Do you have adblock enabled?
 
If you can read this, either the style sheet didn't load or you have an older browser that doesn't support style sheets. Try clearing your browser cache and refreshing the page.

(The Atlantic)   Apparently we're going to be experiencing a software apocalypse soon, so we have that to look forward to   ( theatlantic.com) divider line
    More: PSA, code, Software engineering, software, Computer programming, Programming language, Computer program, Computer, Programmer  
•       •       •

3629 clicks; posted to Geek » on 30 Sep 2017 at 2:24 AM (2 weeks ago)   |   Favorite    |   share:  Share on Twitter share via Email Share on Facebook   more»



65 Comments     (+0 »)
 
View Voting Results: Smartest and Funniest


Oldest | « | 1 | 2 | » | Newest | Show all

 
2017-09-30 02:48:07 AM  
I guess it was inevitable that the doomsayers would move on to that since they've been failing at predicting an actual IRL apocalypse for around 2000 years now.
 
2017-09-30 03:52:33 AM  
"So, that's a big problem."

steamuserimages-a.akamaihd.net
 
2017-09-30 03:59:02 AM  
Resig is a celebrated programmer of JavaScript

Someone celebrates JavaScript?
 
2017-09-30 04:56:37 AM  
I'm looking forward to the day when computer programmers are obsolete.
 
2017-09-30 07:50:35 AM  
They're just now discovering OOP?
 
2017-09-30 07:53:20 AM  
Y2k?
 
2017-09-30 08:01:00 AM  

Spectrum: "So, that's a big problem."

[steamuserimages-a.akamaihd.net image 250x272]


Too bad, because you missed the best part. The first half of the article is a description of the problem. The second half describes the work people are doing that might solve it. TFA+ (the tool in the article, not The F***ing Article) sounds quite interesting.
 
2017-09-30 08:03:31 AM  
Oop

jjorsett: Spectrum: "So, that's a big problem."

[steamuserimages-a.akamaihd.net image 250x272]

Too bad, because you missed the best part. The first half of the article is a description of the problem. The second half describes the work people are doing that might solve it. TFA+ (the tool in the article, not The F***ing Article) sounds quite interesting.


Oops, TLA+, not TFA. I've been around internet-speak too long.
 
2017-09-30 08:30:52 AM  
Didn't Windows 3.1 use an integer clock that would freeze the system after 2^31 ticks?

And who doesn't let the database assign new call numbers?
 
2017-09-30 08:56:40 AM  
,Marcus Aurelius: And who doesn't let the database assign new call numbers?

Say they define call number as an 8 digit integer. There is nothing available after 99,999,999.
 
2017-09-30 09:00:04 AM  

jaytkay: ,Marcus Aurelius: And who doesn't let the database assign new call numbers?

Say they define call number as an 8 digit integer. There is nothing available after 99,999,999.


There's a Herman Cain joke somewhere in there, but I haven't had enough coffee to figure it out.
 
2017-09-30 09:09:26 AM  

jaytkay: ,Marcus Aurelius: And who doesn't let the database assign new call numbers?

Say they define call number as an 8 digit integer. There is nothing available after 99,999,999.


A new standard I've been seeing lately appears to be using 128-bit guids instead of integers. Very easy to generate by simply hashing the stored data.
 
2017-09-30 09:14:14 AM  
FTA: In September 2007, Jean Bookout was driving on the highway with her best friend in a Toyota Camry when the accelerator seemed to get stuck. When she took her foot off the pedal, the car didn't slow down. She tried the brakes but they seemed to have lost their power. As she swerved toward an off-ramp going 50 miles per hour, she pulled the emergency brake. The car left a skid mark 150 feet long before running into an embankment by the side of the road. The passenger was killed. Bookout woke up in a hospital a month later.

First, people need to stop referring to it as "the emergency brake".  The proper nomenclature is "parking brake" or "handbrake".  Second, the best way out of that situation is to shift the vehicle into neutral whereupon the vehicle operator can steer the vehicle into a controlled stop.  Yes, this may involve colliding with a curb or rolling the vehicle in a ditch, but the alternative is obviously worse.  The engine may or may not redline depending on whether or not the rev-limiter activates.
 
2017-09-30 09:21:36 AM  
If I'm reading this correctly the problem is that programmers tend to think they're this guy:
img.fark.net

when in fact they're this guy:
vignette.wikia.nocookie.net
 
2017-09-30 09:40:21 AM  
In other news, Java programmer re-invents the concept of ADA, Decides loading huge linked libraries and going full copy pasta is a bad idea for mission critical software.

Stay tuned when java programmers read the disclaimer license on most software languages, the parts where it says "Hey our shiat totally isn't up to the standards of keeping people alive, or anything where nuclear explosions might happen."

Yes all the major languages have that disclaimer, thee are very few languages that are made to be in systems that can cause havoc, death & loss of life.


>Voiceofreason01
>If I'm reading this correctly the problem is that programmers tend to think they're this
> guy:
 
2017-09-30 10:00:22 AM  
This is why BIT is so goddamn important and log/anomaly management. The issue isn't SW it is failure to assess risk and mitigate risk from a SW prospective.
 
2017-09-30 10:07:49 AM  

Xzano: In other news, Java programmer re-invents the concept of ADA, Decides loading huge linked libraries and going full copy pasta is a bad idea for mission critical software.

