The same can be said of cryptography. Why believe that cryptography is more likely to be used for lawful and social beneficial purposes than for, e.g., terrorism, tax evasion, child abuse, etc.
In fact, I wouldn't be surprised if the latter were in actuality the first adopters of cryptography, especially if you include other illegal uses such as copyright violation.
We could easily say that free market dynamics suggest that the cost+benefit will higher for criminals, especially unless and until cryptography becomes ubiquitous. So why not nip it in the bud? That's been the argument coming from the law enforcement and intelligence communities all along.
Yet engineers are defensive and optimistic about the beneficial usages of cryptography. They're dismissive of usages such as the encrypted iPhones used by the San Bernardino terrorists. And they focus on seemingly esoteric benefits--like improved security from extortionists--that most common people overlook or discount.
I tend to _agree_ with those engineers, am similarly defensive and optimistic regarding cryptography, and place a high value on benefits not obvious to others. I think it's worthwhile to be consistent here. Any distinctions one can concoct to differentiate Facebook and their advertisers from, e.g., non-profits, can easily be turned around later.
The emphasis shouldn't be on _presumed_ abuses, guilt by association, etc. It should be on hard data of actual usage and actual outcomes, with attention paid to implicit biases in collection and reporting. We also need to compare this against alternatives, taking into account (and quantifying) the futility of criminalization or regulation, etc.