Uber halts self-driving tests after pedestrian killed in Arizona

[Chris]

ars technica: Police chief said Uber victim "came from the shadows" –don't believe it

The dashcam footage from the Uber car appears to be very misleading in regards to the darkness on the road. The road is in fact lit and there is much more ambient light than is apparant from the Uber dashcam. The dashcam does show the light bulbs from the street lights (so they were on).

The Uber dashcam also shows only 1.5 seconds between the first view of the pedestrian and the crash. The headlights of the Volvo are also capable of illuminating the road much further ahead than is apparent on the Uber dashcam video. The dashcam video implies that the headlights of the Volvo have a range of less than 100 ft. In reality, this car's headlights have a range of at least 250 ft at low beam.

Regardless of the lighting situation, lidar should have been able to detect the object on the road, this is in fact one of the easiest circumstances, the pedestrian did not just walk from behind a parked car, but had already crossed one lane so was on the road for a considerable amount of time.

So both a human driver and the autonomous systems should've been able to avoid this accident.

The Uber dashcam footage:


Video footage by other motorists of that same location:

[kalvado]

ars technica: Police chief said Uber victim "came from the shadows" –don't believe it

The dashcam footage from the Uber car appears to be very misleading in regards to the darkness on the road. The road is in fact lit and there is much more ambient light than is apparant from the Uber dashcam. The dashcam does show the light bulbs from the street lights (so they were on).

The Uber dashcam also shows only 1.5 seconds between the first view of the pedestrian and the crash. The headlights of the Volvo are also capable of illuminating the road much further ahead than is apparent on the Uber dashcam video. The dashcam video implies that the headlights of the Volvo have a range of less than 100 ft. In reality, this car's headlights have a range of at least 250 ft at low beam.

Regardless of the lighting situation, lidar should have been able to detect the object on the road, this is in fact one of the easiest circumstances, the pedestrian did not just walk from behind a parked car, but had already crossed one lane so was on the road for a considerable amount of time.

So both a human driver and the autonomous systems should've been able to avoid this accident.

The Uber dashcam footage:
(trimmed)

Video footage by other motorists of that same location:
(trimmed)

Well, vehicle is coming from underpass. Road is likely not level, and grade can change pretty quickly. That means distance to edge of headlights spot can vary quite a bit. As a contributing factor, high-end car like Volvo could have an active control of headlights beam direction.
As for illumination.. relatively low posts with fixtures barely over the edge of the road...  Such lights are visibility killers, not aids for me.
And, apparently, pedestrian should yield to traffic while crossing outside of a crosswalk area - in fact in area with a "do not walk" sign. So legal case will not be cut and dry, and likely would be settled out of court.
Of course, tweaking video and comparison with high-gain cameras is useful to create public perception of oh-so-evil Uber and increasing amount of settlement.

Real question is why sensors didn't see pedestrian - or why control software didn't respond properly.

[jeffandnicole]

The two pics were taken at different times; maybe has high beams on, street lights may not have been working the night of the accident, etc.

The comparison photos are meaningless.

[DaBigE]

The two pics were taken at different times; maybe has high beams on, street lights may not have been working the night of the accident, etc.

The comparison photos are meaningless.

Not to mention the different angles/vantage points. Is the Uber really from a dash cam? The angle makes it seem more like it was from a camera mounted in or near the grille. And there's always the potential the image could have been doctored (intentionally or unintentionally), depending on the quality/type of camera used to record the image.

[tradephoric]

The low light capabilities of cameras can vary widely.  Even the best performing low light cameras can produce dark footage at high f-stops.  Here is a picture from a photography website showing different aperture settings of the same night scene:



The point is there is no way of knowing what the human driver would have seen just by watching that Uber footage.  An interesting test would be to replicate the Uber crash as closely as possible with a low budget dash cam... then mount a $2000 Sony A7S II with great low light capability and optimum aperture settings and post both recreation videos side-by-side.  People may have a different opinion about the Uber footage then, and they may no longer believe the Police Chief when she says the victim "came from the shadows".

[jeffandnicole]

The low light capabilities of cameras can vary widely.  Even the best performing low light cameras can produce dark footage at high f-stops.  Here is a picture from a photography website showing different aperture settings of the same night scene:



The point is there is no way of knowing what the human driver would have seen just by watching that Uber footage.  An interesting test would be to replicate the Uber crash as closely as possible with a low budget dash cam… then mount a $2000 Sony A7S II with great low light capability and optimum aperture settings and post both recreation videos side-by-side.  People may have a different opinion about the Uber footage then, and they may no longer believe the Police Chief when she says the victim "came from the shadows".

