the PIP is a pip

February 5, 2021 2:50 PM

Subaru Blames A Single Factory Worker For A Recall (Jalopnik, Feb. 3, 2021)

The Subaru of America limited recall, affecting 383 Outbacks and Imprezas (model year 2021), concerns a continuously variable transmission (CVT) select lever cable nut which may have been improperly torqued during vehicle assembly.

Potentially affected vehicles were identified using vehicle production records. A single associate assigned to a specific production line working between the production dates specified (12/14/2020 – 12/21/2020) was found to be using an improper torque wrench technique. All vehicles potentially affected by this associate are included in the identified population. [...] If the gear select does not function properly, there is an increased risk of crash. (Part 573 Safety Recall Report, Subaru of America, submitted to the National Highway Traffic Safety Administration Jan. 26, 2021)

Check for recalls by VIN at NHTSA

posted by Iris Gambol (40 comments total)

[where PIP = performance improvement plan]
posted by Iris Gambol at 2:51 PM on February 5, 2021 [1 favorite]

For fuck's sake, Steve.
posted by allegedly at 2:57 PM on February 5, 2021 [21 favorites]

Blaming the operator is weak. This is automotive manufacturing. Get your operators trained on basic things like torque tool usage before they assemble anything. Use smart tools that record successful and unsuccessful torques, and lock out the product from moving until all critical torques are successfully complete. Perform torque audits on an appropriate sample so you detect these things before they ship. This operator was able to assemble incorrectly for an entire week.
posted by TrialByMedia at 2:58 PM on February 5, 2021 [82 favorites]

Yeah this seems like a textbook 5 why's problem. Why did the nut come loose? It wasn't torqued correctly. Why wasn't it torqued correctly? Because the operator used the wrong technique. Why didn't they use the right technique? ____

Hint: "Because they're dumb" is probably not the root cause.
posted by muddgirl at 3:01 PM on February 5, 2021 [73 favorites]

Right, but the factory worker was Mr. Bean, soooo....
posted by It's Raining Florence Henderson at 3:06 PM on February 5, 2021 [5 favorites]

Throw me under the fucking bus, that is where I'll be riding..
^ a useful song a former coworker of mine used to sing at moments like this
posted by bleep at 3:09 PM on February 5, 2021 [10 favorites]

Also a functional union system wouldn't have allowed any of this to happen in the first place.
posted by bleep at 3:10 PM on February 5, 2021 [12 favorites]

first comment - "I bet they gave him a stern torquing to."
posted by pyramid termite at 3:13 PM on February 5, 2021 [37 favorites]

well that took a hard turn fast
posted by It's Raining Florence Henderson at 3:15 PM on February 5, 2021 [16 favorites]

(I hope nobody was seriously hurt by this.)
posted by It's Raining Florence Henderson at 3:17 PM on February 5, 2021 [4 favorites]

> this seems like a textbook 5 why's problem. Why did the nut come loose? It wasn't torqued correctly. Why wasn't it torqued correctly? Because the operator used the wrong technique. Why didn't they use the right technique?

I agree. A question to ask is: is it worth changing the process to reduce the chance of a similar issue recurring in future? (maybe it is, maybe it isn't, the optimal process will balance cost of QA & QA effectiveness and will likely have some nonzero rate of shipping manufacturing defects) If it's worth changing, how will the process be improved to reduce the chances of this happening again? Better training? More frequent quality inspections?

I am not too surprised that there isn't a discussion of forward-looking process improvements in the safety recall document -- that safety recall document seems focused on: what was the proximate (immediate) cause of the issue, what sequence of events occurred, what is the impact, what is the plan to fix or mitigate the issue in the potentially faulty products that were shipped?

I don't actually think the safety report is throwing a worker under the bus and assigning blame. It's merely discussing the immediate cause of the issue. If we or Jalopnik are talking about blame, that's our problem, we should stop and let the company internally do its thing -- which is hopefully some kind of five-whys / blameless postmortem mode.
posted by are-coral-made at 3:24 PM on February 5, 2021 [7 favorites]

Subaru employs the blame Meredith defense. An oldie but still loved in management circles through out the world.
posted by cmfletcher at 3:28 PM on February 5, 2021 [1 favorite]

first comment - "I bet they gave him a stern torquing to."

Nobody expects the Spanish Inquisition.
posted by jamjam at 3:31 PM on February 5, 2021 [4 favorites]

The Spanish Inquisition should be able to torquemada making that mistake again.
posted by Greg_Ace at 3:40 PM on February 5, 2021 [18 favorites]

Blaming the operator is weak.

