Shortfall

<chirp, chirp> The non-business line again. “Moire here.”

“Hiya, Sy, it’s Eddie. I’m taking orders for tonight’s deliveries. I got some nice-looking artichokes here, how about a garlic and artichoke pizza?”

“No thanks, Eddie, I’ll stick with my usual pepperoni. Wait, you got any ham?”

“Sure.”

“Let’s go with a Hawai’ian.”

“Sy, we’ve had this conversation. You want pineapple on pizza you open a can and dump some on there after I leave the premises and don’t tell me. I got standards!”

“Calm down, Eddie, just yanking your chain. Yeah, do me one of those garlic and artichoke ones. Sounds more classical.”

“That’s better. I got you in the 6:15 wave, OK? Hey, that reminds me. I read your post series about waves and that got me thinking.”

“Nice to know someone reads them.”

“Well, things are real quiet, just me in the kitchen these days so I’m scraping the barrel, you know?”

“Ouch.”

“Gotcha back. Anyhow, that series was all about wiggly waves that repeat regular-like, right? I get that scientists like ’em ’cause they’re easy to calculate with. But that Logic Curve you wrote about goes up and doesn’t come back down again. Does anybody do math with that kind?”

Logistic Curve — blue line,
Associated slope — red line

“Logistic Curve. ‘Logic Curve‘ isn’t a thing. The mathematicians have come up with a plethora of curves and curve families. The physicists have found uses for many of them. The Logistic Curve, for instance, is one of the first tools they take off the shelf for systems that have both lower and upper limits. You’ve seen a lot about how it’s applied to epidemiology. People also use it for ecology, economics, linguistics, chemistry, even agriculture.”

“What do the top and bottom lines have to do with each other?”

“Ah. Sorry I hadn’t made that clear. OK, find a blank page in your order pad. At the top draw a horizontal zig-zag line like a series of 45‑degree triangles touching corners.”

“45 degrees is easy — that’s an 8-slice pizza. Done.”

“You’ve just drawn what’s called a triangle wave, no surprise. OK, now right under that, you’re going to draw another wave that shows the slope of each triangle segment. Where the triangle line goes up you’ve got a positive slope that goes up one unit for every unit across so draw a line at plus‑one, OK?”

“A-ha. Got it.”

“Where the triangle line goes down you’ve got a negative slope, minus‑one.”

“What about where the triangles got points?”

“Just draw a vertical line to connect the slope segments. What’s the completed second line look like?”

“A zig-zag bunch of square boxes. Hey, wait, we made the second line be the slopes for all the pieces, right? Lemme go check the picture in the ‘Curve‘ post. So what you’re saying is … the red line is all the slopes along the blue line … OK, can I say that the red line is how fast stuff is coming at me and the blue line is the backlog?”

“Half-right. For what we’re talking about, ‘slope‘ is whatevers per time‑unit. The blue line shows how much total has come at you so far. Backlog is a little more complicated.”

“I gotta go back and read those posts again. Now I see why they’re saying ‘flattening the curve‘ — they want the blue line to not climb so fast.”

“That’s part of it.. Flattening that red-line curve as much as we can is important. That’s what the masks and social distancing are about. Maybe as many people get sick, total, but if they trickle in instead of flooding in then they don’t overload the system. Here, I’ll send a sketch to your phone.”

“Got it, but there’s lots of lines there.”

“The red line is your completion rate — pizza orders per hour, patients per day, whatever. The red line goes flat because having only one oven limits your throughput. The gray part above it is pizzas per hour you couldn’t bake or patients your hospital couldn’t take that day. The green line is doable business; the black line shows how more capacity would have improved things.”

“Reduce the incoming, raise the capacity or lose the people. Whoa.”

~~ Rich Olcott

Flattening The Curve

<chirp chirp> My phone’s non-business ring-tone. “Moire here.”

“Hi, Mr Moire, it’s me, Jeremy, again. Sorry for the hold-up. My phone’s on the charger now so we can keep going about the Logistics Curve and all.”

“Logistic Curve, Jeremy, singular. Logistics plural has to do with managing the details of a military or business operation. That’s quite different from population growth which is what the Logistic Curve is about. Though come to think of it, these days we’re seeing a tie‑in. So where were we?”

“We had that S-shaped Logistic Curve with exponential growth at the beginning but then it plateaus and you showed me a humpy curve that’s the slope of the other one and you said the humpy curve is like R = K*S*(N‑S) if N is everybody and S is how many are susceptible to the virus. But you kind of skipped over K.”

“True and I’ll get to K, but that ‘humpy’ curve is important. In the context of the pandemic, it’s people per day — how many catch the virus, how many show up for medical care, how many need ventilators or even mortuary care — there’s a different K for each question. The hump is what we’re trying to get control of. The K factors summarize a whole pipeline of ifs and maybes. Some of them are knobs that we may be able to use to flatten the hump.”

“We can do that? How?”

“Good question. Here, let me send your phone another image. Let me know when you receive it.”

“It’s here, Mr Moire. Looks like you’ve got three Logistic Curves but they’re stretched out different amounts.”

