Hacker News Reader

> Supply Chain is not that big a deal to fix

This is kind of irrelevant, but somewhat amusing, so I figure I'll tell it. My neighbour (in the middle of nowhere in rural Japan) was some kind of automotive manufacturing expert before he retired. I don't even know his name (he tells me to call him "Tom"), but we go out drinking occasionally. He must have been a big wig because he spent a considerable amount of time in the US and even has a personally signed thank you card from George Bush Sr.

Anyway, we were at the bar and I was telling him about the use of kanban in software development. My wife interrupted and asked what kanban was (because in Japanese "kanban" just means "signboard"). I tried to explain the concept, but she remained confused. He said, "I can explain it very easily". He picked up an empty plate and handed it to my wife. "We're out of fish", he said, motioning to the bar. :-)

> "We're out of fish", he said, motioning to the bar. :-)

I don't get it.

I assume it means that the act of _sourcing_ something at the time it's needed is a much harder problem than it appears; simply because a source exists does not mean that it's the right source, or that it's even stocked.

JIT

We had an empty plate. We send that plate back to the bar. The sushi chef slices fish and puts it on the plate. If he goes below a certain amount of tuna, then he sends an empty tuna container to the sous chef. The sous chef takes tuna from the refrigerator and puts it in the container. If the refrigerator goes below a certain amount of tuna, the manager writes a note and puts it in an envelope for the fish monger. The fish monger gets the note and sends more tuna. If the fish monger goes below a certain amount of tuna, he sends a note to his agent at the fish auction. The agent at the fish auction buys a tuna and sends it to the fish monger. When the tuna at the auction is all sold, the fishermen bring more from their boats. When the boats are empty, the fishermen go out to sea to catch more tuna.

Only, we've caught almost all the tuna in the ocean, so please make sure to ask for something else at the bar.

Edit: Just to be clear, that was not Tom's point. His point was that the plate was empty, so by simply noticing it is empty, we can ask for more fish which "pulls" all the activity that ends up producing fish on our plate. In a kanban system, you only really need to worry about how much inventory you want to carry at each station and the connection to the previous station.

I think the joke only works if you know what "kanban" is. Even with your explanation, I still don't know what "kanban" is or how it relates to fish.

You need fish. You ask for it.

If you don't ask for it you don't get it.

Its just a organized way of building a todo list which has a few features like:

1. Priority.

2. Ability to see the work-item move through various stages of work(Accepted, in-progress, blocked-by, done).

3. Means to track the progress of the work-item.

4. Means to the improve the process through methods described by the process.

> I don't get it.

That's actually a very meta reply (like that joke about UDP).

>Supply Chain is not that big a deal to fix. They will sort out there problems there, they just imo need a few vets from a traditional car company.

Tesla have been trying to and they are still running into many problems, many industry veteran hires and tens of thousands of cars sold later.

I have talked to many people who have worked/work in production as well, and unfortunately they don't share with you the same optimism on how easy it is to scale up supply chain for a large scale luxury auto company. Daimler-Benz has had major reliability problems with both Chrysler and Mercedes and it took Mercedes over a decade to get back to industry average, while Chrysler is still in the gutters.

It's because of the fast pace of the company. I doubt they are any worse than any other company introducing a new model. As a new model recedes into the past quality issues should decrease linearly, and maybe faster. Alot of the BSR issues (often caused by worker inaccuracy) that Tesla faces today are probably catchable nowadays with a convnet before they even go down the line.

If I had to guess right now, many years after leaving I would guess that Chrysler will eternally be a fail, it's culture is just wrong at every level. Tesla isn't hamstrung by that, or the UAW either (yet).

"I doubt they are any worse than any other company introducing a new model."

Except they are. The Model S is 5 years old (even the refresh model is a year old) and they are still running into parts and build issues. It has improved a long way but during the whole process they had to cut the number of interior and exterior colors/trims just to streamline the production. Other manufacturers introduces new fit/finish and trim options along a model's life span and Tesla does the opposite.

Case in point, the original P85+ came with sports trims inside (such as the red piping on the leather seats) and some unique fit and finishes, with unique sports suspensions and chassis tuning. But later on they eliminated all of those and now a top of the line P100D have the exact interior trim and chassis tuning as the cheapest S60 that cost less than half as much, all for the sake of streamlined production.

And even then they can't find parts for when cars need service or are in an accident.

It sounds like I won't convince you, and it probably shouldn't be convincing because it's Chrysler, but we were having problems like that and worse on a 20 year old platform. The way I got involved in Federal Torque was because we delivered a brand new vehicle that basically fell apart (lost an axle iirc) as the customer was cruising down a freeway. Even among all the problems we had, that was a pretty big deal. As a result we had many meetings about what a serious issue this was.

Chrysler was a special case. First it was saved in the 80s when it ran itself in a ditch and when rolling up a good head of steam, bought was left of AMC, and had a good partnership with Mitsubishi.

The in 98 along comes that infamous Daimler Mercedes merger where Mercedes basically looted Chrysler to prop themselves up. At that time Chrysler was making money hand over fist with their minivan.

Once bled dry, a lot because most new car projects dried up or worse, the platform sharing using older Mercedes setups, they got sold to an investment firm. There were even claims of value gone missing during that sale.

A more apt comparison is watching GM. After bankruptcy they still pushed out the first generation Volt, the second generation took longer but arrive in 2016. However the real kicked was how fast the Bolt went from conception to production. In 2015 it was shown in concept, Jan 2016 in production form, and in customer hands by end of the same year.

