It's not about throughput or latency. It's about cost. Shipping is _way_ cheaper than anything else. Trains are quicker, but also more expensive. Barge/Train is again cheaper than a truck, but in various cases only a truck is a good option.
> I also think that is true generally today
It's not true in the slightest. If something needs to arrive quicker another modality can be used. What's important is than it is predictable. Meaning, something arrives every week around the same day. Latency can cover up scheduling issues. It's much cheaper and better to control the scheduling issue.
Yes, currently shipping by sea is cheaper, even if it means sending several containers' full of stuff to a place that doesn't need them by the time they arrive. Or worse, sending several containerfulls of stuff somewhere in anticipation of them maybe needing it soon -- because by the time they need it, it's too late to start sending it.
This is of course very wasteful, and the high latency is the source of the waste. But it still comes out ahead in the books.
What I'm saying is that hopefully maybe in the future the parameters of the equation look different, and it's no longer sensible to send huge batches of things just to patch over delivery problems relating to latency.
So effectively what has happened in lean manufacturing and product development flow-based approaches, except on a global scale.
You're again making a lot of assumptions here. What do you mean with "sending several containerfulls of stuff somewhere"? Similarly with "This is of course very wasteful". First you assume that companies often send too much stuff, then use that as a basis that there's waste. You haven't substantiated the first assumption!
> and the high latency is the source of the waste.
Again, this entirely depends on the business (commodity) and circumstances! Further, you take this as a given, while you did _not_ substantiate this at all.
For loads of commodities the time spent on the vessel does _not_ matter. For others it might be a mix. Sometimes it's important that things arrive on a regular basis. Sometimes it's not important.
> So effectively what has happened in lean manufacturing and product development flow-based approaches
Lean manufacturing car companies are happily using container vessel to transport their car parts. I've already mentioned the latency argument is missing the point quite a bit: the scheduling is important. A factory that runs out of parts is way more important than the 2 weeks vs 4-6 weeks. If the 2 weeks cannot be trusted upon it won't work.
Further, substantiate your assumptions. I don't see anything referencing a particular business, the costs associated, etc. Just guesswork.
A tiny fraction is the answer. Shipping back and forth has huge cost implications, including customs tariffs at every country the goods are held at and container/lcl costs (currently up to $20k per FCL!). Businesses aren't generally shipping loads of stuff that they don't need to be moved.
If they manage to shave a week off the overhead elsewhere (e.g. by throwing money at it), they can switch from flights to rail, and likely achieve the same quality of service at a lower cost.