Yes, have you read my post or the original article?
The question was: what do we do when the only evidence (for child pornography, terrorist plot, money laundering/tax evasion, etc) is on encrypted systems? While this doesn't apply to the San Bernardino case (the FBI already has all the evidence they should need), there is and there will be increasingly more people using these systems to hide incriminating evidence. And the question is, what do we do when it's the only evidence against them?
Excellent question, although I'm assuming you mean not evidence of a crime but rather evidence of who has participated in that crime (ongoing threat, ticking timebomb, all those discredited rare scenarios that we are suppose to give up liberty to thwart). I'll play along. What does the FBI do in the following scenarios:
1. That evidence is in the perpetrator's mind. He knows what he did. If only we could get it out.
2. That evidence was written on a stack of papers, the entire plan laid out in precise detail, but then the papers were sent to the shredder, the shreds burnt to ash, and the ashes distributed from the top of Mt Denali.
3. That evidence of a wide conspiracy including hundreds of co-conspirators' identities and the level of their involvements, was definitely in someone's head and on their hard drives and printed out, but they physically destroyed their drives beyond our capabilities to recover (magnetic, physical shred, accelerated combustion, dismantle, disbursed), then shredded and burnt all the printouts, then shot themselves in the head so even our Fringe "read the last thoughts the person had if we get to them before they are dead-dead" super top secret machines can't recover it?
4. That evidence is inside a secure messaging app using a TOR routing mechanism which we had not infiltrated prior to its use, and saved in secure notes encrypted using an app supporting AES-256 encryption.
I would assert that in all four cases, the response is exactly the same. This is called an investigative dead end. It sucks. You make sure there is nothing more you can do, and keep it open in the back of your mind, but in the end there isn't a direct route to that data. Instead, you go around it and investigate the circumstances of the dead end (ex, if the guy didn't put a bullet in his own brain, perhaps someone else did that for him after assisting him with the cleanup; if the messages were sent between two people perhaps there is an artifact of the conversation or the envelope of the conversation on some server; if the data is encrypted perhaps there is a flaw in the encryption protocol).
I would also assert that there is absolutely nothing law enforcement can do to stop any of the above scenarios. The first three have been happening from time immemorial to varying degrees, and the last has been happening for hundreds of years, just getting more common in recent years with computers available to assist in both the encryption and in the brute-force decryption attempts.
Law enforcement has not ground to a halt because of the availability of turn-key AES encryption for the past decade, nor inrecoverable-without-the-key encryption for the past several centuries, nor of the availability of mechanisms to permanently hide secrets from LEO in the past, I don't know, ten millennia. There is no reason to expect that the availability of freely-available encryption being built into a device will trigger that apocalypse, because that is barely a half step removed from what any criminal can put together on their own from component part