Better find one who has actual restaurant experience and not just experience from a lab.
The problem with the idea of engineering-level feedback control is that it is completely unrealistic in the restaurant world. It may work for extremely simple linear physical systems, but not for businesses where decisions are often binary or map to a small number of discrete values. If it’s not worthwhile for a restaurant to be open when limited to 25 percent of capacity, it’s going to be closed if that’s the limit, i.e., it will be running at 0 percent of capacity. Which would be the same as if the limit was 23.7 percent of capacity and likely if the limit was 26.1 percent. The oldest restaurant in San Francisco, Tadich Grill, seems to indicate that they will only reopen at 50 percent of indoor capacity, which means that a limit of 42 percent or 10 percent would both map to 0.
Of course, the actual implications of a limit would vary dramatically from restaurant to restaurant and nobody would have sufficient information to predict exactly what the impact would be of setting limits to some very precise values. And the feedback loop between implementing a measure to seeing a result may be long. Then there is the issue of enforcement. It’s easy to be crtitcal of simplistic measures, like all bars must be closed or all restaurants must close at 9, but at least it’s relatively easy to enforce them. How easy would it be to enforce a capacity limit of 63.75 percent? Or a limit of 80 percent of normal hours? What would the latter even mean? It’s one thing to cut back during lunch rush or prime-time dinner hours and a different thing to close the restaurant 2-4 when there are likely few diners. The whole feedback idea is a pipe dream.
Actually, the existing 4-tier system in CA is a feedback control system. Just a highly simplistic one. That is, a metric is established and measured, and an action is taken at threshold levels. Feedback control is not only not a pipe dream, it’s what’s in place right now.
A better feedback control would be dimmer switch. In terms of restaurants, cutting back or expanding hours rather than cutting back or expanding occupancy, is the better control. For some restaurants, though, allowing some indoor customers, for example people seated next to open windows, would be better than only allowing outdoors or takeout.
The best would be engineering-level feedback. I haven’t in the past talked about that just in terms of restaurants, but rather all businesses collectively.
In terms of restaurants, cutting back hours rather than cutting back occupancy is the better control. That’s what’s being done in some places, like Sweden.
Except that a the kind of smooth adjustments that you’ve been looking for would require information that’s not there. For instance, what would be the impact of letting restaurant X serve a few people near a window? It would probably be different that letting restaurant Y do so if their dining spaces are different. Who would make those kinds of decisions? And what would the rule be for scaling back hours? And why would it be better than scaling back occupancy? If we agree that crowded spaces are bad, wouldn’t occupancy be a more obvious target? I’ve never heard that the virus is particularly bad during a certain time of the day, so why would it make sense to force a bar to close at, say, 9, if the result is that it’s completely packed at 8:30 with people trying to get a last drink?
Most customers at most places eat in a fairly narrow window of a few hours anyway. I think making them close early is just about drunk people getting sloppy about hygiene.
Perbacco in SF has a sign in the window that they’ll open when they can seat 50%, which I believe would be 75 customers (due to their huge upstairs area used mostly for private events).
That’s true, and I think it’s an important point. If there are hours restrictions, each restaurant should be able to decide how to allocate them to enable/maximize business profitability. In other words, you might have four hours that you can be open indoors, but you can decide when those are. If both occupancy and hours limits are too stringent in a given restaurant owner’s assessment, the restaurant could voluntarily be kept closed.
There are lots of restaurants that currently fall into that category. What would normally be the highest grossing restaurant in SF, the Slanted Door, has been “temporarily” closed for about a year.
Hours allowed could be an inverse function of percent occupancy, perhaps. E.g., if you have to be at 50% occupancy to justify indoor operation, then you get fewer hours to be open.
Hours allowed could be a function of average patron separation distance.
There is currently no feedback loop in place - a feedback loop wouldn’t have predefined outcomes, like now in the pandemic, but would adjust depending on real-time data. You are desperately trying to show something which doesn’t exists by stretching the definition to a point nobody is using it
You really don’t have a clue as to what “feedback control” means, so there’s hardly any point in discussing it. But I’ll try one last time.
The Tier levels call out a number of fixed thresholds in terms of measurable metrics (like test positivity). Movement between Tiers is the government response to changes in the actual population statistics having passed (above or below) one of the threshold levels. That’s the control. Restrictions are either tightened or relaxed. After the response, the values of the measured statistics will change. That new information is fed back to the Tier Levels control, and if the modified statistics again cross a Tier threshold (in either direction) action is again taken. And so on, with no limit on how many times this control loop is repeated. The amount of the corrective action depends on how far the measured metrics are from the ultimate desired values. In CA the control consists four discrete levels of corrective action, increasingly restrictive the further the statistics are from the ultimate desired values. The set of ultimate desired values reflects a very low level of covid spread.
That’s called feedback control. Set a desired level. Measure the actual level. Feed that information to the control, which makes adjustments to move the actual level towards the desired level. Repeat. It’s being done “manually” and with crude step-wise thresholds in CA, but that doesn’t change the concept. Look it up for Pete’s sake.
You keep reiterating your feedback control fantasy as if that would make it relate more to the real world, in which we are many orders of magnitude away from having enough data.
If we had lots of data, we could do reverse contact tracing, and there would be no need to impose drastic restrictions on businesses except locally during outbreaks (as in Australia).
There’s no reason to think that adjusting restaurants’ hours would reduce infections. Most customers want to eat within a block of several hours. Off-peak hours when the place isn’t full present less risk so cutting them will have less effect.
Distance is mostly bullshit. People have been infected 20 feet from the person they caught it from. What matters is ventilation.
First, I really wish you’d stop using the disparaging term “fantasy”. CA is exercising feedback control even as you utter that expression. It’s clunky and I don’t agree with all of it, but it’s a form of feedback control, like it or not.
The ultimate objective of controlling virus spread and infection is to reach a condition with simultaneously very low values for all of the primary virulence metrics. Zero cases, zero hospitalizations, and zero deaths is not an attainable objective. We may never be able to stamp out covid-19 completely worldwide, or even locally, just as completely stamping out the common cold or the flu is not a realistic goal. But it’s felt that the spread can be reduced to a very low level, as measured by a combination of metrics taken together; i.e., positivity rate, cases per 100K, ICU usage. That’s the ultimate desired, stable condition.
The further the departure from this desired, realistic, ultimate, stable condition, the stronger the control (restrictions) to reduce spread, as reflected in the four-tier control in CA.