pippellia
Pip the social graph guy

I'm not sure how to continue my programming journey, so I'm asking here:
I've only ever used Python, and I love it, except when it's slow (which is *always* when you're interested in writing very efficient graph algorithms like I am).
I like simple, clean code, I've mostly never used objects or classes, they're not particularly useful for what I'm doing.
So, after some research, I am thinking about learning C or Go. The thing is, C is much more widely used and supported, but I don't want to become a hardcore developer who has to manage memory and shit. I want to be a mathematician who can write good code, that's all.
Also, I like that Go makes concurrency easier, at least that's what I've read, which is something I'll need to use a LOT because of the particular algos I have in mind.
What do you guys recommend? cc @npub1fjq…leku

2024-09-03 09:43:04

- reply

fedimit is an open source project. You are talking about Fedi the app developed by Fedi the company.
Fedimit solves many interesting problems related to consensus, that are absolutely non trivial to solve, that's why I think it's silly to try to recreate them.

2024-09-03 08:34:49

- reply

fedimit exists too. I don't see any reason to hack something into becoming a consensus machine when there is already a battle-tested implementation.

2024-09-03 07:50:33

- reply

What's the difference with Fedimint?

2024-08-29 11:29:00

- reply

not sure. That's true for our "community" of Bitcoiners, but that might not be true in general.

2024-08-29 09:22:34

- reply

not only in that case, but also when we are talking about a new user that doesn't follow anyone.
spoiler, 99.9% of the world is in that situation

2024-08-29 09:21:42

- reply

ofc, but that's why I said "if u don't have a pov", meaning u are a new users that doesn't follow anyone.

2024-08-29 07:52:45

- reply

Creating an identity on nostr is basically free. So imagine an attacker creates a whole new network of nibs following each other, exactly replicating the structure of the real network of npubs, same structure.
The two graphs are isomorfic, so any graph analysis will yield the same result. Hence, if you don't have a pov (e.g brand new npubs that don't follow anyone) you can't distinguish between the two, hence my focus on PoW for this case.

2024-08-29 07:49:07

- reply

It's difficult to game it tough. An attacker can pay a miner 10$ to get his PoW checkmark, however, unless he's followed or interact with many other real an trusted people, he'll lose that initial weight in algos like Trustrank.

2024-08-29 07:34:32

- reply

Agreed.
PoW keys are a step in the right direction, but even better are what I call PoW endorsements.
https://pippellia.com/pippellia/Social+Graph/Navigating+the+social+graph#PoW+endorsement

2024-08-27 09:06:35

- reply

it's not rocket science
https://m.primal.net/KPPC.jpg

2024-08-27 09:03:13

- reply

I think Mallory is always the enemy, spooks are enemies so that's Eve.
Charlie and Dave sometimes are good sometimes are bad

2024-08-27 08:57:58

- reply

Yeah, in some situations Dave is the bad guy, but u are right that more generally it's Eve

2024-08-27 08:05:29

Dave, being the real bad guy, is free to this day. There is a lesson here
https://m.primal.net/KPLn.jpg

2024-08-26 19:36:39

- reply

ahahahahhaha good one.
cc @nprofile…lde9

2024-08-26 19:24:13

- reply

this Europe man

2024-08-26 10:57:32

and parties are superior
#nevent1q…pw4l

2024-08-19 11:56:06

- reply

cc @nprofile…pyug

2024-08-12 18:17:44

- reply

What if I take an ASIC, make it mine, and people can pay 10$ to me to have these PoW keys or even better these PoW endorsements (so you can use the same old key), in order to get a "golden badge" displayed next to their name on clients, as a form of anti-impersonation.
Like, I would pay for it, and other would as well I think.
And if the market for anti-impersonation service gets big enough, it can actually work at scale.

2024-08-12 17:46:27

- reply

okay, so just for fun. I thought you did it as an anti-impersonation mechanism

2024-08-12 17:27:00

- reply

oh, thanks for sharing this interesting story, and sharing the code I used.
However, I was more interested in why did you mind that key instead of using a random one?

2024-08-12 15:28:51

- reply

genuine question:
why did you mine your npub so that it ends with 'mleku'?

