And they're supposed to get prior judicial approval. It's only acceptable to file retroactively in cases where there is an imminent risk of death or great bodily harm, such as a hostage or barricade situation or other 'extreme emergency situation.'

You're assuming the worst possible outcome (police overreach, court apathy) and using that as a justification to omit any mention of the copious rules limiting official use of this power. I find it rather ironic that you decided to leave your readers poorly informed about this, not least because it would put them at a disadvantage in any argument with a supporter of the bill.

The problem is that "imminent risk" tends to creep downward over time, to the point that it becomes "imminent risk that a citizen unit will improperly fail to bow and scrape."

If you're going to shut off several thousand cell phones, you can bloody well get a judge out of bed for that.

Perhaps, but I don't think it's fair to judge this bill on how legislators might amend it in future, or we'll never pass any laws at all.

Laws have unintended consequences. It's good to slow down the creation of laws and weigh tradeoffs. Laws are intended to be solutions. But the cure may prove worse than the disease.

Can we judge it on how the police interpret these things now?

For example, the police are only supposed to shoot people when there's imminent risk of death or great bodily harm. Yet there is routine, if not common, overreach there, including one notorious recent incident that people are currently protesting with great vigor.

If you can't reliably enforce the rule that police aren't supposed to kill people except in dire emergencies, how will you do it for cell phone kill switches?

The only safe way to analyse the risk is to assume the worst case scenario. The Ferguson police are the worst case scenario: to quell a protest they deploy violence, then complain when the protest becomes violent.

As a result of this, the police will feel justified claiming "imminent risk of death or great bodily harm" any time two or more #insert <target_demographic.h> are together in one place.

Then you just do this often enough that the judges don't have time to review every case.

Don't dismiss as ridiculous what is simply an extrapolation from current practise.

That may be true, but it should still be mentioned in the article in the interests of a fully informed electorate.