Stay tuned when java programmers read the disclaimer license on most software languages, the parts where it says "Hey our shiat totally isn't up to the standards of keeping people alive, or anything where nuclear explosions might happen."

Yes all the major languages have that disclaimer, thee are very few languages that are made to be in systems that can cause havoc, death & loss of life.


>Voiceofreason01
>If I'm reading this correctly the problem is that programmers tend to think they're this
> guy:


I'm really hoping that problem doesn't happen to Python or Rust.  It's kind of starting to happen to Python where people rely on 3rd party packages like NumPy.  What little Python I've written has relied on NumPy or PyQT to do the heavy lifting.

Python3 + QT makes for easy to write GUIs....what the actual buttons are supposed to do....still working on that part....

QT Designer is a pretty neat WYSIWYG editor for making QT application layouts and with Python or C++ compatible code.  Decent place to start if one wants to learn programming in their spare time (combined with actual programming books and various online videos & tutorials of course).
 
2017-09-30 10:21:55 AM  

big pig peaches: Y2k?


Kinda.

Arbitrary limits have a tendency to bite people in the ass.  Even if they don't get the same amount of media attention as Y2K did.
 
2017-09-30 10:26:48 AM  

Marcus Aurelius: Didn't Windows 3.1 use an integer clock that would freeze the system after 2^31 ticks?

And who doesn't let the database assign new call numbers?


Lots places have gas pumps that can't handle prices above $99.99.
 
2017-09-30 10:30:40 AM  
"United Airlines grounded its fleet because of a problem with its departure-management system; trading was suspended on the New York Stock Exchange after an upgrade; the front page of The Wall Street Journal's website crashed; and Seattle's 911 system went down again"

One of these things is not as disastersous as the others.
 
2017-09-30 10:41:12 AM  
Heh.  This is shockingly relevant.  I work for a video distribution company.  Just two weeks ago we had a purge job run that deleted tens of thousands of critical video files.  The purge job had run regularly for more than two years without any issue, usually deleting no more than a hundred or so files.  This time it ran simultaneously along with another task that interfered with date and usage calculations.  The result?  Nearly a year of lost work for our encoding team.  It could have been stopped if someone had built in a manual approval check into the script.  Are you deleting more than 500 files?  Stop what you're doing and get manual approval.
 
2017-09-30 10:46:09 AM  

Flab: Lots places have gas pumps that can't handle prices above $99.99.


Gas pumps can go straight to hell.  If you've ever tried programming one, you would agree.
 
2017-09-30 10:51:32 AM  

ScarabDrac: A new standard I've been seeing lately appears to be using 128-bit guids instead of integers. Very easy to generate by simply hashing the stored data.


I often have to dole a number out before the data arrives.  I have SQL code from about 1990 that lets you generate based on object type plus a text-y-m-d-j autonumber format mask.  I've re-written that code into more languages than I care to think about.

Integer keys have always been overrated.
 
2017-09-30 11:09:18 AM  

Gary-L: FTA: In September 2007, Jean Bookout was driving on the highway with her best friend in a Toyota Camry when the accelerator seemed to get stuck. When she took her foot off the pedal, the car didn't slow down. She tried the brakes but they seemed to have lost their power. As she swerved toward an off-ramp going 50 miles per hour, she pulled the emergency brake. The car left a skid mark 150 feet long before running into an embankment by the side of the road. The passenger was killed. Bookout woke up in a hospital a month later.

First, people need to stop referring to it as "the emergency brake".  The proper nomenclature is "parking brake" or "handbrake".  Second, the best way out of that situation is to shift the vehicle into neutral whereupon the vehicle operator can steer the vehicle into a controlled stop.  Yes, this may involve colliding with a curb or rolling the vehicle in a ditch, but the alternative is obviously worse.  The engine may or may not redline depending on whether or not the rev-limiter activates.


You are assuming the software controlled shifter allows shifts into neutral at that RPM, speed, pedal position, steering angle, etc.
 
2017-09-30 11:12:58 AM  
There is a bunch of work on actually proving real programs are correct.

For example the https://en.wikipedia.org/wiki/Coq system uses the Curry-Howard_correspondence (that proofs are programs) and has been able to prove mathematical theorems as well a write a provably correct C compiler.

There is the whole issue of what sort of maths such systems should use. Just set theory (as in Lamport's TLA+) is considered too weak and primitive (a lot of trees with no forest). Current approaches use (higher) category theory to reason about types and behaviors, or even that extended to Homotopy_type_theory where the notion of equality can depend on context in a mathematically precise way.

I recently stumbled across the 5 year INRIA project CoqHoTT (Coq for Homotopy Type Theory). The lead researcher is based at École_des_mines_de_Nantes. Why would a school of mining be interested in leading edge maths? I can't untangle the bureaucracies now but essentially the government Department of Mining became responsible for nuclear power (which generates 40% of France's electricity) so mining schools are where nuclear engineering is taught. Thus miners are really concerned about the correctness of software controlling nuclear reactors.
 
