Oddbean new post about login | logout I like real decentralization, I’m gonna stick with NOSTR 💜🏆
The point is that nostr is not real decentralization, and neither is bluesky. Each model has trade-offs. And can learn from each other. A good path to success is to avoid the "either/or" fallacy aka binary thinking. Satoshi was a master at this. In fact, the word decentralized never appears in the white paper, but distributed occurs 6 times!
What is the line for “real” vs “not real” decentralization? I always understood it as a system has to be decentralized “enough” to overcome censorship, system failures, etc. That’s mostly impossible to quantify until the system is put under pressure in a real world scenario… but it is possible to identify systems which by design are not decentralized.
That line is a psychological line. http is a decentralized protocol, tho some will argue that it is not. But if it is the parent protocol of nostr, so anything nostr can do, http can do, and more. So if you think nostr is decentralized then http must be decentralized. It might be useful to use the term "sufficiently decentralized". The term has become a slogan and a cliche in modern s/w engineering, in any case. You could look at it another way. If the main bluesky node went down it could be critical. If damus went down it would be a pretty bad situation in nostr. If one particular website goes down it doesnt matter too much. If any bitcoin node or miner goes down, it will barely be noticed. If one pubky or bittorrent node goes down, the system is resilient.
Why would it be bad? Half the time I don't even use the damus relay. I deleted it for months. Didn't affect me at all. It seems simple: can I instantly respond to a post unlike by sending with Bitcoin? If so, let's continue discussing. If not, then there's is no reason to discuss.
Will literally nuked the Damus relay and it had almost no noticeable effect (at least for me, or from what I’ve read of others’ experiences). And Damus is my main client, but I also use 1-2 others on occasion, I don’t know much about BlueSky but it sounds like there is basically a single data server.
@matt @melvincarvalho, Since you seem to be defending the current or potential decentralization of BlueSky… how does it handle the current central data relay going down? Or, how do they plan to go from one to many data relays?
Sorry, the second question was meant to be more… How do they play to add more redundancy in the data servers, and in that future state how would they handle it if the ‘top’ server went down?
I don’t think anyone is suggesting that Bluesky’s decentralization is here or substantial. It might be in the future, but it certainly isn’t now. The point is more that we should learn from their feature set, because they have some features that are very decentralizing of power within a social network, even though they run on centralized rails.
Cool, I can totally get behind that. When you say “we should learn from BlueSky”, do you mean ppl focused on Nostr? If no, then I’m lost on your point. If yes, what specifically do you think BlueSky is doing right *as a protocol* that Nostr is doing wrong or worse? And specifically curious in the BlueSky “decentralized features” you mentioned. Thanks
They've actually got more open over time, which has amazed me. More of it, I say! Nostr has become more centralized. You just never know how projects will develop, but all can strive to be better!
Dont forget an important fact that Jack funds nostr to the tune of about 1 million dollars a month just to keep the lights on. Without that, you would see a very different situation. That is a central point of failure right there, that cannot be wished away.
Oh wow I didn’t realize he is doing this monthly, what a an OG 🏆
Its completely false
Is it? How much has Jack donated to Open Sats nostr fund alone? (I honestly don’t know, but wouldn’t be surprised if it was 10+ Million)
nostr needs maybe a couple thousand dollars a month to keep the lights on. Clients already exist and running relays is insanely cheap. Damus will keep working even money runs out. You can’t say the same about bluesky
The problem is software is not static and cannot be You need to update it as the userbase scales and the environment changes, if that was not true, we could just not have NostrDB and still use Branle You also forgot the human cost of operating servers and the gap in user count
I think about 56m but about half goes to btc. I would expect another 21m in march. Let's see.
This is not a serious argument. Jack's generous donations amount to about $1mm a month and makes a massive impact. Especially since most non-payroll devs have left.
there were plenty of clients before jack arrived, even damus. nostr worked fine. this idea that nostr needs 1m per month to survive is extremely misleading and simply wrong.
the truth is that opensats is burning money on gimmick projects
dammit. i say nice things to you and zap you all the time. but you only respond to the retarded things. Thats it. Damus is made from stolen top secret CIA Alien ™️ code and rotten dandelion stems. And it takes at least $7 million opium rubles a month to keep its servers running. And the worst part is, its developed on Xmonad.
Also deepstates are behind the Nostr and almost everyone on the top of Nostr reach is FBI agent!
It's paid annually $21mm a year to open sats + more to other places. You can average it to about $1mm a month. Folks could deny this, or that it makes a big impact on nostr. But that is not a serious argument. https://philanthropynewsdigest.org/news/opensats-receives-21-million-from-twitter-co-founder-jack-dorsey
This doesn't seem correct. $5M in 2023. $5M in 2024. $10M in total since may 2023. Over 18 months. That would be $555K a month on average. But that would be if it's all spent as of today. We know that's not the case though since grants are paid out over the course of a year or several years in some cases. It's far, far less, IMO.
im honestly starting to think you’re legitimately redarded.
