Comment by somethoughts | Hacker News Reader

Comment by somethoughts | Hacker News Reader

Parent Full thread

Don't get me wrong - when I first saw the Vox video (and others like it) - I can't say I disagreed with it. However what it didn't do is present any of the rebuttal issues. Its weakness is it treats all people arrested equally in good faith, when at least some of the repeat offenders might need some culling.

And at this point it's still just a highly correlated but not necessarily causation [1][2] - until more underlying data with more context about the crimes/criminals. But the move fast/break nature of the roll out makes the a/b test feedback cycle challenging. Perhaps it should have been tried in one or two counties first.

And don't get me wrong either - I'm more than happy if the move fast and breakers can stick around and perfect the new feature (i.e. bail reform) by adding some additional checks. It's a worthy goal/feature. Perhaps all that's needed is a check like - "if you are arrested a second/third time we are going to have cash bail".

[1] California Supreme Court Nixes Cash Bail Some Defendants - March 25, 2021

https://www.latimes.com/california/story/2021-03-25/californ...

[2] Original Post - "In October 2021 alone, the increase was 356% over compared to October 2020. Not only do these dramatic increases represent retail product thefts – they include increased assaults and armed robberies of UP employees performing their duties move. These individuals are generally caught and released back onto the streets in less than twenty-four hours... Even with all the arrests made, the no-cash bail policy and extended timeframe for suspects to appear in court is causing re-victimization to UP by these same criminals.

> at this point its still correlation only

There is no correlation I'm aware of. Two things happening around the same time isn't correlation. Furthermore, changes in crime rates across the country are similar regardless of the local DA policy (per the FBI's crime report), which would indicate a lack of correlation with bail policies.

> In October 2021 alone, the increase was 356% over compared to October 2020. Not only do these dramatic increases represent retail product thefts – they include increased assaults and armed robberies of UP employees performing their duties mov

Crime for one private business isn't data supporting a correlation with actions that affect the entire city.

To be pedantic - two things happening at the same time are correlated but what is not necessarily proven definitively is that its causation.

And yes - its still early innings and the single data point is end user collected from a "bug report" by a major power end user (i.e Union Pacific).

But at the end of the day - UP is a power enterprise level user (of California as a Service) who happens to provide a useful add on service to a large chunk of the CAaS end user community.

I'd say as a proactive PM/dev responsible (i.e. Newsome, CA District Attorney, etc.) for championing the new feature (i.e. bail reform), I'd be attentively listening to all users of the community about their feedback/UX experiences and very actively thinking about tweaks to improve site uptime even before definitive data existed that pointed to my feature causing outages. Or concretely ensuring/proving its not.

If I wanted to get that promotion that is.

> two things happening at the same time are correlated but what is not necessarily proven definitively is that its causation.

That part is incorrect: Correlation is defined as (A => B) AND (!A => !B). (I.e., that's a logical AND; both terms must be true.) Every morning, the rooster crows and the sun rises; that's not correlation because if the rooster doesn't crow, the sun still rises. (The actual definition of correlation is, IME, a very powerful tool for analysing through assertions.)

As you say, correlation does not mean causation, even using the proper definition. For example, two people completely unknown to each other may, over the course of an hour, laugh, cheer, and cry at the same time, in a highly correlated manner. But there is no causation between them; they are merely watching the same live TV show.

I don’t see how a year of arrests and no charges is even slightly defensible. We aren’t talking about bail reform here. We’re talking about are we going to do anything at all?

Reply on news.ycombinator.com