2017-09-30 11:16:43 AM  

Gary-L: FTA: In September 2007, Jean Bookout was driving on the highway with her best friend in a Toyota Camry when the accelerator seemed to get stuck. When she took her foot off the pedal, the car didn't slow down. She tried the brakes but they seemed to have lost their power. As she swerved toward an off-ramp going 50 miles per hour, she pulled the emergency brake. The car left a skid mark 150 feet long before running into an embankment by the side of the road. The passenger was killed. Bookout woke up in a hospital a month later.

First, people need to stop referring to it as "the emergency brake".  The proper nomenclature is "parking brake" or "handbrake".  Second, the best way out of that situation is to shift the vehicle into neutral whereupon the vehicle operator can steer the vehicle into a controlled stop.  Yes, this may involve colliding with a curb or rolling the vehicle in a ditch, but the alternative is obviously worse.  The engine may or may not redline depending on whether or not the rev-limiter activates.


This lack of knowledge is the result of our laughably easy driving tests. It's nothing more than a rubber stamp that verifies that you know that a stop sign means stop. There's nothing in it about how to handle the car in an emergency situation - not even scenarios that are likely to happen to everyone at least once, such as a tire blowing out on the highway.

And it's not this way in other countries.
 
2017-09-30 11:22:16 AM  

Chevello: Gary-L: FTA: In September 2007, Jean Bookout was driving on the highway with her best friend in a Toyota Camry when the accelerator seemed to get stuck. When she took her foot off the pedal, the car didn't slow down. She tried the brakes but they seemed to have lost their power. As she swerved toward an off-ramp going 50 miles per hour, she pulled the emergency brake. The car left a skid mark 150 feet long before running into an embankment by the side of the road. The passenger was killed. Bookout woke up in a hospital a month later.

First, people need to stop referring to it as "the emergency brake".  The proper nomenclature is "parking brake" or "handbrake".  Second, the best way out of that situation is to shift the vehicle into neutral whereupon the vehicle operator can steer the vehicle into a controlled stop.  Yes, this may involve colliding with a curb or rolling the vehicle in a ditch, but the alternative is obviously worse.  The engine may or may not redline depending on whether or not the rev-limiter activates.

You are assuming the software controlled shifter allows shifts into neutral at that RPM, speed, pedal position, steering angle, etc.


I'm sorry Dave.  I'm afraid I can't do that!
 
2017-09-30 11:32:11 AM  

Chevello: Gary-L: FTA: In September 2007, Jean Bookout was driving on the highway with her best friend in a Toyota Camry when the accelerator seemed to get stuck. When she took her foot off the pedal, the car didn't slow down. She tried the brakes but they seemed to have lost their power. As she swerved toward an off-ramp going 50 miles per hour, she pulled the emergency brake. The car left a skid mark 150 feet long before running into an embankment by the side of the road. The passenger was killed. Bookout woke up in a hospital a month later.

First, people need to stop referring to it as "the emergency brake".  The proper nomenclature is "parking brake" or "handbrake".  Second, the best way out of that situation is to shift the vehicle into neutral whereupon the vehicle operator can steer the vehicle into a controlled stop.  Yes, this may involve colliding with a curb or rolling the vehicle in a ditch, but the alternative is obviously worse.  The engine may or may not redline depending on whether or not the rev-limiter activates.

You are assuming the software controlled shifter allows shifts into neutral at that RPM, speed, pedal position, steering angle, etc.


Her Camry was old enough that the shifter was still cable controlled, but you have an excellent point regarding newer vehicles.  I drive vehicles that are 10 years old, so I forget about everything being controlled by the CANBUS.  Her applying the parking brake (rear brakes on a FWD vehicle) will cause the vehicle to fishtail, since the locked rear wheels are being dragged and the vehicle is still being steered.  Given the age and features of her vehicle, my original suggestion is applicable.  It would be interesting to know if the vehicle's software would prevent the transmission from shifting to neutral or even a lower gear.  It is possible doing so could override the accelerator, thereby returning control to the operator.
 
2017-09-30 11:46:11 AM  
Really interesting and thought provoking (and long) article, although through the first half all I could think of was

img.fark.netimg.fark.net
 
2017-09-30 11:49:17 AM  

Gary-L: Chevello: Gary-L: FTA: In September 2007, Jean Bookout was driving on the highway with her best friend in a Toyota Camry when the accelerator seemed to get stuck. When she took her foot off the pedal, the car didn't slow down. She tried the brakes but they seemed to have lost their power. As she swerved toward an off-ramp going 50 miles per hour, she pulled the emergency brake. The car left a skid mark 150 feet long before running into an embankment by the side of the road. The passenger was killed. Bookout woke up in a hospital a month later.

First, people need to stop referring to it as "the emergency brake".  The proper nomenclature is "parking brake" or "handbrake".  Second, the best way out of that situation is to shift the vehicle into neutral whereupon the vehicle operator can steer the vehicle into a controlled stop.  Yes, this may involve colliding with a curb or rolling the vehicle in a ditch, but the alternative is obviously worse.  The engine may or may not redline depending on whether or not the rev-limiter activates.

You are assuming the software controlled shifter allows shifts into neutral at that RPM, speed, pedal position, steering angle, etc.

