On: Online databases dropping like flies, with 10,000 ...

[pjc50]

Hypotheses:

- A/B testing of ransom prices will happen

- the person paying the ransom is not the company but an employee afraid for their job

- companies more likely to be able to restore from backup

[saycheese → pjc50]

It would be really hard to run an A/B test of any meaning since the value asked and the data would both vary; to do a true A/B test there may only be one variable, the populous must be very similar, etc.

[matt4077 → saycheese]

The "data" (the unknown variables) always varies. That's why you need n >> 1 in A/B tests. And you are only testing one variable: the asking price.

[saycheese → matt4077]

Thanks, though what does "n >> 1" mean?

(Ask since to me it reads as gibberish and there's no way to Google it myself.)

[thmorton → saycheese]

Usually this notation ("x >> y") means "x much larger than y," so you need way more than 1 person to A/B test.

[saycheese → thmorton]

Thanks. Normally need more few thousand to do a valid A/B test; meaning to me the comment makes no sense.

[Karunamon → saycheese]

You need more than one person to A/B test against, so n in this point is sample size.

[abailin → saycheese]

"much greater than"

[brianwawok → matt4077]

Correct. Running a e-commerce store each buyer buys different items. You need a big n.

[saycheese → brianwawok]

What is a "big n" mean?

(Ask since I'm used to running valid A/B test on groups in the 1000s.)

[PretzelFisch → pjc50]

Once you agree to pay, they may come back with a higher offer. No reason to A/B until they have a live one on the line.