This is the terrible Nostr experience that is fixed by the Gossip model. Listen to the Mike Dilger episode at https://nostrovia.org/.
Discussion
#[0]
Followed the pod on fountain and started the episode for future listen. Thanks, jaf!
Weâre looking at the gossip model and NIP-65 for our client. Thanks for sharing.
unfortunately, the proposal is linked with several China authority apologists based in Hong Kong, gossip, gossip... it won't solve any problems of Nostr
TIL thereâs a nostr podcast!
Link to NIP discussion for those interested:
Taking a listen now
#[0]
Take a listen guys
#[0]
I love watching gossip logs, just to see how it works!
#[1]
By the way, I don't think NIP-65 is a solution to anything or even required to solve anything. It is just one nice addition to the model. It worked already before without that, relying on NIP-05 relay hints, nprofile codes, relay hints in events and keeping track of things.
Watch this relatively old video, it is very enlightening: https://mikedilger.com/gossip-relay-model.mp4
Canât stop thinking of relays past few days - are there mathematic modelling for the logic case and connectivity ? I get numbers better. Also was thinking spams are always filtered at front end and not from source - ie if eventually there are relays catered for porn would that minimise global broadcast of đ , when there are specific interest on politics would that remove govât funded cyberwarriors and further limited by shared blocklist and for all who try to spam free money, would block list prevent it? This might be naive thoughts right now.
On a separate note, relays at the moment focus on accessibility & spams. Iris blocks anyone that you have no contact with or no responses to but still provides accessibility. Paid relays limit by 2nd and 3rd tiers globally and past week I canât see a lot of good posts that used to be from new followers I followed, and this might slow down user adoption, but, might be too early to determine a pattern right now? Itâs been on my mind constantly
I can't quite follow everything, but interesting lines of thought.
Yikes sorry, should have worded it better
Im not entirely sure what youâre getting at. I understand people better than mathematical numbers.
Are you suggesting that if there were xxx specific relays that so long as you werenât connected to one youâd minimise the likelihood of coming across random porn? (And the same with political affiliation relays etc etc).
Perhaps Iâm missing the mark youâre outlining entirely, but Iâm interested in understanding your position.
I really should have worded that better, Iâm sorry -
re mathematical modelling - this on relay connectivity ie how private, safe and spam relays work from a user and x no. of users perspective - in terms of broadcasting, re broadcasting. I get the gist, but I canât fully envision the big picture with all possibilities included such as 2nd / 3rd tier restriction / gossip / Iris models and the outcome of each
re spam (I could be completely wrong on this) - current practice out there is keyword filters, manual filter, image filters.
What if this was looked at differently from the spam motive perspective, for instance
- porn related
- Govâtâs cyberwarrior attacks
- Money and trades
From a porn angle, if there are private porn relays, users who enjoy porn might be more compelled to interact with porn related spam (whatâs spam to one might not be spam to another) - would that then narrow down global broadcast/rebroadcast via engagements ? Somewhat like when Satoshi talked abt reducing spams by Incentivising zombie farms to generate bitcoins instead. What if itâs just about benefiting those who want to view porn ?
From govât cyberwarrior perspective - for example opposition parties or govât who attack on each otherâs users - and hence if there were more politics related relays - would that shift the focus from global to specific users of the relays. And then from there the problems can be further narrowed down via shared blocklist/ users muting ?
From money scammerâs perspective - this most often is quite obvious and hence would shared blocklist help ? I think it requires keyword filtration as well?
The challenge with shared blocklist could be that whatâs spam to one might not be to another so this may need a second filter layer Iâd reckon.
All this might be completely naive thoughts,
You may be interested in this: https://fiatjaf.com/3f106d31.html
I do think there should be porn relays, but we can't have these unless clients start to give users the option to treat relays differently, browse relays, allow users to pick and group and switch between relays, and expose a nice UX for users to publish different things to different relays.
The Nostros, Nozzle and Monstr clients have started playing with these things, I hope they get somewhere.
Coming in time⊠old man here working slowly.
100% âno common relaysâ is a massive problem.
Imho. 1 (or >1) relay must be a catch all no matter what.
Distributed so no 1 entity can take it down would be a double ++(essential).
So long as someone is connected to the distributed âmain netâ and whatâs in the main net isnât human readable/machine decipherable (excepting end point) then I think weâre on a winner.
A massive pain point right now is âhow the f* do you know what relays should be used & for what should each be used
#[0]
solution to discovery, rebroadcast replied to content, publish a 'home' relay in kind 0, indexes of pubkey home relays, clients auto add relays as needed
Went through nip65, gossip, this podcast and your doc - looks like a great way to beat algo and still increase followings and user adoption. I think how paid relays operate would impact this model differently - for example if they only read and write and donât push the event to the relay list, then adoption would degrade. Also the 2nd /3rd degree follower restrictions would reduce the reach. And for ppl who donât follow others as much (passive contributor) but rather read contribution based on topics of interest, maybe interest-based relays or search function, might add value - something else to the mix . I went back to Twitter after a long gap because of the selection by interest. But very interesting growth vision by Mike. Thanks for sharing!
Another scenario - if person A has all paid relays and person B has free relays, some of their 2nd & 3rd degree friends will not be able to view each posts - if, for example, events are not published to the relay list or for other reasons.
An experiment is to set a new npub, follow a few Ppl, add common relays and only read from paid relays. Look for a high following profile and compare from this ânew npubâ vs that high profileâs npub - you might notice some posts missed out and not all people unaccessible via this ânew npubâ. My worry is current mix will slow down global adoption instead of improve it