Her Camry was old enough that the shifter was still cable controlled, but you have an excellent point regarding newer vehicles.  I drive vehicles that are 10 years old, so I forget about everything being controlled by the CANBUS.  Her applying the parking brake (rear brakes on a FWD vehicle) will cause the vehicle to fishtail, since the locked rear wheels are being dragged and the vehicle is still being steered.  Given the age and features of her vehicle, my original suggestion is applicable.  It would be interesting to know if the vehicle's software would prevent the transmission from shifting to neutral or even a lower gear.  It is possible doing so could override the accelerator, thereby returning control to the operator.


I didn't check the year, so that may have been an option for her. But having no actual control over things in my cars bothers me. And then you have "good ideas" like changing the way the shifter even operates because it isn't forced to move a cable anymore, like the Durango that killed Chekov. The worst thing is that the only reason for it is manufacturability and cost cutting. Everything is a module assembled somewhere cheap and installed on the assembly line so that it can be done by the lowest cost employee.  You can't even control the heater anymore, it's all just a set of IO requests.
 
2017-09-30 12:01:29 PM  

Marcus Aurelius: Flab: Lots places have gas pumps that can't handle prices above $99.99.

Gas pumps can go straight to hell.  If you've ever tried programming one, you would agree.


One of the local gas stations has installed brand-spanking new pumps about a year ago. They are the most temperamental pumps I have ever seen. It's a Russian Roulette of whether the machines will be unable to read your card, or get stuck reading your card, or not print a receipt and act like it did, or just plain refuse to work.
 
2017-09-30 12:03:35 PM  

Chevello: Gary-L: Chevello: Gary-L: FTA: In September 2007, Jean Bookout was driving on the highway with her best friend in a Toyota Camry when the accelerator seemed to get stuck. When she took her foot off the pedal, the car didn't slow down. She tried the brakes but they seemed to have lost their power. As she swerved toward an off-ramp going 50 miles per hour, she pulled the emergency brake. The car left a skid mark 150 feet long before running into an embankment by the side of the road. The passenger was killed. Bookout woke up in a hospital a month later.

First, people need to stop referring to it as "the emergency brake".  The proper nomenclature is "parking brake" or "handbrake".  Second, the best way out of that situation is to shift the vehicle into neutral whereupon the vehicle operator can steer the vehicle into a controlled stop.  Yes, this may involve colliding with a curb or rolling the vehicle in a ditch, but the alternative is obviously worse.  The engine may or may not redline depending on whether or not the rev-limiter activates.

You are assuming the software controlled shifter allows shifts into neutral at that RPM, speed, pedal position, steering angle, etc.

Her Camry was old enough that the shifter was still cable controlled, but you have an excellent point regarding newer vehicles.  I drive vehicles that are 10 years old, so I forget about everything being controlled by the CANBUS.  Her applying the parking brake (rear brakes on a FWD vehicle) will cause the vehicle to fishtail, since the locked rear wheels are being dragged and the vehicle is still being steered.  Given the age and features of her vehicle, my original suggestion is applicable.  It would be interesting to know if the vehicle's software would prevent the transmission from shifting to neutral or even a lower gear.  It is possible doing so could override the accelerator, thereby returning control to the operator.

I didn't check the year, so that may have been an option for her. But hav ...


The incident happened in 2007, so her Camry was not equipped with the equipment you mention.  Like you, I agree about the lack of control; however, it's not the fault of the system, but rather the fault of the design.  Why the hell should the entertainment system be able to "talk" to the ECM?  Hackers Remotely Kill a Jeep on the Highway -- with Men Inside  That demonstration pissed off FCA and also demonstrated that the systems are not intelligently designed.  When Chevrolet presented the concept of the Chevy Bolt to a focus group, the people (thankfully) responded that mechanical controls for the environmentals, and other functions, should be mandatory.  Chevy listened.  The reliance on the touch screen for all functions in Tesla vehicle really irks me.  I was taught to drive by feel, meaning I should be able to find my way around the dashboard by memory and sensitivity, because one's eyes should NEVER be taken from the road.  I've witnessed too many drivers veer all over because they are focused on the dash (even before touch screens).  Steering controls, to me, are fantastic.  Being able to control many functions, by touch, without removing a hand from the steering wheel is brilliant.

Sometimes, the glacial pace regarding regulations over what can and can't be changed on vehicles, are ridiculous.  For example, manufacturers have been fighting to get the regulations revamped regarding exterior lighting improvements changed to the better, for consumers, because the current regulations are based on 50+ year old technology.  In the case of all this computer control, and lack of operator control, the regulations (IMO) don't do enough.

There is a subset of folks who think autonomous vehicles will be fantastic.  Just remember, Google wants to do away with hand and foot controls in autonomous vehicles.
 
2017-09-30 12:29:32 PM  

Gary-L: The incident happened in 2007, so her Camry was not equipped with the equipment you mention. Like you, I agree about the lack of control; however, it's not the fault of the system, but rather the fault of the design. Why the hell should the entertainment system be able to "talk" to the ECM? Hackers Remotely Kill a Jeep on the Highway -- with Men Inside That demonstration pissed off FCA and also demonstrated that the systems are not intelligently designed. When Chevrolet presented the concept of the Chevy Bolt to a focus group, the people (thankfully) responded that mechanical controls for the environmentals, and other functions, should be mandatory. Chevy listened. The reliance on the touch screen for all functions in Tesla vehicle really irks me. I was taught to drive by feel, meaning I should be able to find my way around the dashboard by memory and sensitivity, because one's eyes should NEVER be taken from the road. I've witnessed too many drivers veer all over because they are focused on the dash (even before touch screens). Steering controls, to me, are fantastic. Being able to control many functions, by touch, without removing a hand from the steering wheel is brilliant.

