been testing it, testing numerous clients, most web. Loved the vibe, loved the connection with pairs, loved the "there is no algo BS", for the most part its pure non toxic environment. My concern though, will it scale? You can see how slow likes, followers, zaps get updated, I subscribed to some relays, some of them rugleys.. If this tech scales, how will it bridge the cloud gap with dedicated infrastructure providing storage, bandwidth at sub 20ms access to content? even more challenging, how this can be resolved for global south users?
pull down to refresh
1178 sats \ 1 reply \ @nerd2ninja 22 Feb 2023
There's been discussion of a sort of relay pool. Kinda like a mining pool. Relay operators can be part of the pool. On a technical level, all relay operators are behind a centralized load balancer for a single domain name.
Its best to think of nostr as a reimplementation of RSS. Social media over RSS with cryptographically singed messages.
Put another way, its like seeing facebook, reddit, stacker news, twitter and more all in one home page. Each of these websites have different moderation policies, but adding another site to your feed, is easier than logging into and checking another website. So if your favorite content creator says "sorry guys we got banned from twitter, please follow us on minds" or something like that, instead of signing up to minds and remembering to log into it now and then, just add the feed that comes from minds, to your main feed and it'll get mixed in with your feeds from the other websites.
reply
15 sats \ 0 replies \ @kst_winterhodler 22 Feb 2023
This concept makes a lot of sense. Worth exploring!
reply
15 sats \ 2 replies \ @mallardshead 23 Feb 2023
I don't know. Nostr I think will become something different than Twitter et al. Maybe something we don't fully expect. I see a nascent marketplace forming there already. This could get wild if you bring something like Slashtags or other decentralized reputation. To me the future of desirable and entertaining content requires high performance [non-relational] databasing to do all the cool stuff, like snippets of 8k video, complex animations, gaming in app, api's, Ai, etc. Twitter is an entertainment platform as much as it is anything else, and they use a high performance database architecture by Redis, very badass, and Nostr can't compete with that. They would need some serious innovation. Like mining pools except relay pools where they balance, process data in parrallel, and shard. Think MongoDB. And instead of blockspace you might have fixed advertising space. This funds relays. Maybe pays content creators. I don't know, but the innovation would have to be nearly on par with bitcoin. At this point are we just dancing around centralization? Nostr's problem isn't data storage. Its many clients' problem isn't UI/UX. It's database performance. Wen?
reply
60 sats \ 1 reply \ @DeltaClimbs 23 Feb 2023
Do you have any thoughts on how to have Sybil resistant reputation?
reply
10 sats \ 0 replies \ @mallardshead 23 Feb 2023
Holepunch built on Hypercore's append-only architecture possibly. Look into Slashtags by Synonym
reply
10 sats \ 1 reply \ @sebastix 23 Feb 2023
This is a related work in progress NIP, number 67: https://github.com/nostr-protocol/nips/pull/245
reply
0 sats \ 0 replies \ @fedelang OP 23 Feb 2023
I see this can help, but won't solve the issue. A massive scale for social media content distribution requires hyperscaling capabilities, meaning you can make it as efficient as possible, but you need the actual hardware, servers on top of servers + storage spreadout arround the globe. IMO this is not so much of a SW problem to solve but a HW problem in three dimensions, networking, processing and storing at a massive hyperscale.
reply
10 sats \ 0 replies \ @l0k18 23 Feb 2023
Nostr needs some work on consensus to enable this sort of thing. Likely there will be aggregrators at minimum, but I can easily see how there will be likely several different ones all at once.
reply
10 sats \ 1 reply \ @kristapsk 23 Feb 2023
This is overkill, 200 ms should be enough.
reply
0 sats \ 0 replies \ @fedelang OP 23 Feb 2023
Lol, I wanted to write 1ms since we are slowly trending to 5G-gigabit society plus submilli connections to enable VR and RT applications but I thought aiming for gaming ms was ok, guess not =) I see what you say it could be an overkill but here we are stacking ms a lot.
reply
10 sats \ 1 reply \ @TheBTCManual 23 Feb 2023
I don't think it kills it but it will have to change, if you're just interested in niche topics, lets say you're into coin collecting and thats your community, the current model is fine but if you're talking about getting into politics or news where people would all congregate and dogpile in their bad takes and opinions, then the client, will need to offer an aggregator of some kind sourcing from relays and trying to be a bridge.
Not sure how far it can go or how efficient it becomes but lets see
You make a good point on bandwidth which would be a problem, in my global south countries your telecoms providers actually have direct deals with the likes of Facebook and Twitter to route traffic to these sites at special rates or even for free subsidising the traffic to increase DAU, new users and ofcourse ad revenue, doubt nostr can compete with that
reply
10 sats \ 0 replies \ @fedelang OP 23 Feb 2023
Not only in the global south. Competition for smooth and fast user experience is such that most content providers at global scale own even the submarine fiber than runs from continent to continent, relay stations, landing stations, multi megawatt datacenters for in-country or regional distribution of content and the edge layer at the ISP level where based on you IP range they ship free the HW to proxy content and have it fast deployed to users at very low latency.
ISPs give free landlord services (they have no alternative) to this large content providers: rackspace, power, cooling and basic hw operation in exchange of barely cuting their carrier bandwidth costs.
IMHO, ISPs are key to develop Nostr at a higher scale, and instead of leaving them with crumbles on the table they should be running HW for Nostr the way that miners process the BTC transactions. I don't see any other model succeeding that won't include a seat in the table for ISPs.
ISPs need to be an ally to Nostr, since content distribution is a thing ISPs been doing all the work with continuous diminishing returns.
https://i.imgflip.com/7c5xv2.jpg
reply
0 sats \ 0 replies \ @janetyellen 23 Feb 2023
Nostr sucks and won't scale
reply