straycat
david

All this time we’ve been thinking about ways to export a list of tens, maybe hundreds of thousands of pubkeys. We could (sorta) do it all with just one event!

2025-05-15 03:15:42

- reply

I tend to agree. Anything that can improve performance and reduce bandwidth will be useful.
Are people using bloom filters? Is there a NIP or a standard in the works?

2025-05-14 05:25:43

- reply

I haven’t been following the bloom filter convos but if an api is going to deliver a list of npubs, like a whitelist or blacklist or WoT-curated list of whatever, I suppose it could be delivered as a bloom filter. What particular use cases are you thinking?

2025-05-12 17:47:35

- reply

I see it

2025-05-08 17:06:22

- reply

PageRank can also be used to filter the packs
@npub1kpt…nhmz ?
cc @npub176p…vgup

2025-05-08 16:43:20

- reply

I def want to do this with Brainstorm once I address some backend database problems. Only show Follow Packs from users whose GrapeRank scores are above some threshold. And if multiple users create Follow Packs on the same topic, then combine them to create a composite Follow Pack curated by your Grapevine, similarly to the way that Brainstorm (still in alpha) currently uses GrapeRank scores to generate composite, Grapevine-curated NIP-56 reports:
https://straycat.brainstorm.social/nip56.html

2025-05-07 16:58:50

- reply

The purpose of my current iteration (see link in my nostr bio) is to determine whether strfry + neo4j is the stack upon which to build brainstorm. I think neo4j is probably the best FOSS graph db, and strfry is arguably the best relay: well written, battle tested, uses LMDB which is super fast. The challenge is to pipe the data from the relay to neo4j, and there aren’t any off the shelf LMDB to neo4j tools for that purpose. There are, however, SQL to neo4j tools, which makes me wonder whether something like khatru (which offers sql and lmdb as db options, if I recall correctly) might be a better choice.
So far with my current iteration, I got the basic strfry to neo4j pipeline working for follows, mutes and reports, it calculates PageRank / graperank / hops, and has a few features on top of what my last iteration does, like a personalized follows rec page and a curated nip-56 reports page. You can check those features out using the link on my bio. I haven’t advertised it much though bc of some bugs that came up - not sure if the data synchronization is keeping up. I’m wondering if it’s time for me to hire some experts to build out the relay to neo4j ETL pipeline.

2025-05-07 14:16:03

- reply

I had a great time in this convo with Shawn & excited to see it finally drop!! 🔥 💜 🎙️

2025-05-07 14:07:23

WOOHOO!!! 🍇 🧠 ⚡️
#note199j…8qzq

2025-05-07 04:28:43

TaaS — Trust as a Service!! 🔥
#note1ttj…gp6a

2025-05-05 13:54:06

- reply

gm! ☕️ ⛅️

2025-05-05 13:43:04

- reply

GM! ☕️

2025-05-02 22:38:24

- reply

If you filter out the bots, probably nobody has more than ~ 10k or so followers

2025-05-02 20:55:02

- reply

Maybe I should start out with a broader framing: how does a group of entities (meaning people but could be other things - aliens, AI, etc) achieve consensus? I’d say there are a myriad of different methods and a myriad of different things about which we strive to arrive at consensus. Some methods are centralized, some are decentralized, and some maybe a little of both, depending on what we mean by those words.
Rally around a king - that would be a mechanism centralized consensus on political issues. Nakamoto consensus is a decentralized mechanism for consensus on what to use as money. Another category would be linguistic consensus, like: what do I call this thing in my hand? A hat, a pencil, a shoe? The simplest most basic mechanism, and the one that I think we actually use the most, is basic WoT: I take a poll of my closest associates, calculate a weighted average of what they call something, and that’s what I call it. (That’s GrapeRank for vocabulary words!)

2025-05-02 20:25:30

- reply

Just framing the question and defining the terms is admittedly one of its most difficult challenges

2025-05-02 20:24:00

- reply

Warfighters would be a form of centralized consensus, would it not?

2025-05-02 18:31:23

- reply

I suppose I may as well. I mean, people do keep saying you can just do things … 😂
Once I finish the optimization I’m working on, maybe that next.

