Hacker News Reader

Comment by lolinder | Hacker News Reader

Privacy laws don't solve the real problem, they would only solve the fig leaf that politicians are hiding behind when it comes to tiktok.

The actual problem is and always has been control over the content being fed to users. It's not an issue of privacy, it's an issue of voter manipulation. It's just that the US has decided that it's okay with its own plutocrats manipulating voters while it's not okay with the CCP doing so.

On the one hand that's a very rational position for people who owe their election to algorithmic voter manipulation to take, but that doesn't really make it better ethically.

View on HN

The solution for the other half of the problem is anti-trust divestment of client apps from hosted services. Let TikTok (and Faceboot, and so on) keep their own assortments of services. But the mobile and web apps should be spun out into different companies, only communicating with openly documented APIs that are available for every other developer/user.

This won't solve the issue with propaganda that still manages to be compelling in the court of public opinion, but it will at least level the playing field rather than having such topics inescapably amplified for "engagement" and whatnot. There's definitely a mechanic of people realizing specific social media apps make them feel bad, but as of right now it is extremely hard to switch to an alternative due to the anticompetitive bundling of client presentation software (including "the algorithm") with hosted services (intrinsic Metcalfe's law attractors).

This doesn't make any technical sense. The key part of what is at issue with TikTok and other social media companies is editorial control over the ranking algorithm. The ranking algorithm is not done on the client, and I don't see how it can be.

And even if you could do it somehow, that is in direct conflict with the other main complaint people have about social media apps ("selling your data"). FB had an API that they opened up to some degree, then Cambridge Analytica happened.

Discovery algorithms are part of client presentation and should be under control of the client, regardless of beign commonly done where the database lives.

From the technical perspective, for things where your view is limited to friends that is straightforward to do on the client. Global discovery using metadata can be done using deterministic searching which condenses the view of the database enough that the client can then rank those results however it'd like. And global discovery from a pure firehose is likely even within the reach of a modern personal computer.

As far as apps taking a copy of data, the first obvious control is that its only apps you choose which have access to your data. And for the general problem, that is the point of privacy legislation - which was taken as a given by the comment I was responding to.

There's a good argument for separating content from discovery. Originally, Google was pure discovery - all you got was the page's link and maybe a line of description. If antitrust law was used to separate discovery from content, transmission, and aggregation, Google would be forced back into their original niche. They'd be a much smaller company, of course.

The voting algorithm needs to change so that destructive (negative) campaigning is not so effective.

Duverger's law makes campaigns devolve into undermining and destroying the competition, with the two parties hosting primaries to see which of them can "turn the wheel" the hardest before the general election where they claim "don't worry I won't crash the car!" despite their prior incentives.

If we used plurality voting for the inputs to a decision problem that follows the classic tragedy of the commons, we'd see a similar result. If instead of just {+1, +0, +0, ...} without repeats, we instead voted with {+1, +0.5, -0.5, 0, 0, ...} cooperation (or at least constructive competitive frameworks) would at least be at parity with destructive and potentially mutually destructive competition.

>The voting algorithm needs to change so that destructive (negative) campaigning is not so effective.

No algorithm is going to fix this, it's human nature. Negative campaigning has been a constant of elections since elections were invented. At best you can tamp down on the aggressive engagement feedback loops. We should probably do that, but it's good to stay realistic about outcomes.

error_logic was referring to the "Voting system" meaning the way we elect politicians: plurality voting vs. ranked choice/approval voting. Not an algorithm in the social media feed sense.

Other voting systems do promote less negativity, among many other benefits such as giving us real choice and a true multiparty system.

we call that a voting _method_ not system.

The difference is that if your party A spends x amount of resources on negative campaigning on another party B instead of on yourself and B is only one of 4 competitors (B,C,D,E) the other 3 will stomp you.

> It's just that the US has decided that it's okay with its own plutocrats manipulating voters while it's not okay with the CCP doing so.

I think it's more likely that it's able to pass muster when the threat is a foreign state. The US may be more limited in what it can do with it's own entities (I don't know for sure) and would probably receive far more push back from the courts if they tried heavy handed measures before going through the proper legislation regulation (as states are doing) or the FBI route(if it were in their realm of bad). The threat that is more concerning than US owned and operated companies getting US citizens to vote for their prefered US president, is a foreign state slowly radicalizing US citizens(without their knowledge it's happening) against themselves (the US), more than voter manipulation.

Reply on news.ycombinator.com