“My Car does not start when I buy Vanilla Ice Cream” (2015)
digitalrepublik.comOn a related note: The case of the 500 mile email [0]
This immediately came to mind as well and is one of my favorite stories.
Indeed! From the article:
> In 2010, the Pontiac Division of General Motors received...
> This was a cool act by General Motors especially in this world of Internet when news can go viral in matter of seconds
From Snopes:
> This legend surfaced in print in 1978, but an anecdotal sighting places it even earlier than that, in 1971.
Yes. It's a neat story, but it has that smell of an urban legend. What engineer would come back for four nights in a row rather than just figuring out why the car didn't start the first time?
Why do you think he came back four times in a row? Not everyone has a complete and accurate physical model of their car in their head.
The story, as written, says that the engineer want with the father on four different nights. I would expect an engineer to at least try to diagnose the problem rather than spending all the effort of coming back on four different nights outside of work ours. Basically the story feels made up.
I agree only because just reading the problem description I guessed that it was a difference in distance or sitting time and I don't even know anything about car engines. I assume any automotive engineer would have been ahead of me on that and already looking for differences that could matter on the first trip.
An engineer who is using logic, reason, and observation to find the root of the problem instead of just patch it once?
Or an Engineer who likes ice cream!
Do you figure out every bug the first time you see it?
Or do you have to look into it for a while
The story reads like an email my grandfather would send me with 6 pages of fwds and signatures about emails being checked by Norton Antivirus.
If the car didn’t start then how did he come back the next day to get more ice cream? Surely he waited a bit and the car would start. And the timing issue would be obvious from here?
I’m also worried about his health. Should he really be having ice cream everyday?
The engineer also wouldn't wait until the next day to test again.
One of our bug reports was "if we leave the machine for more than two weeks it doesn't deploy correctly." Sometimes you can't test again until the hardware is ready.
> I’m also worried about his health. Should he really be having ice cream everyday?
Why not? As long as you are not overweight, I don't see much of a problem.
Reminds me of my wife's macbook air wifi that was always spotty when she was using it, and always got back to normal when I used it. She thought I had some kind of magic engineer's touch.
After much thinking, we realized that when she was using the laptop in her bed, she would put it on her lap with her legs bent, so that the computer was sitting at a 45 degree angle wrt to the floor, while I would put the computer flat on my chest. This angle made the wifi antenna much less performant.
Humans are basically bags of mostly water, and water is not god to radio signals. Have seen the exact same problem in rooms that are far away from the router. Massive difference if any of my body parts is in the way of the signal
Wifi is weird sometimes. I once had to help my nephew with his network. He had 2 adjacent rooms, but with half a meter concrete inbetween. Needless to say the signal would not penetrate from one room to the other, no matter how close we moved the router and laptop to each other in their respective rooms. But moving the router and laptop back would make it work all of a sudden. Turned out, moving them back would position them in front of the windows, and the signal would then go out of the window and bounce of the neighbouring building to the other room's window.
Did she change her posture or did you get a wifi dongle?
Why do sites use this crappy scrolling override. I's infuriating and makes navigating the article a pain.
I just thought my ctrl-key was broken because I couldn't change the size with ctrl-mousewheel. I whish webbrowsers wouldn't allow this shit. There is no legit reason for a website to fiddle with the basic controls.
I was annoyed when the article text wouldn't display at all without enabling JavaScript -- but I could read it just fine with Firefox's Reader View.
My first car (bought used) had a weird bug: When it had been parked overnight on a steep slope facing downwards, the driver's electric window sometimes wouldn't go back up. Rain would increase the chance of the bug occurring.
All other windows were fine and the problem would go away after a few hours of driving. The bug never occurred if the car was parked on a flat surface.
The mechanics were clueless. I've found that there's a so-called "convenience comfort module" located under a front seat. This electronic module was responsible for controlling the electric windows (amongst other things) and was insufficiently sealed. Water could enter and disrupt certain functions. My dealer wouldn't even believe me that this module exists. :)
Turns out other people had the water problem, too: https://www.youtube.com/watch?v=C9z7iEbg3qY
> parked overnight on a steep slope facing downwards
I had a car which couldn't be parked that way either! Turned out, the reverse gear on a 1985 Honda Civic with a 4-speed manual was too low - the tiny engine simply didn't have the power to back up uphill. Unless I could guarantee that nobody would parallel park close to my front bumper, the car would be stuck in the parking spot!
You parked your car and left the windows open?
No; after letting them down the following day, the one on the driver's side just wouldn't go back up again or would even go down by itself once the car was started.
This story comes up once every few months if we get a really, really weird bug request from support. Issues like „the customer says, the app automatically buys something while he is sleeping. Please investigate where our app has automatic buying“ and we just sit there and stare at this in awe
The point of this story is not the ice cream, it's to take user reports of bugs seriously even if they make no sense or you can't reproduce them.
Vapor lock was a thing in the time of carburetors.
This story is from the 70s, when "internet virality" was not yet on people's minds.
Interesting. Fuel-injected aircraft engines get vapor lock—shutting down to refuel and then restarting can be a serious pain—but I don’t think I’ve ever heard of a carbureted one doing it.