2025-05-02 17:38:00

- reply

Ah, I didn’t realize that (either of those things).
@npub1kpt…nhmz do you want a competitor to offer personalized PageRank via WoT DVM? The point would be to demonstrate interoperability, which I think is important.
Or is the dvm spec gonna die on the vine? I’ve only partly been listening to the dvm debate. Most of my attention has been focused on trying to optimize the sync pipeline between strfry and neo4j.

2025-05-02 16:00:20

- reply

We need at least one client to step up and say: I’m ready to ingest personalized WoT scores from a third party provider. Here’s how I want to use them and here’s an outline of the spec I want to use to ingest them. We could start with personalized PageRank to keep things simple and consistent. It can be used to rank any content whatsoever by author. Basically, an alternative to ordering content chronologically. You’ll quickly have two competing providers: Vertex and Grapevine.
Are there any clients interested in this?

2025-05-02 15:24:22

- reply

That would be easy enough to do but there’s no sense building out a spec that no one is going to use and if the spec is about to be upended anyway.
To my knowledge there are 2 teams eager to offer personalized WoT scores, team vertex and team grapevine, but 0 clients ready to consume them. Why is that?

2025-05-02 15:00:09

- reply

So does npub.world uses your DVM or not? If a competitor were to offer personalized PageRank scores via DVM then npub.world would not accept them?

2025-05-02 03:15:37

- reply

Agreed! 💪🏻😊

2025-05-02 03:07:54

- reply

Almost exactly my grocery list

2025-05-01 21:39:35

- reply

straycat.brainstorm.social currently curates nip-56 reports using graperank. I could do basically the same thing for Follow Packs. Then offer the list of curated follow packs via API which nostrmeet.me could access.

2025-05-01 21:31:48

- reply

Grapevine-vetted Follow Packs

2025-05-01 21:31:03

- reply

New user Alice gets onboarded by Bob
Alice wants to follow musicians
Bob doesn’t know musicians but someone in his 🍇 made a Musician Follow Pack
BOOM

2025-05-01 21:27:07

- reply

💯
If you’re borrowing against your bitcoin, you may as well open a long position on a derivatives market, and if you’re doing that you may as well be a day trader, and if you’re doing that you may as well play slot machines at Vegas.
Stay humble and stack sats is the smartest thing for most people.

2025-05-01 21:17:21

- reply

In this case, you’re talking about providers of personalized WoT scores, right?

2025-05-01 21:06:05

Might Follow Packs be useful for nostrmeet.me?
@npub1man…n9tn

2025-05-01 20:17:59

- reply

That third profile has multiple 🍇-verified reports as an impersonator
https://straycat.brainstorm.social/profile.html?pubkey=990694581c58124489d0b08dfa4d2a7523a713b16a829a0bc161ad82bffdc915

2025-05-01 19:15:01

- reply

So what can you do with the DVM that you couldn’t do with a vanilla API? What problem does it solve?

2025-05-01 18:53:53

- reply

Ok, I can get it to work if I use VPN but not if I don’t — which is weird bc I’m in the US. Even if I VPN into the US, it works. But not without VPN!
I cannot explain it 😂

2025-05-01 18:33:18

- reply

Right now I calculate scores and whitelists based on follows, mutes and reports, and use that to filter spam out of streaming content (kinds 1, 7, etc). So I’m starting to think about how content could be used to create useful metrics. So far, the only metric I use is a “recency” metric. If an npub hasn’t posted within some time frame, it won’t be on the follows recommendations page:
https://straycat.brainstorm.social/follow-recs.html
Have you written up any of the engagement metrics you had in mind? Maybe I could try implementing some of them.

2025-05-01 17:55:52

- reply

a good case for personal WoT relays

2025-05-01 16:36:40

- reply

I thought they used their WoT DVM spec for something
https://github.com/nostr-protocol/data-vending-machines/pull/38

2025-05-01 15:36:25

- reply

@npub176p…vgup and @npub1wf4…dgh9 what say you? Is your WoT API better with or without the DVM part?

