Hacker News Reader

Comment by Spivak | Hacker News Reader

While this answers the question of why we couldn't respond to the bottlenecks, I guess I'm still missing the piece of what caused them in the first place. Like sure, "this silly little global pandemic thing" but it's not really an enlightening answer without more detail as to what the pandemic caused that shocked the system.

View on HN

It was predicted by a lot of people with experience in supply chain management or business.

https://en.wikipedia.org/wiki/Bullwhip_effect

Basically - complex systems behave in complex ways. When you have a small shift in consumer demand, it gets amplified up the supply chain, such that you need potentially big shifts in the production of intermediate components. We had a big shift in consumer demand. The supply chain basically can't cope, and so we get chaotic behavior until consumer demand normalizes and revised sales forecasts get propagated up the supply chain.

People were locked down so they stopped spending money on services (hotels, restaurants, travel, massage, etc) and instead bought more goods. This happened at the same time governments printed trillions and pumped it into the hands of consumers and businesses. The result was a 20% surge in container volumes, which our supply chain infra just wasn't ready for.

As well as rolling labor blackouts as factories had to shut down (repeatedly) because people got sick and they couldn't risk everyone getting sick. And distribution centers. And retail. Etc etc.

Since a bottleneck implies a vessel with fluid passing through, I'll stay with that analogy. There's actually no bottleneck; the pipe is basically the same size as ever. (Actually just slightly smaller because of capacity that was destroyed, rather than taken temporarily offline, during the slowdown.[0]) But the pipe in the first place was only exactly as big as it needed to be, because of the "maximizing ROE" stuff he mentions early in the thread. So when the fluid volume suddenly expands (slingshot effect following the slowdown), the whole system becomes a bottleneck.

[0] Capacity destruction: "Nobody's buying our stuff, we can't afford to keep this one ship/plant/facility/machine running anymore or we'll go out of business." But since every other player in their industry who might've bought it is in the same boat, maybe they're forced to sell to a scrapper, who promptly disassembles the ship/plant/facility/machine. I got my house this way. It was a rental property owned by a company that could no longer afford to keep it, so I bought it, took it off the rental market, and thereby slightly reduced rental capacity in this town. Compare this to an airline temporarily parking a slew of their planes in a densely-packed parking pattern at an un-busy airfield near a nice runway.

In normal times, people mostly shrug off biological transfer with other people. Maybe you avoid the odd symptomatic person who comes to work anyway, but you don't act as if everyone else is trying to get you sick. This allows for close working conditions, less sanitary practices, etc. The pandemic invalidated this assumption, requiring workers to become less efficient in response. This is the fundamental dynamic - the secondary effects of denial/overcompensation created additional problems.

It also didn't help that everyone was expecting an economic slowdown, and then the government gave trillions of dollars to corporations to juice the stonk market. The resulting price signals induced a ton of aggregate demand for production capacity that simply wasn't there.

Have you seen the YouTube video where a waterwheel [1] demonstrates predictable behavior under one set of conditions, but with a small change, the system moves into chaotic behavior?

I figure the supply chain is like that. Under certain conditions, JIT is fine. Given sufficient disruptions, the system descends into chaos. Note that we have humans in the loop, too, so expect non-linearities.

I could be wrong. Corrections are welcome.

[1] https://youtu.be/Lx8gMBJBlP8

Reply on news.ycombinator.com