He might have a point except the money is being spent poorly. NOSTR development is laughably inefficient. If there wasn't so much money available, devs would actually have to cooperate and focus instead of building garbage on top of garbage on top of garbage, before scrapping that and building new garbage. I don't think much of that money is dedicated to hosting regardless, though. I think most people hosting relay and media servers are just doing it out of pocket. It's probably not an overly expensive thing to do at NOSTR's size.
one thing I think about is how the centralization of their funding might have affected their design. their architecture looks like what happens when you have teams that divide themselves into micro-services.
Well, I was just shit posting in a cute way lol 🫣 because I love NOSTR. But, I suppose I should back up my words if asked. I would say decentralization fundamentally revolves around control, architecture, and governance. Real Decentralization - Distributed Infrastructure - Interoperable Standards - User Ownership - Resilience Against Censorship Illusion of Decentralization - Federated but Centralized Governance - Control Over Critical Infrastructure - Opaque Decision-Making - Dependence on Proprietary Systems As far as I know Bluesky does use the federated AT Protocol, which is a step toward decentralization because it avoids a single point of control. However, federation alone doesn’t guarantee full decentralization. In Bluesky’s case, while servers can operate independently, the protocol and its governance are still managed by Bluesky PBC (Public Benefit Corporation). I would say this means that while the AT Protocol provides a decentralized framework, the reliance on a central governing body places Bluesky somewhere between partial and full decentralization. Real decentralization is about giving users full autonomy over their data, identities, and the underlying infrastructure, eliminating points of centralized control. The illusion of decentralization occurs when platforms appear decentralized but still retain centralized governance or critical control mechanisms. Platforms like Nostr align more closely with true decentralization, while Bluesky’s federated model still has centralized elements to address before it can fully achieve that vision.
Thou shalt not effort post in internet discussions on decentralization. https://www.youtube.com/watch?v=CWrMGXwhFLk
These are thoughtful points, but I'd suggest focusing on "verify, not trust."
Nostr governance is highly centralized—one person has overriding control, often contrary to community sentiment ("It's my project, I can do what I want"). This has driven most grassroots contributors away, leaving a small but dedicated group, including great folks like Alex. Meanwhile, Bluesky's interoperable standards outpace Nostr's, though neither is perfect. Nostr's stagnation (e.g., no markdown, editing, or smart widgets) and resistance to scaling innovations have stifled growth.
Nostr absolutely does not have opaque decision making. It has a centralized king-maker effect with very caustic people at the top. Developers are told, "This is dumb", "This is stupid", "We dont need this", "We dont need relays", and even "Go back to where you came from". Overt and blatant racism is tolerated. 90%-95% of the grass roots have left, but thankfully Jack's amazing generosity keeps much of the project going.
Onboarding is another weakness—while Nostr champions user ownership, the steep learning curve drives users to easier alternatives like Bluesky, Mastodon, or Threads. Transitional tools like subkeys or delegation could help but remain unaddressed. Decision-making isn’t opaque—it’s centralized and often dismissive, with a "king-maker" dynamic that's alienating to developers.
Still, Nostr thrives on Jack’s support, powering good projects like Ditto and Chachi Chat. But sustainability remains a challenge, and much of the development feels grant-driven rather than user-focused.
The points you make a are great ones—just temper trust with verification. I've been here from the beginning, contributed heavily, and taken heat for pointing out issues. There's value in Nostr, but its slogans often need scrutiny. I feel like you spend more time complaining about nostr than contributing anything of value. What are you doing to improve these weaknesses? Some of us are very much aware of these issues and are grinding away to improve them. The governance issue is so dumb, there is no governance. Theres a few devs who have control of the nips repo. Guess what! You can ignore the nips repo, i left it a long time ago and have started to spec things on our own. People who find our specs useful can implement them (image metadata, private zaps, quote highlights). All rejected from the nips repo but *gasp* they still exist and people implement them. Not sure why i waste my time with complainers, but i am going to stop wasting my time now. ✌️
I haven’t retracted my offer, will, if you have a change of heart, amigo. https://image.nostr.build/45526cc607c1f00735bb59292e638eedef8d2b5d1592d2cc252ed50f20435ea5.jpg
Bluesky sucks because it is leaning heavily into leftist radical thought as a scaling strategy. It is creating a dangerous echo chamber. Their starter packs idea was really smart. They know how to scale, but people are using those starter packs As tools to block people at scale. they will hit a brick wall and they are not intellectually prepared for it
