Sure. I’m thinking of monetizing the effort. I think there’s more value in curation, filtering, and selective processing that there is in raw data storage. The ‘global’ might become unwieldy and providers will drift into creating services tailored to a segment of users, countering the centralization. Smaller players can be competitive here if they have a creative way to surface a coherent subset of content.
Discussion
I think an "as global as possible" state will always lend itself to attractive monetisation, despite the costs, and as long as the network is alive and of interest to enough people.
There are 16k trusted pubkeys writing events daily according to nostr.band stats. Odell the other day said there are 10k Primal users active daily, also trusted in some way I guess. Assuming most of those Primal users don't jump between clients (low-key Twitter style usage) and taking into account that Primal is one of the few clients that has a monetisation model (in this case subscription), I'd estimate that of the total amount of actual business revenue being generated on Nostr today, Primal accounts for around 80% of it (The rest being split between Damus, nostr.wine, relay.tools and a bucket of others.) Total back of the napkin calculation there, but whatever the number is in reality it'll not be a modest one.
Yes. But I see centralization as a first step only and not the endgame. Aggregation is relatively simple to implement, as you said, especially of structured data. And it’s monetizable as is. I’m looking forward to the next part, the tailoring, which could swing a lot into decentralization. Going back to the start of this thread, global might not be as desirable down the line.
Ah, I see your point. I'm a little different, starting to think that there will always be three levels of quasi-global on Nostr.
1) Global firehoses: real-time event streams with limited historical access, all posts from all keys with some baseline level of trust -- most expensive but still always a market for.
2) Global indexes such as Nostr.band: without the realtime event stream but good for full-active search and other Google-type use cases, all posts with certain trust level archived since dawn of Nostr
3) Focussed indexes: parts or subsets of the network, the curation, filtering, and selective processing stuff you mentioned
For (1) Primal will always maintain their own, they're invested pretty deep in that model now. Other clients may copy Primal. A third-party service may emerge and make their firehose available to clients without their own at cost, these other clients being motivated to pay for it to compete with Primal UX (the exception being the rely-first clients that compete on offering a different experience without certain frills but with other neat things that only a relay-first approach can unlock).
That’s a nice summary.
I’m thinking (3) might take over an increasing chunk of the market and be more interesting 🙂