Hacker News new | ask | show | jobs
by palewire 3013 days ago
I worked on the data analysis for this story, which was published at a Jupyter Notebook of Python code.

https://github.com/datadesk/street-racing-analysis/blob/mast...

If you have any questions, shoot.

3 comments

Do you have suggestions on how I might find information about the _total_ numbers of traffic related fatalities in LA during similar times? (I'm interesting in knowing what fraction of car-related deaths are due to street racing.)

For example, 2016 had 260 deaths from car accidents from one source [0], and only 11 of those were from street racing, less than 5%. (It's possible that the article I linked counted deaths differently than the data you were working with, though, so it's hard to compare the two.) Street racing related deaths are tragic (especially as half of them are _not the drivers_), but I wonder if we are spending proportionally similar amounts to prevent the larger pie-slice of automotive deaths.

Good grief, I feel so callous even asking this kind of question. :-/

0: http://www.latimes.com/local/lanow/la-me-ln-2016-traffic-dea...

I don't think it's callous at all. In fact, it's a sort of higher order empathy. If the goal is to save as many lives as possible, then this is exactly the right way to think about these things.

It's like how most Americans are more likely to die from obesity related diseases than terrorism, but we spend orders of magnitude more resources trying to prevent terrorism.

That is a bad comparison. My weight is under my control, and I don't want the government to interfere. Preventing terrorists from getting in the country and committing attacks is something I expect the government to do.
I think it is highly unlikely to be true, but it could be that the money spent fighting terrorism saves millions of lives every year. That would make it a bargain.

Similarly, if there were no attempt to stop street racing (not even speed limits), we _might_ see thousands more deads due to street racing every year.

If that were true you can compare back to when we didn't spend money on fighting terrorism. Has it changed much? No not really. Terrorism is not a (major) problem in the world if looking at major causes of human loss of life.
Neither is street racing.

Also, I didn’t argue that; I just pointed out that “few people die today, so we shouldn’t spend much money fighting it” isn’t a valid argument because the truth could be “few people die today because we spend so much money fighting it”.

Not callous at all. We all have a limited amount of attention to pay to things, and as a society we have a limited amount of political and monetary capital to direct at problems. So we should focus on those problems that yield the biggest bang for the buck.

The media is frequently a bad actor in this. Since humans aren't generally good at assessing risk, value propositions like "How to keep your kid from being struck by lightning! Film at 11!" are actively counterproductive. I believe media has an ethical responsibility to allocate their resources and grab eyeballs in line with the magnitude and tractability of the problems they cover, but that's not what sells ad spots. :-(

There's a state database compiled by CHP called SWITRS that tracks traffic accidents. You can find it here. http://iswitrs.chp.ca.gov/Reports/jsp/userLogin.jsp

We don't call it out by name in the story, but it's the kind of existing government system that could be expanded to provide a better, official count of street racing deaths.

Our Los Angeles Times team has mined SWITRS in the past for other stories, like these:

http://graphics.latimes.com/la-pedestrians/

http://graphics.latimes.com/la-bike-hit-and-runs/

AP style guidance prefers ‘crash’ to ‘accident’ jfyi
In deaths per mile driven, even a single victim every few years would make street racing the low hanging fruit.
This article (http://ktla.com/2017/04/03/l-a-traffic-deaths-rose-43-percen...) says there were 260 traffic fatalities in 2016 on LA's streets. From the street racing article it appears that roughly 11 of those were related to street racing? So 5% ?
I appreciate your effort to bring some context to the analysis.

However, as we unpack in the story, our total is likely an undercount due to the problems in tracking street racing related crashes.

Just as a reference, I found this statement in a thesis [1]:

> In the event of a collision, neither drivers nor passengers are willing to admit to engaging in street racing. This can partially explain the low incidence of casualties due to street racing reported in the literature. For example, the Office of Traffic Safety in California, in one of its recent publications, acknowledged that fatalities and injuries due to illegal street racing are significantly underreported due to the issues with reporting and suggested a need for a reporting system reform.