When/if they recreate accident scenes, they use humans, not cameras.  Cameras may be part of the recreation to document what they are doing, but they are using actual people to determine what would've been seen, based on the exact conditions that existed; not conditions that should have existed. 

Of course, in what you said, you are already believing that the Chief lied, and that a more expensive camera will prove that.  In of itself, that's a prejudicial remark, because you don't want to think that the jaywalking pedestrian was in the wrong to begin with.  You don't want a study; you want vindication for your own thoughts.

[kalvado]

The low light capabilities of cameras can vary widely.  Even the best performing low light cameras can produce dark footage at high f-stops.  Here is a picture from a photography website showing different aperture settings of the same night scene:

The point is there is no way of knowing what the human driver would have seen just by watching that Uber footage.  An interesting test would be to replicate the Uber crash as closely as possible with a low budget dash cam... then mount a $2000 Sony A7S II with great low light capability and optimum aperture settings and post both recreation videos side-by-side.  People may have a different opinion about the Uber footage then, and they may no longer believe the Police Chief when she says the victim "came from the shadows".

Realistically speaking, most people on this site have their license and thousands hours behind the wheel, so lousy video is not the only source of information for us. Uber's camera is not great, but I would say it emulates eye better than high-end one. Crash happened around 10 PM, about 3 hours after sunset (6.39 PM) - so it was actually dark.
There is a link to the accident location somewhere upstream, let me post it again (slightly back from the crash location): https://www.google.com/maps/@33.435806,-111.9420409,3a,75y,347h,98.91t/data=!3m6!1e1!3m4!1sHTvoeeS6NPEocIyYNvrCyA!2e0!7i13312!8i6656
You can clearly see light fixtures are pretty low, and not wide angle. There are 4 on right side of the road above sidewalk, and a lone one on the left..  I would call those sidewalk illumination, nothing else.
Pedestrian is crossing past that lone light source on the left - and looks to me as if there is a tree blocking light next to that pole. There is a pretty sharp shadow line visible in video. That would make for a challenging detection for a human driver, sure.
It just happened that the oncoming car was a self-driving one; situation would be very clear-cut for a human driver.

[kalvado]

The low light capabilities of cameras can vary widely.  Even the best performing low light cameras can produce dark footage at high f-stops.  Here is a picture from a photography website showing different aperture settings of the same night scene:

The point is there is no way of knowing what the human driver would have seen just by watching that Uber footage.  An interesting test would be to replicate the Uber crash as closely as possible with a low budget dash cam... then mount a $2000 Sony A7S II with great low light capability and optimum aperture settings and post both recreation videos side-by-side.  People may have a different opinion about the Uber footage then, and they may no longer believe the Police Chief when she says the victim "came from the shadows".

When/if they recreate accident scenes, they use humans, not cameras.  Cameras may be part of the recreation to document what they are doing, but they are using actual people to determine what would've been seen, based on the exact conditions that existed; not conditions that should have existed. 

Of course, in what you said, you are already believing that the Chief lied, and that a more expensive camera will prove that.  In of itself, that's a prejudicial remark, because you don't want to think that the jaywalking pedestrian was in the wrong to begin with.  You don't want a study; you want vindication for your own thoughts.

It is natural for people to protect a poor little guy from powerful and rich evil corporation.
The unwanted part of it is effectively saying that jaywalking in challenging visibility is OK. Those poor souls who choose to believe that message (may them rest in peace) wouldn't make it to headlines, though. So side effects will not be understood and lessons will not be learned...

[jeffandnicole]

It is natural for people to protect a poor little guy from powerful and rich evil corporation.
The unwanted part of it is effectively saying that jaywalking in challenging visibility is OK. Those poor souls who choose to believe that message (may them rest in peace) wouldn't make it to headlines, though. So side effects will not be understood and lessons will not be learned...

I also noticed in this case that rather than just saying 'a better camera', one takes the time to mention a very specific camera, with specific capabilities.  I'm sure I can adjust my cell phone camera's settings to eventually produce a general image close to one that will support my opinion. 