“Stretched out on the time axis, and that’s crucial. I generated those three plots by using different values for K. Sooner or later in all three models everyone catches the bug. In the blue-line case, though, that happens over a much longer time interval than in the red-line case. If you’re a public health official or hospital administrator you pray for the blue-line case — the slow initial rise gives you a heads-up and more time to get ready for future incoming cases. Better yet, because the cases-per-day peak is flatter you don’t need as many masks and ventilators to take of the patients and your front-line people are less likely to be over‑extended. Assuming you’ve hired enough in the first place.”

“So the government wants to reduce the K numbers to get to the blue-line case.”

“Absolutely. Keep in mind, K is such a complicated summary of things that realistic models are complex. Experienced modelers know that the more factors you put into a model, the riskier the predictions become. Anyway some of the things that go into K we can’t control, we can only measure or estimate them and try to account for what’d happen if something changes.”

“Like what?”

“Suppose you’re exposed to the virus. What’s the probability that you’ll come down with symptoms bad enough to need medical care? Current data suggests those odds depend a lot on uncontrollable things like your age and medical history. A model for a retirement community almost certainly needs a different set of K-values then a model for a college town full of teens and twenty-somethings. But that gets into a different cluster of factors.”

“That’s for sure. My grandparents are a lot more careful about their health than my crew is.”

“Which gets us into the K-factors we can at least try to manage. Simple example — you can’t catch the virus if you’re not exposed to it. That’s what Social Distancing is all about and that’s why you’re staying at home, thank you very much. Typically, models gauge that piece by surveying what fraction of the population is complying with the stay-at-home, masking and 6-feet-away rules. We need to get to 70% or better to keep the patients-per-day rate down to what the hospitals can cope with. A vaccine, when we get one, will have the same effect but that’s a year away.”

“Yeah, and if someone invents a good treatment so people don’t have to go on ventilators, that’d help the K for that end of the pipeline.”

“Get to work on it, Jeremy.”

~~ Rich Olcott

The Curve To Be Flattened

<chirp chirp> My phone’s ring-tone for an non-business call. “Moire here.”

“Mr Moire, it’s Jeremy.”

“I hope so, Jeremy, my phone shows your caller-ID. I’m glad you called instead of trying to drop by, the city being under lockdown orders and all. What’s your question?”

“Oh, no question, sir, I just called to chat. It’s lonely over here. If you’ve got the time, anything you’d like to talk about would be fine.”

“Mm… Well, I am working on a project but maybe talking it out will help get my thoughts in order. Have you seen that ‘Flatten the curve‘ chart?”

“Sure, it’s been hard to escape. They use it to tell us why we shouldn’t do group stuff while this virus is going around. Are you writing about where the chart comes from?”

“That’s my project, all right. There’re two ways to get to that chart and I’m trying to decide which will work better. I could start from ecology studies of invading organisms taking over a new territory. At first the organisms multiply rapidly, doubling then doubling again —”

“That’s exponential growth, Mr Moire. We talked about that!”

“Just sent you an image. When researchers plot invasions they usually look like the black line, the Logistic Curve. Its height represents the organism’s population as time increases left-to-right. At the beginning there’s that exponential rise. Over on the right the growth rate slows as the plants or animals or bugs use up increasingly scarce resources. The part in the middle’s almost linear. All that’s a familiar story by now, right?”

The Logistic Curve (black) and its slope (red)

“Uh-huh. We talked a lot about ecology back in kid school except we hadn’t learned graphs yet. What’s the red curve?”

“That’s the interesting part I’m trying to write about. One way to look at it is that it’s simply the slope of the Logistic curve. See how where the Logistic is rising, the slope is rising, too? That’s the way exponentials work — ‘the higher the faster‘ as they say. The slope switches direction just where the Logistic switches from growth to slow-down. The Logistic Curve approaches its limit when the organism’s population approaches the carrying capacity of the territory. That’s also where the slope gets shallowest. Very few resources, very little expansion.”

“What’s the other way to look at it?”

“We start with the slope curve itself. It has its own straight-forward interpretation, especially if the organism is a a bacterium or virus that causes disease. Consider the population under attack as the resource. How fast will the disease spread?”

“Uh… what I keep hearing is that if more people get sick, other people will get infected faster.”

“But what happens when nearly everyone’s caught it and they’ve either recovered or left us?”

“Oh, there’ll be fewer people left to catch it so the disease spreads more slowly.”

“Let me put that into algebra. I’ll write N for the total number of people and that’ll be a constant, we hope. At any given time we’ve got S as the current number of people who are susceptible. Then (N‑S) tells us how many people are NOT susceptible. Are you with me?”

“Fine so far.”

“So from what we’ve just said, the rate of infection is low when S is low and also low when (N‑S) is low. One way to make that into an equation is to write the rate as R = K*S*(N‑S). K is just a number we can adjust to account for things like virulence and Social Distance effectiveness. If we plot R against time what shape will it have?”

“Mmm… S is nearly the same as N at the start so (N‑S) is nearly zero then. At the finish, S is nearly zero. Exactly in the middle S equals (N‑S). They each have to be higher than near-zero there. That makes R be low at each end and high in the middle. Ah, that’s sort-of the shape of the slope curve!”

“It’s exactly the shape of the slope curve. So how do we flatten it?”

<click-click, click-click> “Oops, Mr Moire, my phone battery’s about dead. Gotta go get the charger. I’ll be right back.”

“I’ll be here, Jeremy.”

~~ Rich Olcott