2025-05-01 15:12:36

- reply

Have you gotten back to work on it yet? I’ve been making progress in my pairing of strfry and neo4j

2025-05-01 13:28:46

- reply

You’ll be in control of your own trust algorithms. Any piece of data you want to change or update, you do it. Just like real life.
This is why I like the idea of a personalized WoT relay. You’re explicitly control of the data and the algos.

2025-05-01 05:05:45

- reply

We really need a robust WoT, like yesterday, so we can deal with situations like this.
There are certain people I trust more than me to do a deep dive into a topic like this one. And other people would trust me to point out who those people would be. And others who would trust them to pick me to select the OGs who will have good judgment on issues like this. Stuff that the Grapevine is designed to do.
The ability to scale Dunbar’s number from 150 to 8 billion may help us avoid the community splitting in two.

2025-04-30 18:17:22

- reply

My current plan is to fast from time to time up to about 72 hours, keto maybe 50% of the time in stretches of maybe a few weeks.

2025-04-29 23:43:33

- reply

https://nostringer.starknetonbitcoin.com won’t open for me - invalid certificate maybe

2025-04-29 16:36:33

- reply

GM! ☀️

2025-04-29 15:57:39

- reply

Happy Birthday! 🎊🎈🎉

2025-04-29 14:47:26

The Brainstorm personal WoT relay is starting to come together. Have mine running at https://straycat.brainstorm.social. One of the features is curation of NIP-56 reports. From what my grapevine tells me, there are a handful of impersonator accounts that have quite a few followers!
https://i.nostr.build/pyzXlAWeufKx24NI.png

2025-04-29 12:06:33

GM and PV! ☕️ 🌞

2025-04-29 12:03:15

- reply

gm Shawn!

2025-04-29 02:50:58

- reply

Just submitted this as a PR as NIP-97: Ring Signatures
https://github.com/nostr-protocol/nips/pull/1894

2025-04-28 19:06:43

🔥 💜
#note1jtc…9g7l

2025-04-28 15:23:54

- reply

Cool!
I have a draft ring signature NIP — I’d be curious to hear your thoughts Abdel and Leo whether this NIP would be useful with Nostringer. If so I’ll submit it as a PR for discussion.
https://github.com/wds4/nips/blob/master/ring-signatures.md

2025-04-28 01:43:52

- reply

GN and PV! 🫂

2025-04-26 10:52:39

- reply

Yeah but it gets the point across, right? 😂
I reworked these a lot yesterday. Right now there’s frens, groupies, and idols.
Then there’s mutual frens, mutual groupies, and mutual idols.
Self-explanatory?

2025-04-25 19:41:00

Tekkadan Social Graph Analysis:
analysis of the interactions and relationships between the social graphs of two users
- follow recommendations by one user for the other
- mutual follows
- mutual followers
- mutual frens
- mutual groupies
- mutual idols
etc
@npub12zq…99h8
🧠 ⚡️ 🍇

2025-04-24 20:13:18

- reply

Dawn you prompted me to write this draft NIP for ring signatures:
https://github.com/wds4/nips/blob/master/ring-signatures.md
@npub1man…n9tn I made some of the changes we discussed.
@npub19ma…axkl this is the kind of application that will make WoT very powerful.

2025-04-24 19:18:05

🔥
#note1qqq…rn3q

2025-04-24 19:15:09

- reply

Looks from the repo like @npub1ce7…3d7m is using NIP-87 (which isn’t merged) to list and review mints.
https://github.com/nostr-protocol/nips/pull/1110
lists of mints and their capabilities:
38172 and 38173
38000: mint recommendations with ad-hoc 5-star voting:
https://github.com/MakePrisms/bitcoinmints/issues/22
So Brainstorm will weigh the 5-star votes using personalized graperank score, and you’ll have a personalized list of mints recommended by your WoT.
#brainstorm
🍇 🧠 ⚡️
@npub1man…n9tn
@npub10np…tl5h

2025-04-24 18:49:00

- reply

Ohhhhhhhhhh 👀

2025-04-24 18:31:01

- reply

So these are “recommended by Calle” mints? Does nostr have a formal way to endorse things like mints other than a kind 1 note?