Sometimes, the glacial pace regarding regulations over what can and can't be changed on vehicles, are ridiculous. For example, manufacturers have been fighting to get the regulations revamped regarding exterior lighting improvements changed to the better, for consumers, because the current regulations are based on 50+ year old technology. In the case of all this computer control, and lack of operator control, the regulations (IMO) don't do enough.

There is a subset of folks who think autonomous vehicles will be fantastic. Just remember, Google wants to do away with hand and foot controls in autonomous vehicles.


You're right, some things should be updated when tech advances. Lighting these days is getting way better. I even added a CHMSL from a 2000's Volvo to my '72 Chevy, and I'll likely add LED headlamps once they are cheap enough and stop looking like part of a transformer. Other things, like electric steering and brake by wire, I'm not so excited about, nor, living in a place that gets cold in winter, environmental systems that have to ask a computer to turn on the heat. The drive by wire throttle was definitely easier to install in the car than adapting a cable throttle, but it has confused itself and shut down before.

My son has an 04 Avalon that turns off the traction control, stability assist and the ABS when it sets a check engine light due to an emissions problem. The charcoal canister has a leak? Better not allow the stability assist to work. Seems kind of stupid that the computers can't talk to each other enough to make a better call on that. Hope you don't lose a vacuum hose on an icy road. Good luck.
 
2017-09-30 12:36:42 PM  
tl;dr - We need coders working directly in the fields now. The days of hiring out coders passed DECADES ago. Why the fark are we still working this archaic practice?

The programming thing is a real problem that's happening now and affecting a *lot* of people, whether they know it or not.

I can cite a direct source. Every year in my area the business rules for how to get a project funded changes, sometimes dramatically, and always in ways that people don't end up suspecting, because of the native spaghetti code that is the business rules. In my first year, I immediately found this problem and set about building a coded spreadsheet to help navigate the mess and whaddayaknow I was right; I found unintended consequences that required global announcements to the bases with errata on the rules.

This year they hired a contractor to effectively do what I did. But they were contextually clueless as to the why of the code and they built something that didn't save, used VB instead of native Excel language, locked everything and didn't comment a damn thing. They handed over the code and it was passed along as some sort of miracle device.

No one checked the work. I did and found errors. Lots of errors. Because I'm working both sides of the situation (the business rules and the programmatic logic). I sent my response and kept working with my stupid spreadsheet (now hundreds of lines long) because I already had significant hours of debugging in the mix, including a full rewrite to optimize some bad language and clean up spaghetti logic.

Command has my sheet now. They want to use it as a template to standardize the function AF wide. Woo, I guess, but that shouldn't be the bloody exception. The work I do should be the rule.
 
2017-09-30 12:41:44 PM  

Chevello: You're right, some things should be updated when tech advances. Lighting these days is getting way better. I even added a CHMSL from a 2000's Volvo to my '72 Chevy, and I'll likely add LED headlamps once they are cheap enough and stop looking like part of a transformer. Other things, like electric steering and brake by wire, I'm not so excited about, nor, living in a place that gets cold in winter, environmental systems that have to ask a computer to turn on the heat. The drive by wire throttle was definitely easier to install in the car than adapting a cable throttle, but it has confused itself and shut down before.

My son has an 04 Avalon that turns off the traction control, stability assist and the ABS when it sets a check engine light due to an emissions problem. The charcoal canister has a leak? Better not allow the stability assist to work. Seems kind of stupid that the computers can't talk to each other enough to make a better call on that. Hope you don't lose a vacuum hose on an icy road. Good luck


Don't get too froggy on LEDs.  I've read a lot of comparisons and tests that show LEDs aren't the best for headlamps, but they are superior for indicators and brake lamps.  Here is a great article on headlamp tech: Battle of the Headlights: Halogen vs. Xenon vs. LED vs. Laser vs. Conversion Kits

As far as systems deactivating, that boils down to the manufacturer and why the system was designed that way.  Cars are designed like the manufacturing process of the Boeing 787, which is probably why manufacturers have opted for everything communicating with the CANBUS.

Not only is the loss of vacuum pressure a diagnostic problem, but so is proper grounding.  People in corrosive environments encounter a lot of problems with their vehicles due to corrosion at the grounding point or when a ground wire breaks.  Vehicles built since the 1990s are sensitive to the ±5 volts needed for operation.
 
2017-09-30 12:53:21 PM  

thornhill: Gary-L: FTA: In September 2007, Jean Bookout was driving on the highway with her best friend in a Toyota Camry when the accelerator seemed to get stuck. When she took her foot off the pedal, the car didn't slow down. She tried the brakes but they seemed to have lost their power. As she swerved toward an off-ramp going 50 miles per hour, she pulled the emergency brake. The car left a skid mark 150 feet long before running into an embankment by the side of the road. The passenger was killed. Bookout woke up in a hospital a month later.

