Read on Twitter
I’m not a virologist/epidemiologist. But I do #AI and robotics, so I study how to make decisions under uncertainty and how statistics and probability shape decision making.
So every time I hear “only 1% risk” to mean something is safe, I cringe.
#COVID19 #coronavirus
So every time I hear “only 1% risk” to mean something is safe, I cringe.
#COVID19 #coronavirus
I’m not talking about how we end up with a particular number for risk, what models were used etc., because that’s not what worries me.
Let’s just assume there is a 1% risk when we go out (I picked a round number, this works the same with any other number.)
Let’s just assume there is a 1% risk when we go out (I picked a round number, this works the same with any other number.)
It's easy to flip that 1% to mean the odds are 99% that you’re safe, but that ignores the repeated risk.
Let’s forget about viruses for a second and assume there’s a five-foot drop (~1.5 m) in front of you, and there’s a 1% risk you’ll break your ankle if you jump.
Let’s forget about viruses for a second and assume there’s a five-foot drop (~1.5 m) in front of you, and there’s a 1% risk you’ll break your ankle if you jump.
Should you jump?
Well it depends.
If there’s an angry hippo chasing you, I’d say go for it. After all, the odds you’ll be fine are 99% on that ONE jump.
Well it depends.
If there’s an angry hippo chasing you, I’d say go for it. After all, the odds you’ll be fine are 99% on that ONE jump.
What if that drop is on the way to work or school? And you ask whether you should jump it every day to save time?
That’s a different question because repeatedly taking that 1% risk changes things.
By how much?
That’s a different question because repeatedly taking that 1% risk changes things.
By how much?
If you jump it for a month, your odds of NOT breaking your ankle are .99x.99x.99 …. thirty times. That says you have a 26% chance of breaking your ankle that month.
1 in 100 became 1 in 4!
1 in 100 became 1 in 4!
If you jump it for a year, your odds of NOT breaking your ankle become 2.5%.
Yes, that’s right, your odds of breaking your ankle are 97.5%
Still think it's low risk?
Yes, that’s right, your odds of breaking your ankle are 97.5%
Still think it's low risk?
Wait, you say, 1% is too high. What if the risk of breaking an ankle was 1 in 1000. That’s safe right? If you jump every day for a year, the risk of breaking an ankle is …
Over 30%
1 in 3. From 1 in 1000.
(that’s 1 minus .999 x .999 x .999 … 365 times or 1 - 0.999^365)
Over 30%
1 in 3. From 1 in 1000.
(that’s 1 minus .999 x .999 x .999 … 365 times or 1 - 0.999^365)
How did that happen?
Well, you heard about how compounding interest is good and how small amounts of money grow over time, right?
Well, here, you’re compounding risk, so a little risk grows and grows fast.
Well, you heard about how compounding interest is good and how small amounts of money grow over time, right?
Well, here, you’re compounding risk, so a little risk grows and grows fast.
So, do I have a point?
I have two actually:
1. Don’t think of COVID as a hippo you can dodge with one jump. Think of it as having to jump over ruins every day.
So don’t let your mind convince you 1 in 1000 is safe because in the end what gets you is the repeated risk.
I have two actually:
1. Don’t think of COVID as a hippo you can dodge with one jump. Think of it as having to jump over ruins every day.
So don’t let your mind convince you 1 in 1000 is safe because in the end what gets you is the repeated risk.
2. Both 1 in 100 and 1 in 1000 sound low risk so it’s easy to get complacent.
But when you compound those risks for a year 1% becomes 97.5% and 0.1% becomes 30%.
So please reduce your starting risk as much as you can, say by wearing a face covering when you go out.
But when you compound those risks for a year 1% becomes 97.5% and 0.1% becomes 30%.
So please reduce your starting risk as much as you can, say by wearing a face covering when you go out.
