If you create enough data smog the Big Data systems will just give wrong answers. Whether that's a good or bad thing depends on how the data's used.

It seems to be a common misconception that the Googles and Facebooks of the world have someone nefariously looking at your data to profile your every move. No. The average employee at Facebook does not care about you, and even if they have access to your personal data (I know that this is significantly locked down at Google, and probably is at Facebook these days), they don't care who you are or what you do. Instead, they're developing algorithms that mine lots of data from lots of consumers to produce products for lots of consumers. If you put in wrong information, it's basically GIGO. And since the basic models are trained on millions of data points, if just you do it, you only hurt yourself - you're a drop in the bucket compared to everyone's data, but your own personalized interface will be based on the profile you give it, not what you actually want.

You can see this in action with shared Netflix accounts. Share a Netflix account with someone and basically you get garbage for recommendations, because the system tries to make sense of two different "streams" of data that were arbitrarily merged together. (There are ways to separate them out, but they add a lot of complexity to the system.)

Things may be different if you're talking about Palantir or the NSA. They specifically do look at individuals. But then, if they get a bunch of random data their assumption is probably going to be that the person is deliberately trying to obscure their tracks ("tradecraft"), which means that the suspect warrants even further scrutiny.

What happens during wartime when civil liberties are curtailed? I suspect each company will have to follow their law of the land which might involve regrettable use of data. In other words I dont have a lot of trouble believing that the Googles and the Facebooks of the world will be good stewards of data in normal circumstances (Its in their self interest after all).

The other issue worth thinking about is that we are in the midst of the golden age of building networks so the economic models are all viable. What happens 10-20 years from now when the roving eye of economic development has moved onto other sectors leaving the economics of ad supported saas products non viable. I fully trust Google and Facebook, but if they were to ever be liquidated / sold in pieces I dont know if my data will end up in safe hands.

This is a legit concern - there's a long history of organizations doing unscrupulous things with their data & IP when their actual business goes south. (SCO is perhaps the best example.) However, these are usually more self-limiting nuisances than long-term problems. If the business is going south, then it means the organization's actual power is declining, and this limits the damage they can do with their data. They're quite capable of making a lot of noise and annoying a lot of people, but they'd then lack the power to deal with the inevitable blowback.

What would happen if Google goes under in 30 years and sells all the data it's accumulated on all Americans? Let's say they sell it to a worst-case buyer, someone who explicitly uses it for blackmail & extortion. If they do this on a small scale, then by definition it's on a small scale, they can't make any money off it, and they get shut down legally when they pick on the wrong journalist or wealthy individual. (They're in a catch-22 when picking targets: if they pick poor, powerless people, there's no money they can extract from them, while if they pick fat wealthy targets they'll probably get fought hard legally.) If they do it on a large-scale, the public outcry will get laws passed faster than you can say "re-election", and they'll get shut down by Congress/lawsuits. Either one is a no-win for the buyer.

A totalitarian government is more worrying, but the government already has all your personal data anyway.

If they're looking for you, specifically, then yes. It is possible to have the smoke trail have an erroneous local maximum, but it's unlikely that you can pull this off.

If they're just looking for patterns of which you are just a statistic, then it'll merely be hit-or-miss; you're unlikely to generate enough "smog" to make it so they never actually tag you correctly.