First, people need to stop referring to it as "the emergency brake".  The proper nomenclature is "parking brake" or "handbrake".  Second, the best way out of that situation is to shift the vehicle into neutral whereupon the vehicle operator can steer the vehicle into a controlled stop.  Yes, this may involve colliding with a curb or rolling the vehicle in a ditch, but the alternative is obviously worse.  The engine may or may not redline depending on whether or not the rev-limiter activates.

This lack of knowledge is the result of our laughably easy driving tests. It's nothing more than a rubber stamp that verifies that you know that a stop sign means stop. There's nothing in it about how to handle the car in an emergency situation - not even scenarios that are likely to happen to everyone at least once, such as a tire blowing out on the highway.

And it's not this way in other countries.


I had two driving tests

The first one was to get my government issued paperwork.

The second was my old man tossing me the keys to the beater Volvo. We went to an empty parking lot with fresh snow, and said "let's see how you handle getting into and out of a slide"
 
2017-09-30 01:05:06 PM  
Operated by a systems provider named Intrado, the server kept a running counter of how many calls it had routed to 911 dispatchers around the country. Intrado programmers had set a threshold for how high the counter could go. They picked a number in the millions.

Why would you do that? Why would anyone do that? Use a alphanumeric varchar in the database and after you had the first 36^254 combinations set up a warning that cleaning out the data from 200 years ago might be advisable.
 
2017-09-30 01:11:46 PM  
TLDR:
img.fark.net

Victor went one further. He demoed two programs he'd built-each of which took a process that used to involve writing lots of custom code and reduced it to playing around in a WYSIWYG interface. Victor suggested that the same trick could be pulled for nearly every problem where code was being written today. "I'm not sure that programming has to exist at all," he told me. "Or at least software developers." In his mind, a software developer's proper role was to create tools that removed the need for software developers.

IMO, this is pretty stupid. Most of his examples were not even really about programming. Mario Bro's level design, building web pages, and image editing are all tasks that obviously lend themselves to being done with tools that have WYSIWYG interfaces. But not every problem is as repetitive and well defined as those, or as easy to represent visually. And to borrow the hammer analogy, when your only tool is a Mario Level Editor, every problem looks like... a Mario Level? Who develops and maintains these tools? And what happens when the user wants to do something that was never anticipated by the tool developer. A real programmer has to be involved at some point in the process. (Unless it's WYSIWYG and/or turtles all the way down?)

Designing a tool to do a job implies that the job needs to be done over and over, and requires a much greater understanding of all possible areas of the problem. It's usually going to be more cost effective to just solve the problem at hand, by hand, rather than writing a tool to solve all possible future variations of the problem.
 
2017-09-30 01:13:12 PM  

Ghost Roach: thornhill: Gary-L: FTA: In September 2007, Jean Bookout was driving on the highway with her best friend in a Toyota Camry when the accelerator seemed to get stuck. When she took her foot off the pedal, the car didn't slow down. She tried the brakes but they seemed to have lost their power. As she swerved toward an off-ramp going 50 miles per hour, she pulled the emergency brake. The car left a skid mark 150 feet long before running into an embankment by the side of the road. The passenger was killed. Bookout woke up in a hospital a month later.

First, people need to stop referring to it as "the emergency brake".  The proper nomenclature is "parking brake" or "handbrake".  Second, the best way out of that situation is to shift the vehicle into neutral whereupon the vehicle operator can steer the vehicle into a controlled stop.  Yes, this may involve colliding with a curb or rolling the vehicle in a ditch, but the alternative is obviously worse.  The engine may or may not redline depending on whether or not the rev-limiter activates.

This lack of knowledge is the result of our laughably easy driving tests. It's nothing more than a rubber stamp that verifies that you know that a stop sign means stop. There's nothing in it about how to handle the car in an emergency situation - not even scenarios that are likely to happen to everyone at least once, such as a tire blowing out on the highway.

And it's not this way in other countries.

I had two driving tests

The first one was to get my government issued paperwork.

The second was my old man tossing me the keys to the beater Volvo. We went to an empty parking lot with fresh snow, and said "let's see how you handle getting into and out of a slide"


My tests in Pennsylvania circa 2000 involved a multiple choice test on traffic signs (that was to get the learners permit) and then later a driving test that was literally a trip around the block that included 1 stop sign and 1 traffic light, showing that I knew how to turn on the vehicle headlights, and parallel parking in a space two-car lengths long. IIRC my father had to sign something attesting to the fact that I had driven some number of hours with a licensed driver.

I'm not afraid to admit that when I got my DL I was woefully underprepared for driving in urban environments (because most of my practicing was done off major roadways). I eventually took an evasive driving class conducted by ex-State police for fun and it was extremely helpful.
 
2017-09-30 01:22:59 PM  

X-Geek: TLDR:
[img.fark.net image 500x323]

Victor went one further. He demoed two programs he'd built-each of which took a process that used to involve writing lots of custom code and reduced it to playing around in a WYSIWYG interface. Victor suggested that the same trick could be pulled for nearly every problem where code was being written today. "I'm not sure that programming has to exist at all," he told me. "Or at least software developers." In his mind, a software developer's proper role was to create tools that removed the need for software developers.

