• 7 Posts
  • 630 Comments
Joined 3 years ago
cake
Cake day: January 16th, 2024

help-circle







  • You’re really dismissing the reality that the biggest reason dems didn’t show was because the troll farms and their disinformation/amplofication worked.

    Or do you just not understand? Because that’s fine too. That just means it worked on you.

    Let me clarify. They didn’t just hype up the red guy. They also drowned out the blue gal. A lot of the “both sides” talk, all of the “but muh genocide” abstaining voters…some, sure, were real, but most of those that were, were pushed there by the bot farms.

    Propaganda isn’t just glorifying the leader. The other real important part is inhibiting the opposition.

    They all knew, and continue to know, that they didn’t have to get people to vote for Trump. That’s a fools errand. Most everybodies mind was made up about him well the elections.

    But keeping people from voting for Harris is just as effective. Because ultimately the winner has to be one of the two parties. We can hem and haw all we want about the need for third parties (and that’s totally true), but the fact is that it will always be a two party system until the voting system get fixed, and it will always be the same two parties until one inevitably implodes on itself and re-brands. Republicans are very much aware that the easiest way to win isn’t to attract voters, but to disuade people who vote against them.

    They gave up the honey a long time ago, and they’ve gone all-in on vinegar.






  • I don’t disagree.

    The genie is out of the bag with AI. I don’t think there’s any “going back” to a pre-AI world. It cannot get back into the tube.

    It’s a bubble, for sure, but we aren’t going back to pre-dotcom, either.

    In the meantime, though, it’s an arms race. Especially when it comes to things like offensive and defensive cyber security. LLM is especially well-suited for throwing shit against the wall and seeing what sticks, and that’s a perfect use-case for offensive cybersec.

    FOSS has its own set of issues to deal with with regards to LLM code…even beyond the vibe-coding aspects of it, the ultimate problem of licensing and attribution is incredibly complicated.

    That is more of a philosophical issue than anything else. Not to be dismissive if it, it’s quite important…I had a conversation recently with a major FOSS group’s sysadmins about their defenses from web scraping. They have outright blocked a major cloud proxy platform’s ASN because of abuses frok their IP blocks.

    (What follows is my interpretation of what he said, not his wording)

    A lot of that probably isn’t even “offense”, but poorly coded tools run by the platforms customers performing excessive web-scraping…but it exposes the bigger issue that’s more philosophical than anything, because from his perspective, this looks, smells, and has the same outcome, as a DDoS attack. And the only ways to really, truly mitigate such an attack is with CDNs and such, and usually with SSL Decryption on a cloud provider.

    These types of methods are diametrically opposed to something like their mission, putting them into a position where they have to outright block a lot of legitimate use to stop a handful of bad actors who are probably more ignorant than maligned.

    I kinda see AI as the same. Shaming it, in FOSS projects particularly, isn’t a matter of code quality…it’s a philosophical matter, and the problem-users are usually more ignorant than maligned.