Nostr Web Client

Content discovery is a huge pain point on nostr, and this is a problem I want the grapevine to address. The grapevine score is designed to serve as a weight, useful for weighted averages or vote tabulation. If we have a system where you can either self-attest you like to write about topic X or recommend others as good follows for some topic (maybe badges can do this?), the grapevine can synthesize that info into recommendation lists. If Alice wants to find great follows for SomeTopic, her grapevine will have a rec list for that, and you’ll probably be on it if there are indicators you should be.

#WoT

#grapevine

ᴛʜᴇ ᴅᴇᴀᴛʜ ᴏꜰ ᴍʟᴇᴋᴜ 1y ago

one of the root causes of the discovery problem on nostr is a lack of distinction in the protocol between public and private (of interest to a limited group) events

a profile metadata is public, a relay list is public

really, follow and mute lists should be private

or at least, there should be a private form of them, that can differ from the public form

public events should be broadcast everywhere and accepted by anyone who is publicly followed by any paid user on a relay

private events should be only sent to in/out boxes and be accessible only to those who have been publicly declared friends, and not to those who are enemies (really, mute lists should be called enemy lists, and follows, friends)

making these kinds of distinctions in the protocol would help guide implementation a lot better than the mess we ehave now, where certain things are implicit but not really correct

nostr:nevent1qvzqqqqqqypzpef89h53f0fsza2ugwdc3e54nfpun5nxfqclpy79r6w8nxsk5yp0qyt8wumn8ghj7etyv4hzumn0wd68ytnvv9hxgtcpzfmhxue69uhkummnw3eryvfwvdhk6tcqyplkevx558gmz7lh3ut6rqd2q7n6a35up4uzttmw269fduazp9dc23klze0

Reply to this note

Please Login to reply.

Discussion

david 1y ago

There are lots of different potential sources of information (some currently available, some not available bc they’d require changes to the protocol) that could be useful to tease out the distinctions that I think you’re looking for. The challenge is to get everyone to focus on the same sources of information. On the other hand, maybe we don’t have to and shouldn’t try. The grapevine is designed to take information from multiple sources and interpret them individually into a format that is useful to you, so you can synthesize them into a single score.

I’m not sure what protocol changes you’d suggest to indicate that some event is of interest to a limited group. But whatever it is, it could be used as input into the grapevine scoring system.

ᴛʜᴇ ᴅᴇᴀᴛʜ ᴏꜰ ᴍʟᴇᴋᴜ 1y ago

i'm not suggesting any substantial changes in the protocol but rather in the guidance and some distinctions that would help guide making implementation decisions that would not require complex algorithms like social graph analysis

david 1y ago

If the method you’re talking about becomes widely used, the bots will come out of the woodwork to manipulate it, so we will need at least a rudimentary social graph analysis to weed them out. The baseline grapevine score is perfect for that. It ranges between 0 (bot) and 1 (not a bot), and in-between means not entirely sure yet either way.

ᴛʜᴇ ᴅᴇᴀᴛʜ ᴏꜰ ᴍʟᴇᴋᴜ 1y ago

you are confusing my distinctions by equating a lack of public advertising of association with a lack of useful data for your proposed social graph scheme

saying who is welcome to message you and who is not, is a public notice, but who you actually read and who you don't want to read, and what your priorities are for filtering them should and could be private, but is not

it is a zero-metadata-leak aside from "this person is online" when a user stores an application specific data encrypted blob to a relay, and it would solve a lot of the state problems not just with specific applications but with multiple clients and key rotation as well

your assumption is that you should advertise to the whole world who you hate and who you love, and this is actually kinda dangerous and blunt way of approaching the problems

what makse it out to the public is two things:

"you can message me" "don't waste your time messaging me"

outside of that, there is a huge amount of useful data that could easily be added to these graph calculations just by evaluating the number of times your npub and other people's npubs appear in events and graphing that over time as well as on average

my point is that some of the aspects of the utility of this data do not necessarily or, more exactly, are not actually helpful, or privacy-respecting, if you don't make the distinction between what should be private and what must be public

if you expect everyone should wear their heart on their sleeve, this is fine for your personal life but imposing this onto a protocol is irresponsible and actually kinda malicious

david 1y ago

I make no assumption that everyone should at all times wear their hearts on their sleeves.

My assumption is that you, as a sovereign individual, can and should decide what you do and do not want to reveal, and to whom, and when, and under what circumstances.

My task is to help provide the tools that allow you to do make these decisions.