Why Covid and efficiency makes for long queues

An anecdote told to me by a university lecturer has frequently proved useful in anticipating events.

In the days when these things were still new and incredibly expensive, a large utility company bought a highly specialized laser printer to print their bills. This printer was truly impressive, not only printing five million invoices each month, but folding each invoice and placing it in an envelope. This process was much, much faster than any other system available. But it was expensive, so they could only afford one. It could print an invoice for each of the company's five million customers each month with a day to spare.

Things went well for a while.  A whole room of clerks and mail sorters used by the previous system was no longer necessary. 

But one day it failed. A part broke. It took a week for the replacement part to arrive and the mailing system to start working again. 

The question the lecturer asked was this: how long did it take for the system to return to issuing bills on time, and as a result of the delays, how much did the outage cost?

Your arithmetic here will obviously depend upon your assumptions about the size of the average bill and commercial interest rates, but one thing should stand out: it will take years not months for the system to return to billing customers on time. 

This is a common problem when systems evolve (or are designed) which operate at or near capacity: a disruption leads to queue growth, and if the system is "too efficient" , the backlog will never be cleared. The disruptions may be caused by random fluctuations in the rate of new jobs arriving in the queue (think cashiers in a bank, public transit, and hospital waiting lists), or they may be caused by major unexpected events. 

Covid-19 is one such major event. There are a lot of queues in the world, and many of them have been disrupted. A few examples spring to mind:

• Non-urgent medical treatment. Italian authorities warned early on about the risks of shutting down non-essential treatments following their own experiences. It didn't seem to help: the cry everywhere was to prioritize Covid-19 treatment and shut down non-essential services, leaving extended delays and people who will die waiting for diagnosis and treatment.
• Driving license road tests. These are labor intensive (one examiner per student). Nobody likes the idea of driving test examiners having much spare time during normal operations: a consequence is that when there is a disruption, new examiners will need to be hired and trained. There's not a pool of unemployed driving examiners just waiting for the call. 
• Semiconductor manufacture. The car industry has had to shut down production lines or stockpile partly-finished vehicles. This appears to have been a result of not placing regular orders due to an expected downturn, combined with a rapid increase in queue length due to competition from other industries needing semiconductors. (Note that this isn't just due to Covid. There have been fires affecting production as well as major power outages too.)
• Container transport. There are major delays in shipping containers into some US ports due to congestion, containers are not where they are needed, and shipping rates and times have increased as a result. There are also serious delays on exports from China due to Covid disruptions, and even a suggestion of increased US inflation due to the problems.
  

All these long disrupted queues interact with each other. The pandemic hasn't finished, but even when it does, expect the effects to linger for a long time afterwards.

There are two key lessons here:

• Just because something urgent and important is happening doesn't mean that everything else should be halted or ignored. (In a business continuity plan we often add a senior management team whose job is to continue running day to day activities to emphasize this point). It may be difficult or impossible to catch up afterwards.

• If you design a system (or a system has evolved) where equipment or facilities are being used near 100% capacity, anticipate that any disruption will produce delays for extended periods. Remember that the system may not be your own, but that of a supplier, a logistics company, or some other part of your supply chain.
  

It's efficient to have resources operating at 100% capacity. But when time is of the essence, it's pretty risky too.