Basically, in some of the crashes, the participants are unlikely to call 911. And it's possible that some of the normal (non-street racing) crashes/fatalities are actually street racing-related.

[1] https://ir.lib.uwo.ca/cgi/viewcontent.cgi?article=3107&conte...

And I appreciate that folks have gone out and tried to analyze something which has not been really visible before. Let's say it is 3x more likely, that means its 15%.

I think the story is important and meaningful, I feel the narrative it supplies is disingenuous.

A lot of people die on streets due to cars. That we do track, both motor vehicle and pedestrian fatalities. It is a stupidly large number.

The narrative that "we", the community, the society, the people who exist in the same space and time, need to do something about "this" is, for me, a function of impact.

The ability for the community to take action is a limited resource. There is only so much tax money, there are only so many cops, there is only so much technology. And we have to choose where that finite resource is invested in order to maximize the protection, health, and well being of the community from which that resource is drawn. Because I see the problem this way, any narrative that suggests investing a larger fraction on a smaller problem, makes me wonder how much worse the other problems those resources had been applied to becomes when we take them away to focus on this smaller problem.

So I would ask the people, are you prepared to have over all vehicle fatalities double while fatalities due to street racing go to zero? When street racing is 5, 10, 15, even 25 percent of the problem that calculus means more dead people, more people losing children and loved ones to a vehicle fatality.

There are at least 5 direct calls for resources in that article[1]. They are calls for more police, more tracking, more infrastructure. It is a simple and common narrative, "My problems are really important so you should agree to spend more money on them rather than the other things you are spending them on."

Anyone who has lived in California for a while realizes that there are only two choices, raise taxes to increase the size of the resource pool (and then have factions fight over the raises) or let some existing priority slip while you focus on a different one.

It is also annoying (but understandable) to me that these articles cluster around the March/April time frame because that is after the Governor has submitted their budget and the legislators are figuring out where to send the money.

If the article was arguing for more resources on vehicle fatalities, I could understand that. We've got way too many people who don't even follow basic road rules. So if you spent all your taxes on that and no additional on the street racing scene and achieved a 10% reduction in fatalities you would have saved more lives than completely wiping out street racing would. That makes it a better investment overall for a limited resource.

[1] Passages in the article that suggest that applying more resources to this problem would ameliorate it:

A few law enforcement agencies have assigned officers to the task force. But some agencies say they lack manpower.

Efforts to place a similar unit in the Valley, another racing hot zone, were abandoned for lack of staff, a street racing investigator said.

“You can’t solve a problem that you don’t measure,” Englander said.

Neither the Los Angeles County Sheriff’s Department nor the Long Beach Police Department, two of the largest agencies in the region, have officers assigned to the county task force.

County deputies are being trained to recognize vehicles that have been modified for racing, but, Det. Christine Ostrander said, “our deputies are just overworked, understaffed.”

Well said and absolutely agreed.

Based on the numbers presented, even with the "these numbers are conservative" caveat, focusing resources on this doesn't make sense.

"In 2010 and 2011, 161 and 130 drunk driving fatalities occurred in Los Angeles County, respectively." [1]

So almost a 14:1 drunk driving fatality to street racing ratio. Even if the street racing numbers are way off, they'd have to be 14x as large to have the same impact as DUI.

Which somewhat stands to reason. I assume street racers aren't in the habit of (1) wrecking their expensive and carefully modified cars & (2) dying. Given their interest in racing, I'd even hazard to say they're likely better drivers on average than the (admittedly terrible) public.

[1] https://www.sjclaw.com/Resources-and-Media/Articles/How-many... (claims to be NHTSA data)

No questions, I just wanted to say this is really cool seeing the data with analysis in its final form. I wish more articles included data like this.
Thanks!

If you're a Python person, I'd love to hear what you think of how I structured the notebook.

There isn't a set style for writing these and I've been experimenting with different forms lately.

In this case, I pushed back some of the bigger code chunks into separate Python modules.