Thread by @vmyths

Let's talk cybersecurity.

1/6 Florida's official #COVID19 website has a problem: migrants who picked watermelon weeks ago racked up high numbers in central FL counties but have since returned to southern FL counties...

https://twitter.com/hshaban/status/1263332028516306944

2/6 Migrants sent home w/ #COVID19 remain counted in central FL; their data doesn't "transfer" to another county.

And contrary to any PR statements, those migrants who later need hospitalization end up counted as having #COVID19 in *two* counties.

Let's be CLEAR, though:

3/6 Double-counting a case of #COVID19 only occurs when a patient (a) gets tested in one county and (b) seeks hospitalization in another county.

(Remember, we're talking about #cybersecurity!)

For example, Levy County FL cases look like this when it should be ~120 fewer:

4/6 So we can see Florida's #COVID19 data is flawed at the county level.

BUT...

It's not flawed *that* much, right?

And we *need* data to make informed decisions, right?

Hence we #rationalize "in this situation we've no choice but to accept this flawed data as authoritative"

5/6 Note that I neither defend nor dismiss Florida's official #COVID19 website. I'm just a cybersecurity #critic trying to make a point.

And my point is this:

We have NO authoritative data to support global #computer virus damage estimates from 1988 to at least 2016.

Now...

6/6 ...we see the #medical community doing what WE perfected over the last 30+ years:

✔ #PewPew maps
✔ Flawed data
✔ Projectionism
... 🔸️ Anecdotes
... 🔸️ Talking points
✔ Cubic model (!) predictions

In #cybersecurity it's "a necessary evil." In the medical world...?