Nostr Web Client

Jay 11mo ago

Say you could assign an engagement score for each of the following. What do you think of these weights?

Text reply - 1.0

Reaction - 0.1

10 sats - 1.0

100 sats - 2.0

1000 sats - 3.0 (etc. logarithmically)

Boost - 0.3

Quote reply - 0.6

Multiple of the same type on the same note would diminish exponentially. For example, the first reply would be 1.0; the second, 0.5; the third, 0.25 etc. to avoid counting spam.

#asknostr #nostr #plebchain #grownostr

nostr:nprofile1qqsw2feday2t6vqh2hzrnwywd9v6g0yayejgx8cf83g7n3ue594pqtcpz3mhxue69uhhyetvv9ujuerpd46hxtnfduq3kamnwvaz7tmjv4kxz7fwvf5hgcm0d9h8qctjdvhxxmmdqyt8wumn8ghj7erpwe5kgtnwdaehgu339e3k7mgxeu5tj nostr:nprofile1qqsvyv8d6dx2tjp33069j2kq2mx7xage6w2upyzvxl4pcegt3t22wyspz9mhxue69uhkummnw3ezuamfdejj7qgjwaehxw309aekzarnv9nk2tnc09az7qfqwaehxw309ashqmrpvdjkjmn5dpjhxatw9ehx7um5wgcjucm0d5hsaxvfxr

Reply to this note

Please Login to reply.

Discussion

Node Runner 11mo ago

I wonder if you could scale it according to the running cost of sats/vByte to get a tx into the block chain. 🤔

Jay 11mo ago

Following the weather of on-chain fees?

Could be cool, but if I lost my database and refetched all the events to recalculate the scores, it would be a pain to work in the block fees again.

Node Runner 11mo ago

My thinking is that this reminds me of wallet developers trying to adjust fee rates before there was a working fee market. Of course that was based on projected block space, and zaps are based on v4v.

eee8f902... 11mo ago

To me, quote reply and text reply are equal

Jay 11mo ago

I thought about that. Especially because when I store them in neo4j, you can backlink the text note to it's quote with no overhead.

My case for a lower quote weight is that it pulls your note into the quoter's network and out of yours. Heck, you could potentially argue that it's worth more for increasing your reach.

david 11mo ago

I had similar decisions to make when implementing GrapeRank: various parameters including how much relative weight to give to a mute versus a follow. Ultimately I decided I simply had to play with the parameters and see which ones yielded the most useful results. Right now my default is 3 for a follow versus 50 for a mute, ie a mute carries a LOT more weight than a follow. Those values make intuitive sense to me and are also justified empirically.

Logen Kain 11mo ago

Measuring engagement is why sites like YouTube turned into a dumpster.

Saying one type of communication is greater than another leads to disaster.

To clarify, I don't mean positive is equal to negative, but that a smile is equal to a "thanks" or "nice"

That is, if the intent is the same, sats are equal to likes are equal to comments.

Engagement is engagement, now, you may prefer 10000 sats over a like, but the intent may be the same.

Jay 11mo ago

I agree with you, and I want to calculate engagement between you and me, for example, and use it to build a web of trust. Mutual engagement rather than global engagement.

So between you and me, say we talk often in replies and zap every now and then. We build up a balance of engagement and deepen our trust. These weights are to calculate that. But like you said, you might value these different than I do, which adds a wrinkle to any effort to calculate it.

david 11mo ago

This is why we need personal WoT relays: so we can all set weights according to our individual values, preferences, beliefs, etc.

Of course, not everyone will want to muck around with the details. Which means we’ll need default values. But no set of default values will make everyone happy, which means they’ll need to be configurable for those who wish to do so.

Dawn 11mo ago

I like it except i think more of a range per rating for amount of sats sent, rather than individual benchmarks, due to individual monetary restraint, would make more sense to me. Some the best notes get the small zaps... except the 10. That is currently the more spammy level . So maybe like 11-100 carry equal weight, 101-1000 carry equal weight.... and so on. Does that make sense? Am I understanding correctly? And thank you for asking for my opinion! 😁

Dawn 11mo ago

Coming back to amend that... the same concept would also apply to the person sending the sats, not just the notes. More money does not necessarily mean more trust/interest, etc. Lol sorry.. you woke up my squirrel brain 😊

Jay 11mo ago

It's getting late and my brain is off, but I appreciate the feedback 🫂 I'll get back to it later

Jay 11mo ago

Yeah, I see what you're saying: to have bucket-ranges that carry the same weight. I like that idea.

The idea of these engagement scores is to compare engagement between peers and how each user allocates their attention across their network. So things like zaps would be relative to you and me and our balance, for example. My gut hypothesis is that regardless of the weights of any of these note kinds, the engagement between two friends will always be roughly equal. And "followers" are people who expend a lot of engagement with relatively less in return. You can have big influencers with lots of followers and few friends, because maybe they keep their social circles private rather than in the open. I just want a program that lists my friends' npubs tbh 😂

Dawn 11mo ago

I like it. I'd use it. It would be nice to see a visual representation of extending that olive branch and who's been reciprocal, who's acted aloof, etc... and how I have acted too. Follows shouldn't need to be a marriage. Interests change, value varies, trust is rare.

Raw 11mo ago

The length of text reply should also be taken into consideration. If stats have different value for the different amount. Why is text reply considered a singular thing.

Comments are important it shows quality of a post. Also sometimes comments are done by the same person on the same post. It shows a discussion.

b799e60f... 11mo ago

It should work

b799e60f... 11mo ago

The maximum engaging score for 100k people engaging in one way would be around 600k score points.

If the same 100k people would engage in all possible ways exactly once, then we would have a maximum cap of 800k score points.

Would this theoretical cap cause problems in the long run? Like, having the same messages being propagated perpetually at the top of a search result?