[Ninjak8051]

Additionally, notice that the orange data points (actual hospitalizations) are significantly under the "With Intervention" plot. So why even include the slide? Clearly the assumptions behind the "With Intervention" plot do not match reality TODAY, let alone in a few weeks.

Not a good sign if people are making actual decisions based on this.

[xienze]

> Clearly the assumptions behind the "With Intervention" plot do not match reality TODAY, let alone in a few weeks.

I feel like I’m taking crazy pills with the way people are talking about this whole thing still being out of control, when data shows the situation is even better than the most optimistic models were predicting. And by that I mean, pretend I was on HN 3 weeks ago predicting 25K deaths in the USA by today. I’d have been downvoted into oblivion and told I was delusional. But here we are.

Clearly NYC was hit far harder than the rest of the country, but it seems like every state and municipality is simply basing their models on NYC case numbers scaled by population. I don’t get it.

[ridaj]

I don't understand what you're suggesting, that California should ease up faster? It seems to me that reality being better than models had predicted is more the result of everyone flying blind due to lack of testing... California has probably been lucky that it turned out its infection was lower than New York's turned out to be. It is just hard to make a model starting from extremely shitty and noisy base data about facts on the ground, but someone still has to make very difficult public health decisions based on the best models available.

[dehrmann]

> I feel like I’m taking crazy pills with the way people are talking about this whole thing still being out of control

A week ago, I remember reading how the Bay Area was bracing for a bracing for a surge and it was the calm before the storm. The Santa Clara county covid-19 dashboard told a different story: day-over-day, new cases were roughly the same for the past two weeks. 2+ weeks into the lockdown, there was obviously no surge coming.

https://i.imgur.com/LrFZj3S.png

Looking at that chart on the April 7th, nothing pointed to a surge.

[TeMPOraL]

Isn't this just California being in the "3.6 roentgen, not great, not terrible" phase? I.e. number of new cases reported is limited by the number of tests that are able to be made?

[SpicyLemonZest]

Every region (and almost every country) is generally believed to be severely undercounting cases because of limited testing. There's not much reason to believe the problem is worse in California than elsewhere, and a severe surge would be evident in hospitalization and death rates no matter how bad testing is.

[throwaway237683]

In mid-March we were told that we are "10 days behind Italy" and shelter-in-place is not to stop the spread (already too late for that), but to give hospitals a bit of a breather before the surge of critical cases and the Death Wave that follows. That was the justification for such extraordinary measures as a state-wide lockdown.

It's mid-April now. Where is the Death Wave? Either this virus is much less infectuous than it was touted, or much less deadly. Or both.

Sweden never locked down. Denmark is opening up after a lockdown. Italy that was touted as the prime example of hell on earth, had ~21,000 people died by official count. Even if it was undercounted by 50%, that's roughly 40,000 people, mostly elders on their last breath, out of country of 60 mln. And that's the worst case so far, in the world of 7 billion people.

Disregard all that, we're still being told daily that it's "war zone", the zombie apocalypse is just around the corner, etc. Only Big Brother like surveillance is going to save us. The lockdown is not an extraordinary measure anymore, it was an absolutely necessary first step, and now we have to take it further. Much further.

I wonder what it felt like for people in 1932 Germany.

[Jommi]

Sweden didnt lock down and as a result has around TEN TIMES more deaths than its neighbours, and the death toll is raising by hundreds per day, when its hovering around 10 with Finland and Norway. I don't think you want to give Sweden as a good example.

[throwaway237683]

Wikipedia says 1,200 deaths so far in Sweden. Sure that might be ten times more than in Finland and Norway but given the dire predictions of two months ago, and even more dire predictions of a month ago, you'd expect this number to be ten times higher than it is now, or a hundred times. What about exponential growth and stuff? In a country with no lockdowns this disease should go rampant, right?

1,200 people dead of 10.3 mln is 0.012% of the populace. Not good but not scary either, for the Worst Pandemic Of Our Lifetime.

[altcognito]

I think there was genuine concern that the quarantine as implemented would not be enough. You can see lots of stories locally wherever you are about how people aren't obeying the order. Italy was still in a lot of trouble despite having had some quarantine measures in place. Hindsight shows the quarantine was effective enough to get numbers down. The US, despite reacting too late, did lockdown fully relatively quickly.

I would agree that modeling after NYC seems like a bad idea given the lack of density and public transportation. On the other hand, Michigan looked pretty exponential, and as the lagging cases came in from other states, they also began to look fairly exponential as well. Despite cases staying elevated for now, the exponential growth has died down.

The US missed the worst of hospitals being overrun and it running away in multiple cities.

[Loughla]

>The US missed the worst of hospitals being overrun and it running away in multiple cities.

I want to add ", so far" to the end of that sentence. I do not hold hope that this will be true as we fight among ourselves to reopen various portions of the country/economy.

[xienze]

> I do not hold hope that this will be true as we fight among ourselves to reopen various portions of the country/economy.

But things won’t be the same after reopening as they were before. People will be wearing masks (I see a lot of that already), keeping distance, etc. Couple that with common sense restrictions like “keep the elderly and sick at home”, “no big public events”, “no sporting events”, and “lots of restaurants have gone out of business”, it’s clear the opportunity for infection spread will be lower.

[glaugh]

Whenever I see the IHME charts[1] (which are great) I think “I wish I could see the what the model predicted before today’s date so I can assess it.”

This shows laudable transparency, in my view.

(Also the model was directionally correct, it’s not like it was entirely the wrong shape.)

[1] https://covid19.healthdata.org/united-states-of-america/cali...

[wjossey]

The IHME model is proving to be wildly inaccurate on the downslope due to bad data coming out China. Their projections will continue to creep up in terms of deaths over the next few weeks because the predicted drop offs won’t be happening, sad to say.

If you want to see examples, check out their predictions on Italy. Their figures have been underestimating by over 50%, and the gaps will widen each day over the upcoming week. Heck, the ranges for the last few days have been outside their confidence interval.

[altcognito]

A significant portion of this probably is compliance, but I'm with you completely on "the numbers out of China, particularly at the tail end are farcical". There was a drop off a cliff in cases.

[jf22]

Could you share how you would assess the model?

Do you have a background in disease modeling or anything?

I'm curious.

[ISL]

I think the point of the plot is to suggest that intervention efforts are being quite successful relative to expectations at the time the "With Intervention" model was generated.

The brighter dashed line is there to show what happens if intervention is disabled completely.

I really like that the milestones are placed as questions, rather than dates. It is the right way to message the unknown.

[empath75]

This is ridiculous criticism. There are so many unknowns with this virus. We don’t know how many people have it, how fast it is spreading or what the mortality rate. Everything is a fairly rough estimate.

It turns out that social distancing is working better than expected, or the virus isn’t as contagious as expected, or slightly less severe than expected. That doesn’t mean the lockdown was the wrong thing to do, or that it isn’t still necessary. What it does mean is that we’ll be able to start easing restrictions as the hospitalizations continue to drop — and perhaps we can stop it completely with testing and contact tracing.

[mikekij]

My former physics professors would have not approved of a graph like this. D+ on the lab at best.

[dehrmann]

https://i.imgur.com/dJ7kHch.png

It's pretty sloppy to show surge capacity on the same graph as total hospitalizations, not active active hospitalizations.

[Retric]

The line is the average of high and low estimates. The issue we are dealing with exponential growth so 1/10th x to 10x makes it look like the estimate is wildly off even when it’s surprisingly accurate.

Further, these estimates are much older than the data so showing their accuracy is useful.