The Volt was rushed. As a hybrid, it's pretty good. As a car (ergonomics, user-friendliness, etc) it's not so hot. The B-pillar is in your way when you turn your head to look left. The use of capacitance switches in a car ought to be a firing offense (no tactile feedback and muscle memory is useless because there's nothing for the fingertip to guide the last millimeter of motion). Honda is learning this with the low ratings on their new volume knob-less radio.

The Bolt on the other hand, seems to addressed all those. The local car show is in a couple of weeks and I hope to see one there.

Totally agree about "The merger of equals". No such thing. The Germans took Chrysler for everything they could, like you said - to prop up product quality disasters like the ML and E class (a coworker went through 5 window regulators on his E-430 over 2 years. I had something break on my ML-320 on average every 8.5 weeks)

You forgot the biggest scam: 40 YEARS of rolling back odometers on used cars, and selling them as new (or substantially less used) http://articles.latimes.com/1987-06-25/news/mn-10607_1_odome...

My mom bought a Chrysler LeBaron that, in one year of ownership, the brakes, exhaust, cylinder head, a/c and shock absorbers required service. We were later informed by a federal court that her car was one of the "rollback" vehicles. Chrysler denied it and basically told us "see you in court."

I don't dispute your facts at all, but I think you've strongly missed the key point of how disruption works. Not having as many trim and color options because they can't shove them out the door fast enough is a textbook example of "good enough". It is this very industry's not too distant history from which the quip "they can have it in any color they like, so long as its black" comes from.

What I mean is, you appear to be trying to apply the standard of a very broadly experienced car magazine critic, which yes, will absolutely 100% find a laundry list of grave shortcomings. And they won't matter.

You've strongly missed the point he was making. Tesla has such problems with their supply chain management that they have had to reduce parts and SKUs to the point of using mass grade finishings on lower-luxury class vehicle.

Except in most cases disruption doesn't work. The canonical examples from "The Innovator's Dilemma" all fall apart under scrutiny (as in the "disrupted" companies end up eating their disrupters for breakfast). GM is now doing a better job of executing in Tesla's space than Tesla, and that situation is not going to get better for Tesla.

http://www.newyorker.com/magazine/2014/06/23/the-disruption-...

Isn't Elon a sort of supply chain pariah at this point though? His learnings from SpaceX and Tesla so far probably make him one of the most experienced people in modern supply chains.

Not sure 'pariah' was the word you were looking for?

Trying to figure out what word he wanted... prophet?

Paragon?

Expert.

Messiah?

Guru?

It unintentionally might be accurate, though. ~90% of the parts in cars nowadays are manufactured by huge companies that sell to all the major automakers, worldwide. Tesla just doesn't have the volume to get the consistency of supply and quality that the other majors do.

Yes, this is a chicken-and-egg problem. But a real one nonetheless.

Pariah is not the word you're looking for, but either way I'd say the answer is certainly not. Making a few very expensive rockets and luxury cars is not anything like mass producing vehicles on a scale like VW/Toyota/GM/Ford/etc.

I'd say a company like Foxconn is probably a much better example. They are able to mass produce millions of widgets with astonishingly few defects and even begin doing it in secret. Very impressive logistics. If just one component of the iPhone isn't ready the whole line grinds to a halt.

I'd like to know what vehicle programs you were on at DMX during that time, because other than WK1, L cars were a mess, and now that I think about it, nearly every other DMX B/C segment program I was involved with was a mess with respect to quality too.

fyi, I cut my teeth on WK1 and WK2. I love JNAP, and Jeep - I found a home there.

Nice to hear. I worked on B/AB, D. My impression was that everything seemed to be a mess, but especially management. My direct supervisor (an area manager) wanted me to fudge the numbers and lie to his boss (a center manager). By then, I'd had enough and left.

I'm very curious how you were able to save the company millions of dollars.

It was an exercise in creative problem solving. I'd query the data to find out where on the line the most (costly x incidence) problems were occurring and then devise a solution. If a worker was scratching the paint every time he put his tool down the secure a fastener, we'd get a heavy rubber guard fabricated, change his man assignment and have him put down the guard before he did his little peice, or even better change the tool if that was possible. If you are curious, yes there were workers who would damage the vehicle every time and let it go to rework. Don't even get me started on labor. :)

>Don't even get me started on labor. :)

Wow.

Well, programmers do the same thing.

Blah blah make better software blah blah world makes better idiots.

Let me guess when asked why they said "that's how we've always done it"

Haha, yep. That or a complaint about their man assignment changing and it being hard to keep up.

How did they gather comprehensive data like that? Seems like you'd need an employee to watch every employee, recording everything they did.

The line is Body Shop->Paint->Trim->Chassis->Final Vehicle.

Final Vehicle is concerned with testing and certification and whatever other finishing touches need to be performed. It also operates the repair department, and the 'yard' with all the defect jobs. Our yard was so packed the company was renting new yards away from the factory.

The repairmen who worked in the bays and yard were stone cold experts on the vehicles and they kept written logs. Those logs were typed up and put in the database by the office clerks. These were mostly signal. There were also networked terminals at some stations but not all down the line.

Finding the root cause of an issue was basically like git-bisect. You knew where a part was added and where it ended up. So you just needed to narrow it down. For a high priority issue, I'd go do a visual inspection myself at the mid station before a part got covered by another part. Broken? Go to the mid station between here and start. Not broken, go the other way. We got stumped sometimes, but mostly we could narrow it down pretty fast. For the workers, it's impossible to lose your job so there isn't a care factor about covering up shoddy workmanship.

Binary search :)

My wife works in supply chain (top 5 cosmetics company in Brazil) as a senior analyst and she alone managed to save several million dollars in the last couple of years just by doing the right stuff with Microsoft Excel, the right stuff nobody was doing before she joined them. You would be surprised how amateurish big rich companies can seem sometimes.

What did you do save them money?