2024-08-12 15:11:57

- reply

I agree paying a relay solves the problem of spam for that relay, but that's not necessarely applicable to every service. Some services aren't directly monetizable (not because it's impossible, but because people won't pay for that), and thus another spam prevention mechanism must be used.
Also, many people have PoW keys (you included), so there is clearly demand for these things. My intuition is to use these and/or PoW endorsements for anti-impersonation and spam prevention.
> unless there is a difficulty consensus you are fucked, whoever has the most compute at their disposal can drown your "spam protected" channels.
yes, it can happen, but:
a) that's not going to kill nostr, unlike these shitcoins you mention
b) it's going to cost the attacker a lot of money, and the attack will be public, so people will stop using this spam prevention mechanism and the attacker would have accomplished very little

2024-08-12 15:03:48

- reply

> turning elliptic curve signatures into a new inflating monetary asset
this is not what I am saying; What I am saying is mining over a nostr event.
the event:
{ target: npub1xxxxxxxxxxxxxxxxxxxxxxxx
nonce: 59h998543...
}
If you sha256 this event, the result starts with a certain number of zeros.

2024-08-12 15:01:17

- reply

> That sounds like Damus leaving sats on the table.
idk, maybe. Not every service is directly monetizable I think, even in a bitcoinized world. Plenty of examples from non-bitcoin product and companies.

2024-08-12 14:38:12

- reply

thanks for the zap mate.
For example, Damus relay is free to use because Damus the company monetize in a different way. But they still need some DDoS prevention, and here is where PoW keys/ endorsements are useful.

2024-08-12 14:35:43

- reply

wo wo calm down cowboi lmao.
What I have in mind is to use
npub_pow / total_pow.
So, if a miner heavily attacks this system, it will increase the denominator. Unlike shitcoins that need to mantain global consensus and are "defended" by the total pow online at that moment, this considers the total pow over the whole history.
So it's a strict improvement.
Then by re-using SHA256 one can use ASICS for Bitcoin (I think) and offer these PoW endorsements I talked about here https://pippellia.com/pippellia/Social+Graph/Navigating+the+social+graph#PoW%20endorsement.
So, you mine blocks with the ASIC, but when the price people are willing to pay for PoW endorsement is slightly higher, you go and mine that. This should result in having basically the same price for hashrate.

2024-08-12 14:26:48

- reply

yes, they would, but not every service is *directly* monetizable.

2024-08-12 13:48:06

- reply

more generally, you can use someone else's curation as your own only to the extent you trust the curator. PoW has the benefits you don't have to trust anyone, which is I argue an important property if you take the PoV of service providers like relay operators.

2024-08-12 11:53:09

- reply

yes, and in some contexts this is another tool in the wot toolbox.
However, any payment or zap is signal *if and only if* one is absolutely sure the sender and recipient aren't colluding or aren't the same person.
So, if you already trust Fran as moderator, then you can assign a non-zero level of trust to anyone who:
1) has posted on Fran community
2) hasn't been removed
However, this system is a bit fragile when you try to use it as an anti-spam measure for relays for example.
Why? because one moderator (like Fran) becoming malicious is enough for open up a DDoS attack vector.

2024-08-12 10:25:55

- reply

Generally agree. My take on how I use it quite effectively with my work:
1) List generator. I am (and I would argue most humans are) kinda bad in making long lists of things, so I ask the LLMs to generate it for me like:
"rewrite this phrase 20 times".
"Give me 20 synonyms for this word"
"Give me a list of 10 inequalities I can use here to prove this result"
2) Do simple things I don't want/don't have time to learn. For example I wanted to make a simple redirect from pippellia.com to the Obsidian link. Don't know shit about the web, so I asked and it did the job.
3) Replace Stack overflow / looking at the documentation for beginners questions like
"What's the pythonic way to write this"
"how can I get the position of an element in a list"

2024-08-12 10:12:08

- reply

all of the things you mentioned requires time and social interaction, which defeats the purpose of having a disposable key that you can use only once and then throw away.
If we are talking about long lasting identities, then yes these approaches are good solutions.

2024-08-11 19:38:09

- reply

lol

2024-08-11 19:37:48

Pineapple on pizza is disgusting.
Mushrooms on pizza are acceptable.
(I am Italian so you can either agree with me or being wrong)
#nevent1q…6q3t

2024-08-11 19:29:51

- reply

+1 on this question

2024-08-11 14:58:22

- reply

doesn't add anything. The thief can post something outrageous with your nsec and 2s later can steal the money.

