Hacker News Reader

Comment by codecamper | Hacker News Reader

Maybe technology could help this situation?

It sounds like a big part of the problem are too many arrests for things without enough evidence.

Wouldn't everyone benefit for better logs & statistical analysis? If a certain police officer had a 15% conviction rate, and he actually liked his job, then he'd be sure next time to be more sure about his arrests. Such a tool could track recent trends as well.

Also, it sounds like there needs to be a system (email??) to contact employers. All employers required to have an official address and courts are required to contact it with current case status.

Anyone else have ideas about how tech could help the situation?

View on HN

> If a certain police officer had a 15% conviction rate, and he actually liked his job, then he'd be sure next time to be more sure about his arrests.

...or get better at pretending to be more sure so that the innocent defendants he arrests don't screw up his numbers by getting acquitted.

You get what you measure. If you measure the conviction rate then the conviction rate will go up, but that isn't the same thing as the percentage of convicted defendants who were actually guilty.

What about on the other end: an automated system to assign bail amounts? Plug in charges + criminal history + financial status into a system that determines bail. That way it's in the hands of a computer and there is no blowback to a particular judge/prosecutor/etc.

Then any judgment is removed from the system. Judges can abuse that power, and they can have bad judgment, but an algorithm has no judgment at all. Worse, you can get people turning off their mind as to whether it made any sense, because "that's what the computer said".

People are in the system for some good reasons...

Hmm. Actually, the algorithm could be very transparent. It would be there to consistently apply the rules. If the rules were too harsh, and not effective in some way, they can be tuned.

The algorithm could be as simple as:

first time offender, non violent case, 0 bond second time offender, non violent case, 100 bond etc.

Another idea that may benefit all would be to force people who have gone through the system to create a bond "deposit account". Like a health savings account. This amount could be collected slowly over time -- forced deductions from pay.

People without jobs or money -- must be released without bond then, in the case of non-violent crimes.

That sounds like a great idea. YC, are you listening? :)

If such a system could be fed with information like.. did the defendant run, then it could also figure out what bond levels work.

Another part of this article mentions that below a certain level, bail bonds won't handle it. Is that because a human is required to make a decision?

I don't think arrest to conviction conversion rate is very meaningful by itself since in many jurisdictions the prosecutors get to choose which arrested persons to prosecute and how vigorously.

hmm, the arrests are the input. The prosecutors would be a second filter step. And that step would be more or less constant.

There are several steps towards conviction and many variables, but I think there are statistical methods that can tease out the conversion rates at each step along the way.

Reply on news.ycombinator.com