Hacker News Reader

Comment by oconnor663 | Hacker News Reader

> For me was 50/50 or even 60/40 against innocence.

Totally, but the standard for conviction is "beyond a reasonable doubt". By that standard, a 40% chance of innocence means "not guilty". I don't know whether the law has ever translated "reasonable doubt" into a specific percentage, but I have to imagine it's less than 10%?

View on HN

Alexander Volokh wrote a tongue-in-cheek law review article titled "N Guilty Men" that tried to quantify that percentage:

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=11412

It's obviously subjective, but a YouGov poll in the UK found that 47% of pelt required 99% accuracy or higher to be considered beyond reasonable doubt. Only 12% set the bar below 90%.

Source: https://yougov.co.uk/topics/politics/articles-reports/2019/1...

20% said they would require 100% certainty, which is equivalent to stating that not even the most obviously guilty person in existence deserves to be convicted. I'd posit that this is more so evidence that the general population's knowledge of probability is abysmal.

Or their concept of “certainty” or “doubt” is different from yours.

I'm not even sure what that is really supposed to mean. If 100% precludes the potential of an event with infinitesimal likelihood (mass delusion, possession by ghosts, etc.), then as I said nobody ever gets convicted.

Otherwise, any "concept of certainty" that uses "100%" to refer to something that might not occur, is either gravely misusing mathematical notation or self-contradictory.

While technically correct this is obviously not how people speak. If you ask someone 'what is the percent chance that drinking this glass of water I just got out of the tap in New York City will kill me within a few seconds?' They would respond '0%', even though by a mathematical definition that is not true.

People generally consider context for their choice of words. Alongside 100% they were also given options such as 99.9999999%, it's clear that 100% in this case is not just used as "very likely".

I didn't know they were given specific options.

Can you statistically evaluate such a thing? Its all black swan.

If you look at it in a bayesian way this is not an issue. "I can't put a number to it" would definitely be a legitimate answer, but "100%" is just wrong.

I don’t disagree. I guess it’s kind of like an additional psychological bias that one plays into when he’s already been convicted and in prison.

Reply on news.ycombinator.com