2024-08-11 13:00:06

- reply

Interesting take, thanks for sharing it, I didn't know it before

2024-08-11 12:49:46

- reply

why do they require that?

2024-08-11 11:42:44

I've watched the Fedi annoucement video and it's cool and well made and inspiring.
But still I have no clue about how Fedi the company is going to make money.
Maybe they'll use their earned reputation to build a Bitcoin bank? No idea, but I've always wondered how companies like that can go years without any monetization avenue. Do you have more clues than I have?

2024-08-10 11:50:34

- reply

sounds really promising. Do it bro

2024-08-10 11:37:34

- reply

why did you copy plan B tweet?

2024-08-09 19:40:55

- reply

Ops, I re-read it. PoW keys / endorsements solve the reputation part I believe

2024-08-09 19:11:36

- reply

what do mean exactly with reputation?

2024-08-05 23:01:28

- reply

a) It's not volatility, it's long term trend into nothingness
b) why creating a new money supply for paying for storage? money creation is tight to ethics.
c) I haven't come up with a solution (yet), just pointed out that this one is stupid
d) If Bitcoin were just a store of value, that alone would make it the most valuable asset because the inability to save IS the problem of this century

2024-08-05 11:44:12

- reply

I do, master Yoda

2024-08-05 11:03:37

- reply

not I, but you*

2024-08-05 11:03:28

- reply

why wife is happy when I down?

2024-08-04 15:02:06

- reply

It's like using a tank to move a piece of furniture in your living room

2024-08-04 13:14:45

- reply

Blockchain to solve spam? Sounds stupid

2024-08-03 17:55:02

- reply

I get that, but for something key related to be used, there needs to be some review and consensus. Being in the Nip repo gives that

2024-08-03 17:50:50

- reply

yes, go for the nip

2024-08-01 18:12:58

- reply

and some of these metric can be improved with training, so are we going to create a category for those that don't train? lmao

2024-08-01 16:08:35

- reply

#nevent1q…ja3e

2024-08-01 16:07:28

- reply

great idea, why using common sense when there is *clearly* a way to measure the impact of every body measure like:
- weight
- muscle mass
- arms' length (which is important in boxing for instance)
- hormones levels
- density of the skin...
Who is more likely to win in a boxing match, someone who is 2kg heavier or someone whose arms are 5cm longer?
You think the solution is making 10k categories? lmao

2024-07-31 12:31:43

- reply

Yes, this is a useful function that you can delegate to your social graph.
E.g. if an npub got 10 mutes from the set of npubs that are 3 hops away from you (follows, follows of follows...), then it's muted.
However, I think about the other way around. Everything that is not in the whitelist is not shown.
Whitelist can come prepopulated by social graph (again, think about all npubs from 3 hops from u)

2024-07-31 11:06:18

- reply

hey, would you like to beta test some recommendation systems?
I'll give you 2 lists of npubs (let's say 20 npubs you don't follow each).
After getting your feedback, I'll tell you how these are derived

2024-07-28 19:00:48

- reply

Nostr and TBD's Web5 have a lot in common. Loosely speaking, DIDs are like npubs and DWN nodes are like Relays.
DIDs vs npubs
- DIDs have the benefits that the Identity itself contains some metadata about the entity, so you could have multiple keys, and endpoints for contacting the entity (endpoints can be your own DWN, which is like a private relay)
- DIDs allow for key rotations, which means they are more secure
- DIDs are more complex
DWN vs Relays
- DWNs are mostly designed for private things, data is encrypted with your DID's key.
- DWNs seem quite complex for me, you can do a lot of stuff with these, like running Apps like you can on Umbrel OS or other personal server OS. So I think about a DWN as Personal server OS + Relay.
So, my take is that I don't know, different tradeoffs that can be summarised in a couple of points:
- more complexity
- more features (are these important?)
- users are required to have a DWN, either self hosted or rented. (Google Cloud is currently helping bootstrapping that network)
- less network effect

2024-07-27 15:23:17

Absolutely beautiful 😍
#nevent1q…s4te

2024-07-27 10:18:14

- reply

I am sorry dude, I hope it will be returned

2024-07-26 19:05:30

- reply

@npub1nxa…a6q8 you can use this for speeding up the Pagerank computations your DVM

2024-07-26 18:48:14

- reply

@npub1h0u…rwx8 you guys are really thoughtful and fucking focused 🔥

2024-07-26 15:16:18

Nostr can't win without solving spam, discovery and trust.
How? I don't have the complete answer, but I’ve got a new tool for you:
Subrank – Pagerank reimagined for the post-truth world 🧵
https://m.primal.net/JfFb.png

2024-07-26 09:07:19

- reply

I had a similar argument to a dude saying we "need to explain users about Relays and how Nostr works".
It's obviously good if your users know about the technical details, but that can't be necessary.
Like, wtf are we talking about, we don't know how 99% of the tech we use daily works, why would Nostr be different.