[kalvado]

It is natural for people to protect a poor little guy from powerful and rich evil corporation.
The unwanted part of it is effectively saying that jaywalking in challenging visibility is OK. Those poor souls who choose to believe that message (may them rest in peace) wouldn't make it to headlines, though. So side effects will not be understood and lessons will not be learned...

I also noticed in this case that rather than just saying 'a better camera', one takes the time to mention a very specific camera, with specific capabilities.  I'm sure I can adjust my cell phone camera's settings to eventually produce a general image close to one that will support my opinion.

The way I read trade's message is "better camera could have saved the day - and look, those are available". And frankly speaking, Sony matrices are among the best (with all my dislike to that brand). That is a pretty reasonable thing to say (although I assume camera is not the primary source of data)
And yes, why electronics didn't see pedestrian is THE question for me. And I'm not sure if the camera providing this recording is in fact a part of a system, or just a simple off-the-shelf camera acting an independent witness for cases when things go wrong (or very wrong, as in this case). Which is  a reasonable thing to do - setup an independent recorder in addition to main system.

[tradephoric]

Of course, in what you said, you are already believing that the Chief lied, and that a more expensive camera will prove that.  In of itself, that's a prejudicial remark, because you don't want to think that the jaywalking pedestrian was in the wrong to begin with.  You don't want a study; you want vindication for your own thoughts.

When watching the Uber video, it does look like the pedestrian "came out of the shadows" .  At face value, the Chief's initial description of the crash seems totally valid.  But I question if the Chief's opinion of the crash would have been different if the fatal crash was recorded on a high-end camera with better low light capabilities.  Does the pedestrian "come out of the shadows"  if the aperture settings on the camera had been set higher?  Unfortunately, the video camera settings used may largely dictate how the public views this fatal accident.

[kalvado]

Of course, in what you said, you are already believing that the Chief lied, and that a more expensive camera will prove that.  In of itself, that's a prejudicial remark, because you don't want to think that the jaywalking pedestrian was in the wrong to begin with.  You don't want a study; you want vindication for your own thoughts.

When watching the Uber video, it does look like the pedestrian "came out of the shadows" .  At face value, the Chief's initial description of the crash seems totally valid.  But I question if the Chief's opinion of the crash would have been different if the fatal crash was recorded on a high-end camera with better low light capabilities.  Does the pedestrian "come out of the shadows"  if the aperture settings on the camera had been set higher?  Unfortunately, the video camera settings used may largely dictate how the public views this fatal accident.

Unlike most of us, chief did attend the spot of accident in person and seen it in pretty much same lighting conditions as it was at the time of accident. No need to guess camera settings, everything was presented first hand for careful observation. You don't have to trust that opinion - but you cannot deny that it is an educated one, as police chief likely seen tens, if not hundreds, of such situations during years of police career (and "chief" means some experience for sure).

[tradephoric]

Of course, in what you said, you are already believing that the Chief lied, and that a more expensive camera will prove that.  In of itself, that's a prejudicial remark, because you don't want to think that the jaywalking pedestrian was in the wrong to begin with.  You don't want a study; you want vindication for your own thoughts.

When watching the Uber video, it does look like the pedestrian "came out of the shadows" .  At face value, the Chief's initial description of the crash seems totally valid.  But I question if the Chief's opinion of the crash would have been different if the fatal crash was recorded on a high-end camera with better low light capabilities.  Does the pedestrian "come out of the shadows"  if the aperture settings on the camera had been set higher?  Unfortunately, the video camera settings used may largely dictate how the public views this fatal accident.

Unlike most of us, chief did attend the spot of accident in person and seen it in pretty much same lighting conditions as it was at the time of accident. No need to guess camera settings, everything was presented first hand for careful observation. You don't have to trust that opinion - but you cannot deny that it is an educated one, as police chief likely seen tens, if not hundreds, of such situations during years of police career (and "chief" means some experience for sure).

The footage inside the Uber of the fatal crash is obviously a key piece of evidence in the investigation.  The Police Chief made her remarks of the pedestrian "coming out of the shadows"  after viewing that Uber video.  I do believe she got ahead of herself with those comments, but the facts will come out when the NTSB releases their final report (after all it's not just the Tempe Police investigating this crash).   My guess is the NTSB will recreate the crash, using high end cameras that far exceed the low light capabilities of a dash cam.  I'm curious to watch their recreation video if/when it is released.

