Doesn't the sheer amount of noise (in the intelligence agencies' eyes) produced by Glass make it a fairly useless data source? If I worked for the secret shadowy three initial organizations I rather have access to call logs and email. At least it would have some curation.
I think that's the key point here. The sheer amount of data being stored is surely almost completely useless when it comes to preventing crime, but would be invaluable when it comes time to investigate a crime, or if you want to target a specific individual to harass.
I'm sure there is some useful information they could glean, but in large part key indicators of pre-crime are already blatantly obvious or present too great a risk of false positives.
The latter is what worries me. If they come up with what they believe to be the right algorithm and run it against their data, what then? Are we then in the position where we have to a) prove our innocence in a crime that hasn't even been committed, or b) become the targets of increased scrutiny and restrictions?
There's more that can be done than pre-crime, though. For example, statistical analysis/datamining of things like ambient sound levels, vocal stress levels, positive/negative sentiment, etc. could reveal trends even before they reach Twitter, allowing them to be preempted or directed.
