[rm445]

I worked for a company whose factory really was just a room - the company was a machine-builder, a B2B manufacturer of custom production equipment, but with very little investment in special-purpose machinery for its facilities or even machine tools, just an attitude that the company is the people and we can do or buy anything that our customers need.

In some ways that attitude seemed admirable, but ultimately it didn't help the company win or keep consistent business. You'd have gaggles of smart people building custom prototypes but nothing scaled up. The customers couldn't see the vision of scaling their production there, or just saw that they could get better pricing going to factory which had already invested in the right special-purpose machinery.

That's what a factory is to me - ideally reconfigurable, but a place with capital investment for production. It's good to show kids what's behind the curtain but don't get it mixed up with a prototype shop.

[calvinmorrison]

Factories, were like that. Giant Mills, Planing machines, vacuum forming tooling, welding stations, etc. Configurable, yes. Tooling, yes. It's why ford, singer and a hundred other american factories could start making bombs, guns and anything during WW2. You had machinists who could read a drafting diagram, and drafters who could draft anything up.

Today, could we do that? probably not. Not even - we don't have the basic bootstrapping tools in capacities needed, we don't have a wide group of people with the skillset.

So yes, you can make anything in a factory designed to make mostly anything

With specialization, especially like in the auto industry, you'll have one shop in mexico that gets an order 6 weeks ahead of time and has to deliver down to the day on the production schedule of ford to supply say a car headrest, and thats it. So, could we... today... maybe?

[tsss]

China can do it. No car that leaves a factory there is the same. Everything automated and yet customizable.

[calvinmorrison]

Right but big robot arms do need to be reprogrammed. In short we traded higher throughput for less flexibility. Tooling and such is expensive and complicated. A sheet brake and a welder and a mill and lathe can produce a LOT of stuff without any expensive tooling.

[detourdog]

This has always been the trade off. One of my first design jobs was working for Simco. They made boxes with fans and ionizer and you could design what ever you want as longs as it’s bent sheet metal. I’m trying to develop a small run manufacturing facility. Last step is a 9 month composite construction certification course in September. I think composites will provide freedom in the final form and flexible tooling.

[fragmede]

> big robot arms do need to be reprogrammed.

The cost of that reprogramming goes down over time, no?

[HeyLaughingBoy]

Custom machine building is a very difficult business to be in. As you mentioned, it's hard to scale. I knew of a particular service provider (custom motion systems) whose philosophy was to break even on the first job and profit on any followups.

[teitoklien]

>of a particular service provider (custom motion systems) whose philosophy was to break even on the first job and profit on any followups.

It's called Learning Curve Based Pricing / Experience Based Pricing First popularized by BCG (Boston Consulting Group) and Texas Instruments

https://commoncog.com/c/cases/texas-instruments-learning-cur...

Helped Texas Instrument become what it is, and dominate the decade that it implemented this strategy against its peers.

The guy who invented it there, was who later founded TSMC Morris Chang

[HeyLaughingBoy]

TIL. Thanks!