So I could be wrong, here but I have some experience in the (non-automotive) regulated world. To me this reads as explaining to your regulator how you know exactly which vehicles need to be recalled. If you for sure know problem is "Steve and only Steve fucked up," you can't cover for Steve and say "Our training programming is bad and we had crappy QC" because then your recall is much larger here.

As you suggest, IME we'd also be adding things like another QC check or a post-training check so a second untrained employee couldn't make exactly the same mistake. But also honestly if everyone at a big company like Subaru except Steve was doing it right I'd be frustrated writing a new friggin' SOP just because Steve showed us that no, we couldn't trust people to do a good job.

I'm not in the regulated world anymore. Good for everyone, I think.
posted by mark k at 3:45 PM on February 5, 2021 [23 favorites]

The mere existence of this failure mode seems like a massive design-engineering blunder.
posted by heatherlogan at 3:52 PM on February 5, 2021 [9 favorites]

If a single person can be responsible for a life-threatening fault, several things are in play:

• You're not training well enough.
• You're not testing well enough.
• Potentially you're not paying enough for what is obviously some sort of skilled role that you're pretenting is unskilled.

posted by krisjohn at 4:09 PM on February 5, 2021 [27 favorites]

If you for sure know problem is "Steve and only Steve fucked up," you can't cover for Steve and say "Our training programming is bad and we had crappy QC" because then your recall is much larger here.

then you come up with an action plan and perhaps remedial training for everyone and additional QC eyes on the process, after which everyone just does what the hell they want and management ignores it because they don't feel like spending the day fighting with people or hassling with the union and everything's ok until the next time

rinse and repeat

disclaimer - i know nothing about cars - (however those people who found jobs with us after chrysler or gm knew nothing about what we make either) - but i know a lot about management and how they like to appear as if they're doing something about quality when they're not
posted by pyramid termite at 4:16 PM on February 5, 2021 [3 favorites]

first comment - "I bet they gave him a stern torquing to."

Nobody expects the Spanish Inquisition

new connotations for auto-da-fe
posted by logicpunk at 4:40 PM on February 5, 2021 [7 favorites]

More like spanner inquisition amirite?
posted by ardgedee at 4:48 PM on February 5, 2021 [18 favorites]

You know, an awful lot of these concerns seem to be directly addressed in the linked Safety Recall Report. This document isn't very long and I found it pretty easy to read as a non-expert. The problem was discovered during routine random sampling of the manufactured component, and the cause was immediately identified. They then performed an audit and determined that every other employee in the shift was performing the assembly correctly, and furthermore that the training procedure was correct. They then issued a safety recall for all of the 400-ish vehicles the employee had worked on, offering to correct the problem for free. The whole thing took about a month (a month which included the winter holidays!); I don't know how quickly newly-manufactured cars sell to the end owner, but it wouldn't surprise me if most of them were still on dealer's lots at that time. Honestly this seems exactly like the system working as intended, and as someone who doesn't know all that much about the regulatory system around car manufacturing safety, this little window into it gives me a feeling of confidence that the system is well-designed.

I'm not really prone to siding with a corporation, but the notion that the problem was "pinned on one poor factory worker" by Subaru appears to be an invention of the Jalopnik reporter. The Safety Recall Report just explains how they know the scope of the problem is limited to the 400-ish vehicles they're recalling. And as for the idea that it's unlikely that the problem could really just be with a single employee... if you've never trained someone on a standard procedure and had them appear to understand it correctly, only to later discover they're doing it wrong, possibly because they've decided they know better than you and have "discovered" a "better" technique... well, count yourself lucky.
posted by biogeo at 4:50 PM on February 5, 2021 [63 favorites]

first comment - "I bet they gave him a stern torquing to."

Yeah, if that guy isn't already a MeFite, we need to extend an invitation.
posted by Naberius at 4:56 PM on February 5, 2021 [2 favorites]

There's another good comment later in the thread about a qc looking at the timestamps on the torque wrenches and figuring out that the operator was torquing a single bolt to spec multiple times instead of torquing the multiple bolts they were supposed to torque. Never underestimate the laziness of human beings.
posted by clockwork at 5:18 PM on February 5, 2021 [5 favorites]

There's another good comment later in the thread about a qc looking at the timestamps on the torque wrenches and figuring out that the operator was torquing a single bolt to spec multiple times instead of torquing the multiple bolts they were supposed to torque. Never underestimate the laziness of human beings.

We ran into this on my assembly line years ago. The fix was implementing a minimum time limit between torques before they would be accepted. It became more of a hassle for the operator to fake the system than it was to just follow proper torquing procedure.
posted by TrialByMedia at 6:28 PM on February 5, 2021 [11 favorites]

Also a functional union system wouldn't have allowed any of this to happen in the first place.

