Depends on what you mean by decentralized. If there is a giant de-facto operator of spam filtering that’s on by default, it may be akin to centralized in practice.
That said, I am very bullish. I don’t think every node needs to be equal, so to say. Instead, you need good incentives.
For instance, if the spam list starts mission creeping and flagging political speech, for instance, it’s important that someone else can start up a competing spam list. Open source (transparency), and the ability to choose lists (consumer agency) is what keeps power in check.
Also, I really like the idea of keeping lists separate by type (eg crypto spam, 4chan type brigading, and perhaps even more niche things like shallow and edgy takes). It’s the one-size-fits-all that’s so infuriating with legacy social media. It leads to a washed out common denominator speech.
Lastly, I think the challenge is quite technical. As mentioned, creating a Nostr node costs nothing, but distributing spam lists cost bandwidth and space. So compressing and encoding these rules is critical for these lists to have a chance against Sybil attacks..
(Sorry, this comment got more fragmented than I intended)
> If there is a giant de-facto operator of spam filtering that’s on by default, it may be akin to centralized in practice.
The article shows that you can disable Primal's spam filter. And I think they are saying that in the future, you can use other spam filters or none at all and just stick with your friends' mute lists.
It shows what is possible for content moderation in decentralized networks