In Jefferson County, Colorado, the stretch of US 40 up into the foothills from the plains used have a lot of cars stalled by vapor lock along the shoulders. But I don't think I've seen a car stalled by vapor lock in forty years.
Neat. Sounds like I’ll be jumping down the rabbit hole to figure out why carbed cars vapor-lock but airplanes are the reverse. :)
It may be hard to recall but before the internet people shared stories, and the really good ones would spread quickly across the entire population. There were people back then too who would try to come up with things that would spread, maybe not with the same intent (or profits) as now, but still the overall effect was the same
There is a GitHub repo which lists loads of interesting bugs (vanilla ice cream car, 500 mile email, etc.) with better quality writeups than this. I think it's been posted to HN before. Anyone?
I once had the strange problem that shaking my Apple Magic Mouse helped to reestablish the lost blue tooth connection to my Mac. This happened quite often. It took me quite a while to figure out, why shaking the mouse should help with the bluetooth connection:
The reasons is simply that the batteries in the Magic Mouse are often loose and so shaking seems to help moving the batteries around: https://apple.stackexchange.com/questions/15709/magic-mouse-...
Hmm, got 500 Error code, Internal Server Error. Here is a backup:
https://web.archive.org/web/20191213101529/https://www.digit...
I had a similar problem on my previous car. It would start again without issue after stopping if the pause was shorter than 15min or longer than half an hour but there was a spot in between where the car had cooled down too much and not enough at the same time to start. It's nice to see it was not just a fancy from an old car and to know the reason.
I still have this problem on my 2004 Opel Astra. When it sits for about 15-20mins I have problem starting, but I just have to start it for about 5 seconds instead of instantly. Longer or shorter pauses are ok.
I had this same issue as a teen in the late 90s with an old 1984 Pontiac Phoenix. If it sat longer than 2 hours it would start just fine or if I started the car within 15 minutes of last using it. In that ~2hr gap if I needed to use the car I would have to spray some carb cleaner into the engine to get it started.
Maybe not important, but I would have put (2010) in the title.
The article was posted on May 10, 2015 but the story is from 2010.
more like ~1978
That's why it took the engineer so long to figure it out. He wasn't expecting a "solved" problem like heat soak.
Not the point of the story, but I wonder if engineers would reach the solution faster than non-engineers due to their practice debugging. As soon as I read the guy was driving to the store to pick it up I knew it had to be timing due to store layout.
If the story is to be believed then a smart engineer took several days to figure it out while it only took you a minute. I suspect there are other factors than simply being an 'engineer' that will help you arrive at the solution. I personally would never have guessed that store layout would be a factor. However, one question I had immediately was "how do you get home if your car doesn't start?" and if the reply was "I wait and eventually the car starts" then I would probably have closed in on the answer after that.
The situation described is basically a textbook perfect example of heat soak.
If you tell someone with experience repairing vehicles that your car starts fine when cold and find when you shut it off for a short while (e.g. buying gas they are going to nearly instantly hone in on "something is hot that shouldn't be".
> I knew it had to be timing due to store layout.
Even the "store layout" part is nonsense. Really, only vanilla is quick to pick up, and even for chocolate and strawberry (or whatever the next most popular flavors are) you have to take the slow path? Nobody would organize their store like that.
(Also, are dedicated "ice cream stores" a thing in the US? As in, not ice cream parlors that also sell stuff to go, but something you would really call a "store", that is dedicated to ice cream specifically?)
Well remember that this is second hand and paraphrased, so "ice cream store" could just mean "the place where my family goes to pick up ice cream"
And as for the layout - I have definitely seen freezers with a pile of easy-to-access vanilla ice cream in one section then a "miscellaneous" section where chocolate/pecan/whatever are either mixed in a pile or grouped (both require a couple of seconds to pause and search).
The story does seem too perfect, but this part does definitely check out
I knew it immediately too, but only because I'd heard this story or something very like it before. For the same reason, I might quickly figure out a problem related to janitors and vacuums, or light sensors on the inside of cabinet doors. That's probably the only correlation with being an engineer - not any systematic skill, but exposure to a certain kind of folklore.
My guess is yes. Especially if you're doing debugging, your whole day is hypothesis testing. You have a model of how a program works and from there you come up with potential explanations, you try to reproduce the problem under various conditions, you collect data, and so on.
I'm going to guess no.
Had the guy just asked anyone who's used to driving old beater cars from the late carb/early EFI they probably would have guessed vapor lock or heat soaked ignition components right off the bat (probably with a little skepticism because new cars very rarely do that) and then tried to confirm/rule it out.
Vapor lock is a basically solved problem (and has been for 30+yr, the OEMs test for those kinds of things before they scale up production) the engineer will likely weight it lower as a possibility than the more oddly specific things. It's kind of like how a 2nd line IT person may miss rebooting a device as a potential solution because none of the problems they chase down have easy solutions like that.
And also equipped with a firm knowledge that flavor of ice cream can't be the problem. Thus naturally looking at all other variables
One for Car Talk!
Actually when I read the title, my first gut reaction was "Vapor Lock".
Correlation vs causation, eh?
There is causation, just not directly from ice cream to vapor lock. It's a diagram with three entities, ice cream -> time to cool -> vapor lock
That's the nature of all situations in which "correlation is not causation".
It actually is causation. There just happens to be a causal step in between.