/dev/chat

Faheem Mitha

05:49

Well, Bombay is officially closed. They just shut down the local trains for the general public.

For this city, that's basically the nuclear option.

Faheem Mitha

07:16

@Kusalananda For 20 people on a video conference are there lots of little windows, one for each person, like they have in those movie split screen moments?

Michael Homer

The default is usually to cut to whoever's speaking, but gallery view is pretty common

Faheem Mitha

@MichaelHomer I see. Thank you. Gallery view would be lots of little windows?

Michael Homer

^ Gallery view

Faheem Mitha

08:23

@MichaelHomer Yes, I see. Is this a "real" image, or a stock photo? Thank you.

Michael Homer

You'd have to ask them

Faheem Mitha

@MichaelHomer Oh, is it from the company web site? In which case, never mind.

Correction, the Indian Govt has shut down all train services in the country.

This is probably unprecedented.

Correction, all passenger trains. Goods trains of course will continue to run.

Confirmed cases currently heading towards 400.

Tim

11:01

good morning

I was wondering if any of you have used HPC cluster managers such as PBS, SGE, or modern ones such as Kubernetes, Mesos, YARN.

Are there good references for introducing the architectures and features of cluster managers?

Tim

13:03

@fra-san sorry to hear about the ongoing crisis in Italy. I thought it was recently under control

The disease has been spreading around the world at a nonstop speed, except in East Asia.

@FaheemMitha Sorry to hear about the same crisis in India. I too guess the reality is underestimated.

I have also been under personal crisis. sometimes sentimental and break-down

fra-san

13:30

@Tim Thanks. The general situation is not significantly worsening here, but unfortunately it's still not improving either.

Tim

14:28

Don't know how long the lockdown will last

Is there something that can be learned from the earlier experiences in some countries?

Speaking of something else: kubernetes vs openshift

old.reddit.com/r/kubernetes/comments/fm37x0/…

Faheem Mitha

14:49

@Tim Things in India still hang in the balance.

The govt has pulled out all the stops. Whether it is going to be sufficient; well, we'll find out.

Shutting down all passenger rail travel - well, that's quite something. I wonder if it has ever happened before.

Tim

Has any country end its lockdown? How long does it take usually?

Faheem Mitha

@Tim Usually?

I'm not sure where worldometers.info/coronavirus/#countries is getting its numbers from, but they often list India higher than the news articles do.

Of course, the news articles date very quickly in this sort of situation.

Currently showing India as a little under 400.

Tim

Is it correct that new case # is the best indicator of the state of disease spread?

If it is, Italy and France and UK and Japan do a great job

I think the new case numbers are not accurate. At least I know some place has significant amount of new cases but not reported in worldometers.info/coronavirus/country/us

Isaac

19:20

Some thoughts to share:
There was a way to lockdown covid19, let's call it "aggressive testing".

There are two places a virus outbreak could be detected, inside a country or outside.
If it is outside, the easiest to explain, what should be done is to test everyone that comes in from areas in risk in the world. If everyone is tested, the ones confirmed should be quarantined. And everyone that got in contact should also, in turn, be tested. Given enough intensity to this process, most of the outbreak would be detected and contained. Few, if any, would get infected inside the country. Of course, the process could be also applied to the internal cases detected. That is what South Korea has…

That approach works for a few infected, that's called containment. But if the number of infected raise too much, that approach can't work. Then we need mitigation, measures to reduce the effect and consequences of an spreading outbreak. Ideas like "flatten the curve". Or, as South Korea did in March 12: lockdown.

But testing is anyway essential to measure the state of the outbreak. You can not know if you have fever without a thermometer.

Having an app that informs everyone of an infection 1km or 100 meters away would be really useful to everyone. Or better yet, what is the closest infection detected. The distance will grow as the outbreak goes away.

Isaac

21:57

The true measure of the outbreak is the "true infections". That's the real number of people infected. But that is not directly measurable. It has to be estimated.

When an outbreak starts some people get infected but there are no symptoms. In covid19 it seems that that period is from 5 to 10 days. And, in that period people are already infecting others. So, no symptoms and the outbreak is (rapidly) spreading.

As there is little indication that there is a problem, few people is tested. While the real number of infected people "true infections" is exponentially growing.

Then, for some reason, people start to be tested and the number of "detected cases" grows very quickly. Just because there is a lot of people infected.

And see that they grow from 100 to a peak of 3700 (per day) between Jan21 to Feb4. Despite the fact that a lockdown was set on Jan23 (just two days after some cases were detected) the number of daily detected cases kept growing. And, of course, the accumulated number of confirmed cases (what everyone is looking at) was also growing.

As real world example, one that is disapearing now, you can look at the yellow bars in this graph (from China) published in JAMA (The Journal of the American Medical Association) a very reliable source of information.

Source page

And see that they grow from 100 to a peak of 3700 (per day) between Jan21 to Feb4. Despite the fact that a lockdown was set on Jan23 (just two days after some cases were detected) the number of daily detected cases kept growing. And, of course, the accumulated number of confirmed cases (what everyone is looking at) was also growing.

But, when anyone was tested, they were asked (among other things) "Since when did you have symptoms?" that allows the yellow bars to be shifted left to when the individual really got infected. Those bars are the blue ones.

And, amazingly enough, the blue bars peaked just three days (Jan26) after the lockdown. It was very effective.

So, we look at everyday detected cases in the news, but the **real cases** could show a very different picture.
Would it be possible to draw the same type of graph for every country?
If so, we would know when the outbreak has reached its peak, and when we could start to get back to normal everywhere.

Please download and take a look at Figure 1 from jamanetwork.com/journals/jama/fullarticle/2762130

Tim

22:42

Make sense. I am now looking at new case number for how effectively the disease is under control, and active case number for how severely a place is infected

Does this chatroom less crowded recently have anything to do with the virus?

Transcript for