[kalvado]

The footage inside the Uber of the fatal crash is obviously a key piece of evidence in the investigation.  The Police Chief made her remarks of the pedestrian "coming out of the shadows"  after viewing that Uber video.  I do believe she got ahead of herself with those comments, but the facts will come out when the NTSB releases their final report (after all it's not just the Tempe Police investigating this crash).   My guess is the NTSB will recreate the crash, using high end cameras that far exceed the low light capabilities of a dash cam.  I'm curious to watch their recreation video if/when it is released.

And high end cameras may show everything in finest details - but it wouldn't be what driver saw.
Tempe police definitely was on the spot of accident as first responders, either Tempe or AZ state evidence collection team did work the site. Issue probably escalated pretty quick - but NTSB cannot dispatch investigation team at midnight.
Overall situation is pretty clear, reconstruction wouldn't add much to understanding here.

And I think we're loosing focus here. Here is how I see situation:
-Uber was testing self-driving system. Such system should, by all means, notice pedestrian. It didn't. Why? 1st contributing factor.
-Uber sort of anticipated such problem and did due diligence by putting a human backup driver. Who may or may not paid full attention, but generally provides a layer of backup.
-And backup driver failed to react - possibly due to being unable to spot pedestrian on time. 2nd contributing factor. Was it actually so difficult to spot?(i suspect yes).
-Accident could be avoided if pedestrian would be more noticeable to backup driver. Jaywalking in dark clothes in the dark without yielding to traffic. 3d contributing factor
I think 1st one should be investigated in great depth, second and third are more common than we want them to be. Since video provides insight into 2 and 3 only, we tend to discuss those.. but honestly speaking - such accidents probably occur daily without attracting too much attention.

[tradephoric]

And high end cameras may show everything in finest details - but it wouldn't be what driver saw.
Tempe police definitely was on the spot of accident as first responders, either Tempe or AZ state evidence collection team did work the site. Issue probably escalated pretty quick - but NTSB cannot dispatch investigation team at midnight.
Overall situation is pretty clear, reconstruction wouldn't add much to understanding here.

And a low end camera could show everything in poor detail - but it wouldn't be what driver saw.  But in my humble opinion, a high end camera with good low light capabilities would more closely match the details of what a human pupil would see...

The media narrative is that nobody could have seen the pedestrian in time, therefore the pedestrian would have been struck even if an attentive human driver was behind the wheel.  That narrative is based on the dash cam footage of the crash that was released to the public.  But if the camera had better low light capabilities that clearly showed the pedestrian crossing the street several seconds before being struck, that narrative would quickly change.  That's why a detailed recreation of this fatal crash could change people's perception of what happened.

Look at it this way.  Imagine if a vehicle plowed down a pedestrian in broad daylight, but the vehicles dash cam was set up so it looked nearly pitch dark and very little detail could be seen.  Would that dash footage be effective in court to prove that the pedestrian couldn't be seen and that it wasn't the driver's fault that the pedestrian got struck?  Of course not.  It's important to understand the limitations of the camera that was being used when this fatal crash occurred since so much opinion of the crash is gathered from that video. 

[kalvado]

And high end cameras may show everything in finest details - but it wouldn't be what driver saw.
Tempe police definitely was on the spot of accident as first responders, either Tempe or AZ state evidence collection team did work the site. Issue probably escalated pretty quick - but NTSB cannot dispatch investigation team at midnight.
Overall situation is pretty clear, reconstruction wouldn't add much to understanding here.

The media narrative is that nobody could have seen the pedestrian in time, therefore the pedestrian would have been struck even if an attentive human driver was behind the wheel.  That narrative is based on the dash cam footage of the crash that was released to the public.  But if the camera had better low light capabilities that clearly showed the pedestrian crossing the street several seconds before being struck, that narrative would quickly change.  That's why a detailed recreation of this fatal crash could change people's perception of what happened.

Look at it this way.  Imagine if a vehicle plowed down a pedestrian in broad daylight, but the vehicles dash cam was set up so it looked nearly pitch dark and very little detail could be seen.  Would that dash footage be effective in court to prove that the pedestrian couldn't be seen and that it wasn't the driver's fault that the pedestrian got struck?  Of course not.  It's important to understand the limitations of the camera that was being used when this fatal crash occurred since so much opinion of the crash is gathered from that video.

Well, you have a statement of someone who likely saw everything first hand. Regardless of how shity video is, I been in similar situations as a driver, I know how it looks from behind the wheel  - and that part of the story doesn't bother me too much. Uber will pay the family anyway.

