Because innocent people getting swept up by the legal system is potentially disastrous for those people, particularly if they are people of color. Take the Kalief Browder case as an example: http://www.newyorker.com/magazine/2014/10/06/before-the-law

Yes, this is one case.

However, injustices happen all over the world in varying degrees. It's important to figure out the balance of damage done / good done. If one out of 100 million people is wrongly convicted each year due to these practices, but 1 out of 50 million is rightly convicted each year, then this tool net good (even though innocent people have convicted.) This is just a simple view though, it's up to society to balance this equation (but that means not having a puritist and irrationally stubborn point).

What is the prevalence of accidental conviction due to the police having access to your limited private information?

Apart from gigantig societal issues around chilling effects and free speech related issues. Whenever you cast a wide net you must either do significant statistical modelling or disqualify the actual results of the warrant from being considered as evidence. Ie you can not both use the results to filter the suspects _and_ use it as evidence without knowing exactly what you are doing from a statistical point of view. As badly understood statistics generally are, the required level of rigor is unlikely to be achieved.

I understand where you're coming from.

But I believe this is just like where LE collects a list of suspects (from people the victim knew) and narrows down from there.

The net isn't as wide, nor is it used as the main evidence for conviction (atleast I would hope so).

For real?

It's not about this case, it's about the precedent.

The fight is about privacy...at all...online. Google needs to fight this for everyone because if they lose we all lose.

It sets an interesting precedent for more broad fishing expeditions. Slippery slope and that sort of thing. Have you watched the film Minority Report?

Have you? The concept was "pre-crime" which is stopping crimes before they happened - not investigating crimes that have already conclusively been committed. Additionally, upon finding the system in use was faulty it was dismantled and everyone affected by it given a pardon.

Not saying the slippery slope argument is invalid, but the use of Minority Report is better kept to arguments where the intent is to use "big data" to predict and prevent crimes - not for when crimes have been committed and the intent is to figure out who did it.

I mentioned it because it is the bottom of the slippery slope.

Once search data is open game for any law enforcement purpose, why not preemptively tag searches that are common before a crime takes place?

That said, I'm a bit confused as to how a "happy ending" would invalidate moral concerns raised by an illustrative story.

Also FWIW, leading with "Have you?" reduces the chance I might read what you had to say with an open mind.

No, but I get the general premise.

I don't see how it directly meshes into the issue at hand though.

Sweeping through large swaths of data on what people do in order to single out one potential person in the data.

Once you're good with that, it leads places.