2024-07-18 18:27:35

I've been recently interested in applying known algorithms in new "post-truth" contexts.
For example, Pagerank.
Pagerank is prone to manipulation, we all know that, but did you know you can overcome virtually all limitations by constraining it?
Just don't run it on the whole graph, but on a subgraph.
Who cares about the whole anyway? There is no whole, there is no global. Embrace it.
One user, one relevant subgraph. Seems like a lot of computations? Not really, not anymore.
Stay tuned because next week a new cool algorithm will be published. This time with functioning code!

2024-07-17 18:49:09

- reply

I am sorry mate, it must be a nightmare.
Don't dispair, I'm sure you'll sort this shit out eventually

2024-07-14 07:38:53

If you don't know the mint, npub or any other entity really, either directly or indirectly (social graph), we have to assume it's not trustworthy.
This will become more and more true as Nostr grows
#note1m7k…jfs5

2024-07-13 06:42:41

WE'RE GONNA HAVE THE BEST UX
#note1qzy…qz34

2024-07-10 07:50:57

yes
#note12cs…fy0e

2024-07-08 18:50:49

- reply

Yes. Both are protocols, both started fairly decentralised.

2024-07-08 18:48:07

- reply

why?
Doesn't your mom use email?

2024-07-08 18:18:52

- reply

does your mom knows how email works?

2024-07-07 18:23:28

- reply

yes, the design space is vast for social graph stuff.
There is, ofc nuance in the discussion, for example to whether the default should be private lists of IRL friends or public ones, and what the optimal user experience looks like

2024-07-06 19:52:45

- reply

@npub1t3g…upxq

2024-07-05 07:29:06

- reply

I'm working on it. Downsides depend on the use case

2024-07-04 21:39:15

- reply

ah I see now. I wasn't aware of this scheme, such a shame!
The difference is that for each chat, there is a shared secret key that encrypts it.

2024-07-04 21:22:19

- reply

can't the receiver just decrypt the event and share the result?

2024-07-04 21:20:10

- reply

Lyn is great.

2024-07-04 21:18:36

- reply

I didn't read Kind 4 DMs spec, but normally in encryption:
- the sender encrypts with his private key
- and encrypts with the recipient pub key.
So the recipient:
1. decrypt with his private key
2. decrypt with the pubkey of the sender
So, if the recipient just does 1., he can share the message which is "signed"

2024-07-03 17:45:27

does anyone know promo codes for Baltic Honey Badger?

2024-07-03 17:12:12

- reply

@nprofile…azrn Lewis Hamilton is vegan. Good counter example I think.

2024-07-03 17:07:47

- reply

I've checked and I was wrong. He's just intolerant to gluten.
This article based on his book seems to suggests he avoids red meat.
https://www.gq.com.au/health/wellness/heres-what-novak-djokovic-eats-in-a-day/image-gallery/7b423e086cfa05addc3463faba61224b
https://image.nostr.build/f75cc9f07e4407056a6857dcc1b48bf091af5e63f494379cd1e180fdff2c9f31.jpg

2024-07-03 16:57:50

- reply

AHAHHAHAHAHAH.
I have always wandered how Djokovic can compete at that level being vegan. Maybe an exception, idk

2024-07-03 16:53:25

- reply

yeah do it. I am a symmetry autistic

2024-07-03 13:23:36

It's coming, it's just taking more time than expected.
I'm not going to release something unless I think it's truly excellent, even if that means delaying the release.
#note19m4…gh9k

2024-07-02 21:11:45

- reply

for sure. Spanish 2nd

2024-07-02 20:09:20

- reply

thanks for the encouragement

2024-07-02 19:42:54

- reply

AHAHHAHAHAH @npub1t3g…upxq and I Will play this in Riga

2024-07-02 19:39:23

- reply

social graph stuff?