Hmm... Gather 'round, kids and listen to the tale of the UAW and the General Motors (plug in your favorite location) plant...
posted by 2N2222 at 7:50 PM on February 5, 2021 [6 favorites]

Yeah, as both a former UAW member and a former quality engineer, the existence of a union in no way would prevent an issue like this. Not even close. At best, it's totally irrelevant to the situation. At worst, if you had an employee that was deliberately and maliciously trying to sabotage product quality, the union would make it harder to eliminate the issue.
posted by TrialByMedia at 8:46 PM on February 5, 2021 [3 favorites]

looking at the timestamps on the torque wrenches

those are some classy bougie ratchets
posted by flabdablet at 9:05 PM on February 5, 2021 [6 favorites]

I didn't say "existence of union" I said "functional union".
posted by bleep at 9:06 PM on February 5, 2021

Yeah this seems like a textbook 5 why's problem.

Well, that's a Toyota thing, and this is Subaru 😏
posted by pwnguin at 9:33 PM on February 5, 2021 [5 favorites]

We are forbidden from using "man" as the root cause of a problem from a pool of the 5M and Es.

Man, Method, Material, Measurement, Machine, and Environment.

If "man" is your problem then your problem is actually method. Or at least that's what we find 99% of the time. (1% of the time it is a "man" issue but the "man" isn't the faulty worker it's the manager that didn't fucking do anything about them)
posted by Slackermagee at 10:27 PM on February 5, 2021 [7 favorites]

>A question to ask is: is it worth changing the process to reduce the chance of a similar issue recurring in future? (maybe it is, maybe it isn't, the optimal process will balance cost of QA & QA effectiveness and will likely have some nonzero rate of shipping manufacturing defects) If it's worth changing, how will the process be improved to reduce the chances of this happening again?
This is in my playbook for "sideline the things you learned from this mistake."

People are terrible at assessing the chances of this happening again. People consistently amplify minor risks and exist blind significant risks. And particularly don't rely on cost-benefit: the script says "cost-benefit" is the trade-off that this workflow or line of reasoning is optimising and it doesn't have room for repilutational impact of (+) transparency / (---) car can leave factory with undetected crash-capable flaw from the system letting manufacturing steps go unchecked by a single incompetent or malcious colleague.
posted by k3ninho at 11:39 PM on February 5, 2021 [1 favorite]

That's totally to weird. They (the company) should have caught this error on the first car that passed that worker's station. Where were the QC people?
posted by james33 at 4:14 AM on February 6, 2021

my new torquing technique is unstoppable
posted by scruss at 5:02 AM on February 6, 2021 [7 favorites]

Toyota owns something like 30% of Subaru.
posted by postel's law at 5:54 AM on February 6, 2021

That's totally to weird. They (the company) should have caught this error on the first car that passed that worker's station. Where were the QC people?

I'm curious if this is a statement based on specific knowledge of the car or a general expectation?

I know squat about car assembly lines and there are some failures you should of course catch. So this could be one.

But one of our mantras in QC type stuff, one I've carried with me, was "you can't test quality into a system." If you can't trust that people are not just blowing off procedure, or it's a bad procedure, it's impossible to test enough to be sure you have no defects. QC testing samples places you expect to find variance, then takes abnormalities very seriously (like the proverbial canary in a coal mine); a sign something else could be wrong. But it'll never check every nut and bolt.
posted by mark k at 9:36 AM on February 6, 2021 [5 favorites]

And here we just bought a Subaru. :-(

2016 though, so no problem! :-)
posted by sneebler at 9:43 AM on February 6, 2021

Something to add to the taxonomy of mistakes.
posted by HotToddy at 12:16 PM on February 6, 2021 [2 favorites]

“It would be a poor day at work to come in and find out that you personally were responsible for a recall at the multinational automaker that you work for, but then again it would also be just another one of life’s unique experiences.”

God I love Jalopnik sometimes.
posted by midmarch snowman at 12:52 PM on February 6, 2021 [3 favorites]

A single associate assigned to a specific production line working between the production dates specified (12/14/2020 – 12/21/2020) was found to be using an improper torque wrench technique.

Subaru gives its US workers a week and half of paid vacation at Christmas, so probably this person was a floater filling in for a regular employee on that line who was taking another week before Christmas break off to give themselves almost three full weeks of continuous vacation, and the floater found that they just could not keep up.
posted by jamjam at 12:44 AM on February 7, 2021 [2 favorites]

If this was a place I worked, either the bolt, the assembly, or the torque wrench would be named after this guy.
posted by Huffy Puffy at 6:47 AM on February 7, 2021 [1 favorite]