[jeffandnicole]

And I think we're loosing focus here. Here is how I see situation:
-Uber was testing self-driving system. Such system should, by all means, notice pedestrian. It didn't. Why? 1st contributing factor.
-Uber sort of anticipated such problem and did due diligence by putting a human backup driver. Who may or may not paid full attention, but generally provides a layer of backup.
-And backup driver failed to react - possibly due to being unable to spot pedestrian on time. 2nd contributing factor. Was it actually so difficult to spot?(i suspect yes).
-Accident could be avoided if pedestrian would be more noticeable to backup driver. Jaywalking in dark clothes in the dark without yielding to traffic. 3d contributing factor
I think 1st one should be investigated in great depth, second and third are more common than we want them to be. Since video provides insight into 2 and 3 only, we tend to discuss those.. but honestly speaking - such accidents probably occur daily without attracting too much attention.

Exactly.  The investigation really shouldn't focus too much on what the driver saw or not see, but rather what the car's internal systems saw or didn't see.  The accident that occurred is fairly common, and we're fortunate enough to have any video evidence of it in this case.  If the vehicle was a normal vehicle, most of us wouldn't even be aware of this accident, much less trying to talk about camera quality decide how much lighting there was.

Remember, in Trade's numerous roundabout postings, he commonly deflects accident causes at roundabouts to the fact the roundabout was there in the first place. In nearly every case, the investigation won't conclude that the roundabout itself was the cause of the accident.  In this case, he's trying to deflect the accident towards a very pricey camera that will have settings to make it seem brighter than what it was.

The key here is the pedestrian shouldn't have been there whatsoever, which is going to largely be taken into account.  While we can debate about videos, the pedestrian could easily have seen the headlights of the vehicle approaching.  Even if conditions allowed the driver a full view of the pedestrian, the moment the driver processed that info and slammed onto the brakes may be too little too late.  Could the driver have swerved out of the way?  Maybe, maybe not.  Was there a curb or wall there to the right, preventing the driver from fully missing the pedestrian, or being bounced back into the pedestrian?  Could the driver have swerved the other direction, anticipating that the pedestrian would have continued on her path?   If the driver swerved to the left and the pedestrian stopped, it would appear the driver aimed towards the pedestrian.

And yes, I know the first thing that'll come to mind: The driver should've been driving slower.  The vehicle was going about the speed limit.  Might as well make every roadway a 5 mph speed limit to prevent such tragedies.

But again, the key here is the pedestrian was jaywalking.  The 2nd key is that the car's electronic system didn't pick that person up.  Why?

[kalvado]

Remember, in Trade's numerous roundabout postings, he commonly deflects accident causes at roundabouts to the fact the roundabout was there in the first place. In nearly every case, the investigation won't conclude that the roundabout itself was the cause of the accident.  In this case, he's trying to deflect the accident towards a very pricey camera that will have settings to make it seem brighter than what it was.

As with roundabout, fact that the road was designed in such a way definitely creates the framework for an accident. Certain design solutions do create extra hazards. Which would be avoided most of the time, but sometimes somewhere something doesn't add up. For this particular case:

There is a peculiar feature of this roadway as discussed above - an X shaped paved island in the median. Which apparently creates a jaywalk-prone spot, enough to warrant some signs. Is that a contributing factor here? Somewhat. Does it deserve investigative efforts? Probably not. Would you recommend installing those elsewhere? Me not...

[jeffandnicole]

There is a peculiar feature of this roadway as discussed above - an X shaped paved island in the median. Which apparently creates a jaywalk-prone spot, enough to warrant some signs. Is that a contributing factor here? Somewhat. Does it deserve investigative efforts? Probably not. Would you recommend installing those elsewhere? Me not...

Very interesting - looks like they were there back in 2007; with the current curbing being installed just after 2008.  I'm guessing they built and/or refurbishing the bridges just to the south of this location.  The No Ped Crossing signs have been up the whole time.  Not sure why they would leave those crossovers there.

That said, the signage certainly does reinforce that pedestrians shouldn't be crossing there. And depending on the exact point of crossing, shows that a driver would've had very little time to react if the pedestrian came out of a marked "No Pedestrian Crossing" area hidden by existing shrubbery and foliage.  Bicyclists and pedestrians wouldn't benefit from crossing at this point with a signal so close.

