Currently moderation in the sense of modern communication is moderated, where a few apply a filter to the channels of interactions they're in charge of moderati

Currently moderation in the sense of modern communication is moderated, where a few apply a filter to the channels of interactions they're in charge of moderating. The personal agency model eschews this in favor of per-node subjective filtering, so the workload is distributed across nodes that receive the message. To parse or not to parse. As I have observed, initially, the personal agency model suffices. Channels are small, low amount of nodes communicating over it, what messages do exist likely belong there. Over time the channel scales, more nodes are added, more hysteresis follow, and things move to a moderated model where a filter is applied to limit the flow through the channel seeking better ratio of signal to noise. Where this model fails is at scale, as the channel scales, the capacity of the moderator fails, and more layers of moderation are added. Eventually the overhead of moderation consumes a larger part of the resources dedicated to the channel cumulatively on both sides, signal and filter. The nature of my assumption and this question is that, perhaps as scale consumes our channel, it's best to adopt a least-resistance flow of signal, unimpeded, and a return to the personal agency model where the filter exists locally rather than delegated to a centralized discriminator. Interested in what you all have to say about this.