2025-04-22 20:42:12

- reply

Precisely what the baseline GrapeRank score is for!
Now imagine a personalized WoT relay that calculates “red flag” scores based on NIP-56 reports, weighted by GrapeRank scores so you can trust the reports. You decide which reports (nudity, profanity, etc) go on your blacklist. Primal queries your relay via API (WoT DVM) to make sure your blacklist is respected.
Whenever a new category of bad actor pops up, the community can invent a new NIP-56 report tag. You tell your personal relay whether to add that tag to your personal blacklist. Easy peasy.
🧠 ⚡️ 🍇

2025-04-22 14:23:39

- reply

Hehe🤭

2025-04-22 14:12:22

It’s insane what vibe coding can do today
Grok only knows what it’ll be able to build a year from now
#note1yuc…p88s

2025-04-22 03:06:48

One of the things that makes GrapeRank different than most (all?) other WoT scores out there in the wild right now is that it is designed to function as the weight in trust-weighted sums and averages. This may not sound very exciting, but it’s pretty important. Kinda indispensable, I think.
#wot 🧠 ⚡️ 🍇

2025-04-22 01:30:09

I’m in the process of incorporating NIP-56 reports in a big way into Brainstorm, your personal WoT relay.
For each user and for each report type, your grapevine will calculate the total number of reports weighted by GrapeRank. Reports authored by spam accounts are ignored bc their baseline GrapeRank scores are zero by default.
For any given report type, you’ll have the option to blacklist users who are reported over some threshold number of trusted accounts.
If people start reporting with custom report types (not limited by what’s in the NIP-56 spec), those will be automatically processed by the grapevine.
#wot
🧠 ⚡️ 🍇

2025-04-21 23:42:07

- reply

My current version of Brainstorm runs a router to stream events and has several streams set up by default. I have a toggle button to edit strfry-router.config when your WoT is calculated (including your whitelist) and you’re ready to start streaming kind1 notes and other content.

2025-04-21 23:38:27

- reply

The strfry router command does exactly what you’re asking for, if I understand you correctly. Just use your WoT to decide which friendly and trustworthy relays you want to open streams with. Use a different filter with each relay if you wish, like if you trust relay1 for wikis, relay2 for zaps, relay3 for kind1, relay4 for profile data, etc. One of many reasons I’m happy with using strfry for Brainstorm.
https://github.com/hoytech/strfry/blob/master/docs/router.md

2025-04-21 15:32:34

- reply

Zaps can definitely serve as a trust signal and contribute towards WoT calculations. Many other trust signals: reactions, replies, labels, badges, reports, mutes … there’s no limit to the number of trust signals we might want to use. This is why one of the most important steps in GrapeRank is something we call interpretation: you pick your signal, write a script to convert the signal into a format ready to be ingested by the GrapeRank algo, and then use it in contextual GrapeRank score calculations.
For now, for Brainstorm (a personal WoT relay under development), I’m interpreting follows, mutes, and reports into a baseline GrapeRank score, the purpose of which is to filter out bots and spam. At some point I’m thinking I may interpret zaps, reactions, and replies and use them to create one or more “engagement” GrapeRank scores that can be used in a variety of ways. Eventually you and your community will have the tools to pick your own signals, create new interpretation scripts, and create new GrapeRank scores. All of these scores will be made readily available to any nostr client that wants to use them.

2025-04-21 05:10:20

- reply

I think we’ve evolved to internalize logical fallacies under certain specific conditions, particularly if they’re promoted by someone we accept as authoritative. If our ancestors hadn’t done that (and known when to do it and when not to do it), they would’ve been rejected by the tribe and weeded from the gene pool.
Virtue signaling is one method of decentralized consensus formation. Love it or hate it, it’s an important and necessary function. Logical fallacies can serve a very specific purpose, which means they shouldn’t reflexively be chalked up as evidence of stupidity.

2025-04-21 04:31:52

- reply

As a general rule, the rationale for flaunting one’s acceptance of obvious logical fallacies is that it demonstrates one’s loyalty to the tribe. You have to drink the kool aid to gain acceptance to the group. The more bitter tasting the kool aid, the more loyal you gotta be to drink it.