[kalvado]

There is a peculiar feature of this roadway as discussed above - an X shaped paved island in the median. Which apparently creates a jaywalk-prone spot, enough to warrant some signs. Is that a contributing factor here? Somewhat. Does it deserve investigative efforts? Probably not. Would you recommend installing those elsewhere? Me not...

Very interesting - looks like they were there back in 2007; with the current curbing being installed just after 2008.  I'm guessing they built and/or refurbishing the bridges just to the south of this location.  The No Ped Crossing signs have been up the whole time.  Not sure why they would leave those crossovers there.

That said, the signage certainly does reinforce that pedestrians shouldn't be crossing there. And depending on the exact point of crossing, shows that a driver would've had very little time to react if the pedestrian came out of a marked "No Pedestrian Crossing" area hidden by existing shrubbery and foliage.  Bicyclists and pedestrians wouldn't benefit from crossing at this point with a signal so close.

Yet it is a feature which possibly affected the situation. Now since that wasn't roundabout, it was not critical...

[DaBigE]

Since human vs. electronic capabilities have been brought up a few times, I'll throw this out there: It's pretty-much universally assumed that electronic means of detection are more sensitive and better able at detecting objects and scenarios that a human cannot or would take much longer to detect/process. What about in the eyes of the law? Are the electronic devices only responsible for what the average human would be capable of detecting and responding to? In the Tempe case, if a human would not have been able to react in time, are the electronic devices legally held to a higher standard? Are there Federal standards/laws or industry standards?

[kalvado]

Since human vs. electronic capabilities have been brought up a few times, I'll throw this out there: It's pretty-much universally assumed that electronic means of detection are more sensitive and better able at detecting objects and scenarios that a human cannot or would take much longer to detect/process. What about in the eyes of the law? Are the electronic devices only responsible for what the average human would be capable of detecting and responding to? In the Tempe case, if a human would not have been able to react in time, are the electronic devices legally held to a higher standard? Are there Federal standards/laws or industry standards?

I suspect this is THE question in legal domain. Something similar was discussed for quite a while, although nobody envisioned such stupid and simple accident.
Probably one of the reasons Uber would try to settle out of court is that a precedent - or lack of it - would affect lots of things. 

[vdeane]

I do find it interesting that everyone is focusing on the jaywalking pedestrian.  Not really surprising, though.  The government (Arizona in particular), the media, and the car and technology industries all have vested interests in getting self-driving cars deployed as quickly as possible.  I have NEVER seen the media blame a pedestrian for jaywalking before, but again, not surprising - it keeps people from questioning the technology.  The people who want self-driving cars want to downplay that part of this story because they know that self-driving cars could be set back a decade or more if public outcry is big enough.  Just look at the AV START act.  It legalizes self-driving cars nationwide and preempts any state or local regulation regarding them, and allows the industry to essentially self-regulate.  Just a few months ago, it was expected to sail through Congress unopposed.  Now it's already bogged down in the Senate, and depending on how things shape up with respect to this incident, could very well fail.  They certainly do not want that.  So instead they have everyone talk about the jaywalking.

[jeffandnicole]

I do find it interesting that everyone is focusing on the jaywalking pedestrian.  Not really surprising, though.  The government (Arizona in particular), the media, and the car and technology industries all have vested interests in getting self-driving cars deployed as quickly as possible.  I have NEVER seen the media blame a pedestrian for jaywalking before, but again, not surprising - it keeps people from questioning the technology.  The people who want self-driving cars want to downplay that part of this story because they know that self-driving cars could be set back a decade or more if public outcry is big enough.  Just look at the AV START act.  It legalizes self-driving cars nationwide and preempts any state or local regulation regarding them, and allows the industry to essentially self-regulate.  Just a few months ago, it was expected to sail through Congress unopposed.  Now it's already bogged down in the Senate, and depending on how things shape up with respect to this incident, could very well fail.  They certainly do not want that.  So instead they have everyone talk about the jaywalking.

When I see accidents involving pedestrians in the paper, I tend to find this pattern:

If the motorist was at fault, there will usually be one, if not several, follow-up articles.

If the pedestrian was at fault, there will never, ever be another follow-up article.

[vdeane]

Around here, there have been several jaywalking incidents leading to many media stories blaming drivers.  One in particular stands out because the media refused to admit that the person who died was jaywalking until months after the fact.