buy buying an additional 60M shares (the 15% greenshoe provision) from FB at IPO price, therefore incurring zero loss.
2b) If FB stock had gone (or attempted to go) below IPO price, then the underwriters would cover their 60M short position buy buying in the open market at or below IPO price thereby stabilizing the stock price (purchase up to 60M shares). This is what actually happened. There was some underwriter support (buying) just before midday and then from 3:30-4:00, and the volume purchased seems very close to the 15% (60M shares) greenshoe position.
The net result is that the underwriters were able to artificially support the stock by massive buying, but did so at zero risk to themselves, and ended up owning zero stock! In theory they might have had to buy more than the greenshoe provision amount, and therefore been left actually holding some stock, but on the day the numbers seem to have worked perfectly for them.
So - it's a rigged game. The underwriters can't lose whether the stock goes up or down, but the public can. Facebook got greedy with the price such that there was no upside to the stock, so the suckers buying it AT THIS PRICE were the ones taking all the risk and almost guaranteed short term loss.
There were certainly NASDAQ software problems (acknowledged by NASDAQ), but AFAIK that was in order execution confirmations, not order execution itself (i.e. you submitted an order to buy/sell FB and the order executed in timely fashion, but your notification of the order execution was in some cases significantly delayed).
You can't say 100% (vs 99.99%) for sure that the massive buy orders at $38 were the underwriters, but:
1) That's the definition/job of underwriting - to be the buyers of last resort if there's insufficient demand for the stock, either:
a) Pre-IPO (underwriter holds stock it can't sell), or
b) Post-IPO to support IPO price, else massive loss of face due to having mispriced the offering
2) The price action can't realistically be anything other than due to underwriters. A normal stock buyer trys to buy at the lowest price possible, not to support a given price. If you're going to put in a limit order, you put it in below market, not at market. Those "unlimited" buy orders at $38 at 3:30-4:00pm when the stock was trying to go lower were therefore not from someone trying to buy for profit - they were from someone trying to prop up the price at that level), and the only people who do that are underwriters.
You may see more of that same artificial flatlining in the next couple of weeks to some extent, but the underwriters certainly won't support it for too long.
Who knows how long the underwriters will hold that stock before trying to sell it. Quite likely they may wait until FB is included in various indexes (QQQ, S&P 500) which creates artificial demand (index tracking funds MUST buy the stocks in the index)... but in any case that's a lot of stock waiting to get sold...
Total volume for the day was 570M shares traded - the most for an IPO ever. Volume in the last half hour (when all the buying was by the underwriters) was 50M+ (maybe 70M - I think it started around the 500M mark).
Nope ... it was overpriced.
The only thing that prevented the stock from going below the IPO price was MASSIVE underwriter support (i.e. the banks that got paid to do the IPO buying the stock in large volume whenever it hit the IPO price of $38).
Look at the price chart here:
http://finance.yahoo.com/echarts?s=FB+Interactive#symbol=fb;range=1d
See the flatline at $38 from 3:30-4:00pm? That's due to the underwriters buying the stock in massive volume providing support at $38. If you monitored the detailed stock quote during this time (as I did), you'd have seen that the bid volume, normally single or double digit lots spiked to a continuous 99999+ lots (i.e. 10,000,000+ shares) during this time - i.e. the underwriters were essentially buying unlimited volume of the stock at $38 (to artificially support it). From the trade volume during that final half hour, I reckon the underwriters bought well over 50M shares (10%+ of the 420M shares floated).
So... stumbled out of the gate is being kind. It really slithered out like a giant wet turd.
Maybe, but I'd guess that prints have a better chance of surviving generations rather than getting lost due some "mishap" than digital copies. Sure prints *could* get lost due to say fire or flood, but the family photo album doesn't need any more maintainance than stuffing in the bottom drawer.
The current problem with digital photos is that they need ongoing ACTIVE maintainance to not be lost - you need to copy them to new media every few years to avoid media failure and have an adequate backup system for when a failure does occur anyway.
Maybe for geeks digital photo preservation isn't a problem, but who's to say that your (maybe non-technical) heirs will be up to the task, or that you'll remember to put the online back up account details into your will, or ensure the bill gets paid after your death, etc, etc.
I'd say that the optimal strategy is a traditional printed photo album (only containing best photos) in addition to an attempt to preserve everything digitally.
Yep - didn't last long.
Nokia's stock isn't doing so great either!
Yes, it's ranked #1 & #2 in the "Cell phones with service plans" category, but the (lower) top 100 rankings in the broader "Cell Phones and Accessories" category is more telling because there it's competing for sales rank with some very cheap high volume accessories.
Either way, it appears (if we can take it at face value) to be selling very well so far.
Actually, the Lumia 900 black and cyan models are both in the Amazon "Cell phones & accessories" top 100 best sellers (accessories includes cheap plastic skinns, etc), which from previous phone releases is indicative of 1M/mo sales rate, assuming it's for real (not just Microsoft/Nokia buying).
It's odd that Nokia is *still* pushing this QML stuff, although it did make sense (to Nokia at least, if not to most Qt developers) while they has been pursuing a Qt-based smartphone strategy.
So the car is able to evaluate the age of a human figure and guess it's predicted motion based on that (i.e. middle age folk arn't likely to step/run in front of a moving car, but young children and old folk may do).
How about recognizing a person in dark clothing sitting in the edge of the road at night time (I've seen it)? Humans excel at visual recognition tasks; computers not so much.
Nope. The problem isn't reaction time, but rather predicting human behavior. The automomous car would likely only break if it saw the child running out into the road, by which time it may already be too late to brake.
The human has the edge because they can also predict when a dangerous situation is going to happen and react before it does. An autonomous car whose intelligence amounts to an expert system rather than a advanced general intelligenve (AGI - yet to be achieved) is always going to have situations where it fails because the situation didn't fall into it's playbook.
It's not a matter of the situations one can easily anticipate, such as pedestrians in the road, but rather the ones that havn't been anticipated. You may not have noticed, but human behavior is rather varied and can't simply be enumerated as a checklist.
For example, if a human driver sees a ball rolling out into the street, you know to slow down immediately since there's a good chance there's a child about to come running out after it, but the computer -if it hadn't been pre-programmed for that scenario (or had human level AI) - wouldn't react until it saw the child, when it may be too late to brake.
So you program that one in... but forget to program the fact that an oncoming driver may swerve to avoid an animal, etc, etc. In general it's the cases where you need to predict human behavior (not react to current conditions) that are going to be a safety nightmare.
The trouble is the real world is messy and unpredictable and you can't simply enumerate and pre-program optimal responses for all situations - you ultimately need human-level AI to fallback on if you're not going to fail in cases where a human would not.