2025-04-20 21:48:43

- reply

Something along those lines seems inevitable

2025-04-20 21:18:31

- reply

Seems like this would be a remarkably simple NIP.
Maybe the tag should be a single letter instead of “ring-signature”. Perhaps R. I’m not sure how many single letter tags are taken …

2025-04-20 20:22:39

- reply

Does that make sense? We have two sigs: the usual nostr one in the “sig” field and the ring sig as the third item in the tag.

2025-04-20 19:52:16

- reply

It’s a ring signature spec bc the content field is verifiably signed by one of the <pk>s in the tag, without revealing which one.

2025-04-20 19:42:37

- reply

I’m thinking:
NIP: vanilla ring signatures
event type: any
authored by a one time, throwaway pubkey
special tag:
[
“ring-signature”,
“[<pk1>, <pk2>, …]”,
<ring-signature>
]
The message that is signed by the ring signature will be whatever is in the content field.
The fact that we’re using a one-time use throwaway pubkey means that WoT will not be a feasible method to prevent spam. It will therefore be recommended but optional to use something like PoW for spam mitigation. If spam becomes a problem, relays can reject any note with the ring signature tag that does not meet the PoW threshold.
And that’s it. Pretty simple.

2025-04-20 17:55:47

- reply

I want this 😋

2025-04-20 17:45:29

- reply

Once a handful of people are using Brainstorm, you’ll have the option of syncing with them. If you don’t know who they are, sync with whatever relays that you know about, like primal or damus, and them switch to decentralized relays once your WoT tells you which ones exist and can be trusted.
With Brainstorm, the initial baseline GrapeRank WoT scores are to sort the real users from the bots, nothing more complicated than that. Nothing contextual, doesn’t require all that much data. For now, I’m using follows, mutes and reports and those can sync up pretty quickly. Once the bots are filtered out, you can start to build a few contextual trust metrics, and one of the first will be trusted relays.

2025-04-20 17:19:06

- reply

By nodes you mean relays?
Here’s a business model: Personalized Webs of Trust relays. You’ll want to run one as an individual because it will keep track of who you trust and in what context. Running the relay yourself means you’re in full control of the algorithms that calculate your WoT scores. You’ll sync your relay with big relays like primal or damus but also with trusted friend relays which will make the big centralized ones redundant / unnecessary.
The above product is what I’m building now. Calling it Brainstorm. 🧠 ⚡️ It uses strfry under the hood, arguably the powerhouse relay of nostr. I’m keeping negentropy activated so you can sync your Brainstorm relay with trusted friendly Brainstorms quickly and efficiently. Whatever content you care about goes into your relay. Personalized WoT scores are calculated using a variety of methods, including PageRank for noncontextual trust and the much more versatile GrapeRank 🍇 for contextual trust.
You’ll be able to run your own Brainstorm. But as a business model, a service like @npub1fvm…szfu can host your Brainstorm for you.
Imagine if things like Trending content and Recommended Follows are communicated to primal via your Brainstorm. Personalized to the max. No more global feeds!

2025-04-20 16:19:46

- reply

The closest I could find to a draft proposal is NIP 404 Ghost Events — seems to me we could make one much more simple than that. A vanilla Ring Signature NIP, nothing else. Maybe a special event kind, or maybe any event kind but with a special tag to indicate that the sig is a ring signature and to specify the list of pubkeys in the ring. The difficulty would be that relays would normally reject these events unless they know to check for valid ring signatures, and that would require a pretty big change in how relays work. And whose pubkey goes in the event pubkey field? Maybe we’d need a throwaway pubkey as the author to act as a wrapper for the ring-signed message.
Shall we try writing one up?
https://github.com/nostr-protocol/nips/pull/1676

2025-04-20 15:46:50

- reply

From the Wikipedia article on ring signatures:
In the original paper, Rivest, Shamir, and Tauman described ring signatures as a way to leak a secret. For instance, a ring signature could be used to provide an anonymous signature from "a high-ranking White House official", without revealing which official signed the message.
I could see this being useful. In some instances, people would feel more comfortable issuing a NIP-56-like report if they could hide behind a ring signature. This would provide an additional source of data that we could feed into a graperank interpretation engine.

