Just a scenario: Say you want to galvanize an entire population into submitting to autocratic rule. It could be done like so -
1) Grab the attention of the public with a salacious controversy. Bolster widespread memes and juvenile conversation regarding that controversy on social media, engaging a large share of the population.
2) Perform some action which causes street protests (fire a well known investigator, etc.) Use social media to brigade violence on the protestors. "On the street" protesting dwindles as people would rather be safe, causing protest and attention to move online.
4) The online realm would now have high attention, split into a contrived binary contest: protest-related content versus salacious controversy-related content. At this point it is easy to start to drown out protest content by brigading salacious memes and cracking normalizing jokes.
5) As the population watches this happening, start hacking power outages in cities which have the most protest-related content, causing riots and chaos in those regions. Seeing this, hold-out regions will convert for the sake of their own safety. As hold-outs are marginalized, their scale becomes more manageable and it becomes possible to direct energy toward individuals.
6) People begin to understand that the online vote is basically a demand for loyalty at the threat of violence. Videos of Hold-out individuals being attacked begin to surface.
... without going further down a Black Mirror-esque rabbit hole, I just mean to point out the worst-case type danger that the apparatus of social media and internet tracking can present. The scale of social control that could be on the horizon should be quite concerning.
Of course attempts to say public opinion have been going on for thousands of years but what is new is the scale and ease at which this can be done. It was much harder back when there were hundreds of independent newspapers and when information exchange was much more costly both in terms of time and in expense. Now that vast amounts of information about hundreds of millions of people have been collected it's much easier to influence opinion.
What implications does this have for free speech? For instance, should algorithmically generated text garner the same free speech protections as I do? Should a couple of companies be in charge of the vast amount of data collected about us?
I'm reminded of an Isaac Asimov story about the future where only one person votes for President. The person has been chosen by a computer as being representative of the general population and he/she has the sole vote.