IMO, this is pretty stupid. Most of his examples were not even really about programming. Mario Bro's level design, building web pages, and image editing are all tasks that obviously lend themselves to being done with tools that have WYSIWYG interfaces. But not every problem is as repetitive and well defined as those, or as easy to represent visually. And to borrow the hammer analogy, when your only tool is a Mario Level Editor, every problem looks like... a Mario Level? Who develops and maintains these tools? And what happens when the user wants to do something that was never anticipated by the tool developer. A real programmer has to be involved at some point in the process. (Unless it's WYSIWYG and/or turtles all the way down?)

Designing a tool to do a job implies that the job needs to be done over and over, and requires a much greater understanding of all possible areas of the problem. It's usually going to be more cost effective to just solve the problem at hand, by hand, rather than writing a tool to solve all possible future variations of the problem.


I didn't RTFA that far, but it brings to mind the difference between using a GUI versus Powershell.  The command line will prevail every time.
 
2017-09-30 01:29:26 PM  

X-Geek: TLDR:
[img.fark.net image 500x323]

Victor went one further. He demoed two programs he'd built-each of which took a process that used to involve writing lots of custom code and reduced it to playing around in a WYSIWYG interface. Victor suggested that the same trick could be pulled for nearly every problem where code was being written today. "I'm not sure that programming has to exist at all," he told me. "Or at least software developers." In his mind, a software developer's proper role was to create tools that removed the need for software developers.

IMO, this is pretty stupid. Most of his examples were not even really about programming. Mario Bro's level design, building web pages, and image editing are all tasks that obviously lend themselves to being done with tools that have WYSIWYG interfaces. But not every problem is as repetitive and well defined as those, or as easy to represent visually. And to borrow the hammer analogy, when your only tool is a Mario Level Editor, every problem looks like... a Mario Level? Who develops and maintains these tools? And what happens when the user wants to do something that was never anticipated by the tool developer. A real programmer has to be involved at some point in the process. (Unless it's WYSIWYG and/or turtles all the way down?)

Designing a tool to do a job implies that the job needs to be done over and over, and requires a much greater understanding of all possible areas of the problem. It's usually going to be more cost effective to just solve the problem at hand, by hand, rather than writing a tool to solve all possible future variations of the problem.


Yes it's WYSIWYProgram all the way down.  That was kind of the whole point. Abstracting it to a layer that generates underlying code in the same way that typical compiling languages generate machine code.  Hopefully and theoretically possible perfectly.  Procedural specs/pseudocode that becomes real software.
 
2017-09-30 01:40:50 PM  
It's because software doesn't know when bad programming tells it to fark something up that we need true AI. An intelligent software entity that would know when following a given command would result in an unwanted out come.
But we should start small. Maybe have an AI who's function is entirely to open and close doors. We could name it Durandal. After we establish that, we could move up to giving AIs control of deadly neurotoxin.
 
2017-09-30 01:58:39 PM  

Far Cough: Yes it's WYSIWYProgram all the way down.  That was kind of the whole point. Abstracting it to a layer that generates underlying code in the same way that typical compiling languages generate machine code.  Hopefully and theoretically possible perfectly.  Procedural specs/pseudocode that becomes real software.


The "all the way down" part implies that the tool you're using to solve your problem was in turn developed by using a tool that helped someone solve the problem of developing a tool that would help you solve your problem... and so on, and so on, back to... what?  The great Master Control Program that came into existence through a software singularity and/or God?
img.fark.net

I just thought his point was kind of... well... pointless. Having tools that help with development is great, but these tools don't develop themselves, so I don't see how programmers are going away any time soon.
 
2017-09-30 01:59:33 PM  

ajgeek: Every year in my area the business rules for how to get a project funded changes, sometimes dramatically, and always in ways that people don't end up suspecting, because of the native spaghetti code that is the business rules. In my first year, I immediately found this problem and set about building a coded spreadsheet to help navigate the mess and whaddayaknow I was right; I found unintended consequences that required global announcements to the bases with errata on the rules.

This year they hired a contractor to effectively do what I did. ... clueless ... VB ...


It sounds like sort of what you need is some sort of  formal Business_process_model rather than encoding the business rules in a spreadsheet. That is probably what the management thought they were getting from the contractor rather than a hack recoding of your spreadsheet into VB. Then again picking some formal system and using it for just one smallish project makes little sense.

The USAF could really use a common business process modeling system. Getting there depends on navigating all the levels of bureacracy and the hot air spewed by contractors. There is a webpage USAF Acquisition Process Model Home that is currently dead (cached version) and I can't tell it that project achieved anything. There are probably way too many players in the game to reach a consensus in the next 10 years and if one is reached it could so complex (as to satisfy every interest) that it is unusable.

Meanwhile contractors can still make bucks selling I don't know what to deal with some of the problems. E.g.
https://www.govconwire.com/2017/07/air-force-taps-booz-allen-for-pote​n​tial-140m-business-area-process-support-idiq/
Air Force Taps Booz Allen for Potential $140M Business Area Process Support IDIQ
 
2017-09-30 02:03:07 PM  

Far Cough: Really interesting and thought provoking (and long) article, although through the first half all I could think of was

[img.fark.net image 425x613][img.fark.net image 425x318]


8 of my 12 pinballs are Electromechanical
Limelight
Twin Win
Wizard
Captain Fantastic
Klondike
Space Time
Student Prince
Full House

Solid State pins

Blackout
Flash
Taxi
Lost World
 
2017-09-30 02:03:53 PM  

MythDragon: It's because software doesn't know when bad programming tells it to fark something up that we need true AI. An intelligent software entity that would know when following a given command would result in an unwanted out come.
But we should start small. Maybe have an AI who's function is entirely to open and close doors. We could name it Durandal. After we establish that, we could move up to giving AIs control of deadly neurotoxin.