2025-04-20 15:17:47

- reply

I’m imagining that if I call up customer service for some company and tell them I’ve lost access to an account and need them to restore access, I could ask N people in my WoT to attest that I’ve told them in the real world that my request is real, I’m not under duress, etc. This would enable the customer service rep to process requests more efficiently.

2025-04-20 15:09:45

- reply

Hmm, that’s interesting. I bet there could be some applications there. Like if Alice wants to report Bob, she doesn’t want to reveal her identity as the reporter, but she wants the report to be trusted so she makes a list of N trustworthy pubkeys where she’s on the list and signs using a ring signature.
Perhaps FROST could be used for this. Not exactly ring signatures, but related. Any thoughts @npub1man…n9tn ?
Definitely something to keep in the back of our minds!

2025-04-20 05:27:50

DCoNR:
Decentralized Curation of NIP-56 Reports
#WoT
🍇 🧠 ⚡️
#note16np…9un4

2025-04-20 05:14:06

- reply

You could use a NIP-56, kind 1984 event to report a pubkey using “paid shill” as a custom report type.
Your personalized WoT relay could gather all reports with this type (or similar types), discarding reports authored by pubkeys not in your WoT to mitigate gaming the system. If paid shills becomes a problem that people actually care about, they can start using this system.
https://github.com/nostr-protocol/nips/blob/master/56.md

2025-04-20 04:17:35

- reply

Imagine a personalized WoT relay that decides what goes on your feed, trending lists, recommended follows, etc, and the client fetches that from your personal relay. Which means that if you want mute to mean mute, then it does bc you control your relay.

2025-04-20 03:58:47

- reply

🎯
And hopefully, the people in my grapevine who are paying attention will be able and willing to flag the bad actors as such. (This would be an example of what I mean by explicit contextual trust attestations.)

2025-04-20 03:56:19

- reply

💯% agree
This is why we need explicit contextual trust attestations IN ADDITION TO proxy indicators of trust. If someone I trust flags someone else as doing specifically the bad thing you mention, I’ll know not to trust that person in that context.

2025-04-19 22:16:33

- reply

2 more hours to go in my 48 hour fast. Just espresso & electrolytes

2025-04-19 19:29:44

- reply

@npub1aka…l9dl agree?

2025-04-19 19:28:58

What can WoT do for businesses?
How about this: address the problem of hacked accounts via social engineering. Empower customer service to require social proof before providing access to sensitive data.
@npub1man…n9tn
@npub1clk…pup9

2025-04-19 01:22:46

- reply

I think I have that book on a shelf somewhere along with 20 other books in the freedom tech space, all on my to-read list lol. I hadn’t heard Jason talk about WoT but I’d be curious to know how deeply he’s thought about it

2025-04-19 00:43:23

- reply

Did you find them useful? Still use them?

2025-04-19 00:28:12

- reply

I’m eager to see you make more progress!
I’m making good progress on my personal WoT relay. Baseline trust scores depend mostly on follows (with some dependence on mutes and reports), but more algos will be incorporated; having multiple ways to calculate trust scores is the idea.

2025-04-18 17:17:09

- reply

This is my theory why we haven’t met any aliens. Because they always send their nanobots first, and in all those timelines, we’re dead.

2025-04-18 16:23:57

- reply

How long until we build self-replicating AI nanobots and send them through space in every direction at 0.99 speed of light

2025-04-18 13:10:22

- reply

Gm! 💜

2025-04-18 13:08:23

- reply

Gm! 🧡

2025-04-18 12:59:23

- reply

gm! ☀️

2025-04-18 00:21:38

- reply

Would be cool to show a big convo where one of the axes is time so you can see where it branches off into threads. Each node is a note, and maybe the radius of the node represents some WoT-derived importance score.

2025-04-18 00:16:25

- reply

@npub12zq…99h8 you were just telling me about this article

2025-04-17 14:51:26

- reply

Replace it with a verified follower count, where verified means in my WoT.