A named AI? Why not have a nameless AI pass butter? That would be a far better testcase for rampancy.
 
2017-09-30 02:13:48 PM  

MythDragon: Maybe have an AI who's function is entirely to open and close doors.


I'm sure nothing could possiblai go wrong with that.
img.fark.net
 
2017-09-30 02:17:32 PM  

X-Geek: Far Cough: Yes it's WYSIWYProgram all the way down.  That was kind of the whole point. Abstracting it to a layer that generates underlying code in the same way that typical compiling languages generate machine code.  Hopefully and theoretically possible perfectly.  Procedural specs/pseudocode that becomes real software.

The "all the way down" part implies that the tool you're using to solve your problem was in turn developed by using a tool that helped someone solve the problem of developing a tool that would help you solve your problem... and so on, and so on, back to... what?  The great Master Control Program that came into existence through a software singularity and/or God?
[img.fark.net image 200x90]

I just thought his point was kind of... well... pointless. Having tools that help with development is great, but these tools don't develop themselves, so I don't see how programmers are going away any time soon.


Well first the concept is not even unusual, since of course a classic test of a language compiler is whether it can compile its own code properly.  And second, yeah, at least one of the proponents in the article is saying out loud that the goal should be removing the need for traditional programmers.

I guess the ultimate idea is to express the business or other requirements in human language and let the MCP sort it all out.  Ironically though the current TLA+ solution is apparently heavily mathematical or arcane.

/only know what I read in TFA+
 
2017-09-30 03:12:07 PM  

HairBolus: ajgeek: Every year in my area the business rules for how to get a project funded changes, sometimes dramatically, and always in ways that people don't end up suspecting, because of the native spaghetti code that is the business rules. In my first year, I immediately found this problem and set about building a coded spreadsheet to help navigate the mess and whaddayaknow I was right; I found unintended consequences that required global announcements to the bases with errata on the rules.

This year they hired a contractor to effectively do what I did. ... clueless ... VB ...

It sounds like sort of what you need is some sort of  formal Business_process_model rather than encoding the business rules in a spreadsheet. That is probably what the management thought they were getting from the contractor rather than a hack recoding of your spreadsheet into VB. Then again picking some formal system and using it for just one smallish project makes little sense.

The USAF could really use a common business process modeling system. Getting there depends on navigating all the levels of bureacracy and the hot air spewed by contractors. There is a webpage USAF Acquisition Process Model Home that is currently dead (cached version) and I can't tell it that project achieved anything. There are probably way too many players in the game to reach a consensus in the next 10 years and if one is reached it could so complex (as to satisfy every interest) that it is unusable.

Meanwhile contractors can still make bucks selling I don't know what to deal with some of the problems. E.g.
https://www.govconwire.com/2017/07/air-force-taps-booz-allen-for-poten​tial-140m-business-area-process-support-idiq/
Air Force Taps Booz Allen for Potential $140M Business Area Process Support IDIQ


It gets worse. In the last several years, the AF stood up the Installation & Mission Support Center (IMSC), so now base engineers formally have two bosses, who each have different agendas. This means one asks for things one way, the other asks for things another way, and in the end I just pick which way more closely meets both sets of criteria and do the work once and have actively told higher ups to deal with it; they literally pay me to think, so stop thinking I'm another uniform.

Now the Air Force Civil Engineering Center (AFCEC) is desperately scrambling to justify its existence. They've turned into the DoD equivalent of Realtors.

/I read and obey International Code Council, Unified Facility Criteria, and Air Force Instructionals, in that order, top to bottom.
//Your petty bickering and power struggles don't interest me.
 
2017-09-30 03:21:19 PM  
Well, I can't argue that the software apocalypse isn't coming.  To write software you have to account for every possible input combination and make sure the algorithm does the right thing.  And we seem to have completely abandoned the KISS principle so that's an impossible task.  But I do have to take exception to:

Bret Victor does not like to write code. "It sounds weird," he says. "When I want to make a thing, especially when I want to create something in software, there's this initial layer of disgust that I have to push through, where I'm not manipulating the thing that I want to make, I'm writing a bunch of text into a text editor."

An algorithm is analogous to a recipe for accomplishing a task.  So all this time we've been cooking wrong because our recipies are written in cookbooks rather than what?  Made out of food?  I think ol' Bret is just in the wrong line of work.
 
Displayed 50 of 65 comments


Oldest | « | 1 | 2 | » | Newest | Show all


View Voting Results: Smartest and Funniest

This thread is closed to new comments.

Continue Farking

On Twitter





Top Commented
Javascript is required to view headlines in widget.
  1. Links are submitted by members of the Fark community.

  2. When community members submit a link, they also write a custom headline for the story.

  3. Other Farkers comment on the links. This is the number of comments. Click here to read them.

  4. Click here to submit a link.

Report