That $50M/yr server budget that Signal has is looking riskier more and more every day…
Discussion
And $20M/yr for verification SMS alone if I remember correctly.
I wish I could easily run my own instance and still communicate with the users who only use Signal servers.
Where we are going you don't need an instance.
Lol Technology promises are frequently made and rarely kept.
There will always be an instance. You may not be funding and maintaining it, but someone will be. They can be distributed or centralized, but there will be instances.
Software isn't magic. Resources are required.
Instance of what? It’s all client side except for the relays.
How do you think of signal application relays nostr:npub1zuuajd7u3sx8xu92yav9jwxpr839cs0kc3q6t56vd5u9q033xmhsk6c2uc?
Who is paying for these?
Signal app servers perform three functions.
One, they are the source of truth for your identity in the system. Dangerous point of control.
Second, since the signal protocol isn’t efficient at all for groups, they have to do the re-encrypting and fan out to all group participants. This is heavy work.
Third, they process all media. We’ll leave that aside though since we’re going to put that on blossom servers that will almost certainly need to find a business model.
> they are the source of truth for your identity in the system.
Interesting point.
> Second, since the signal protocol isn’t efficient at all for groups, they have to do the re-encrypting and fan out to all group participants. This is heavy work.
Whats the solution on white noise : your protocol, and who handles and pays for this compute? Client, relay, or both?
> Third, they process all media. We’ll leave that aside though since we’re going to put that on blossom servers that will almost certainly need to find a business model.
Very interesting. Maybe users are willing to pay for image hosting, this is a perk of being a paid supporter, and this is how blossom servers make ends meet 🤔
There is no compute. MLS is just a way more efficient protocol so when you send a message to a group of 2 or 200,000, you just have to send a single message to relays.
For group messaging, blossom servers definitely don't need to keep the files around forever, they should be downloaded/cached to the client device. I think an expiration flag on blossom blobs would be a good addition to that protocol. Then White Noise and other chat clients could signal that the files should only be kept around for a month, for example, when uploading.
We have barely begun to scratch the surface of what business model innovation looks like in this space. ALMOST everyone who is building on nostr is still trying shoe-horn legacy business models (pay-to-play, saas, etc) into a structure that is fundamentally different.
Digital bearer tokens make non-interactive micro-payments possible. No one is bullish enough on that (except maybe nostr:nprofile1qyt8wumn8ghj7etyv4hzumn0wd68ytnvv9hxgtcpzemhxue69uhks6tnwshxummnw3ezumrpdejz7qpq2rv5lskctqxxs2c8rf2zlzc7xx3qpvzs3w4etgemauy9thegr43sugh36r ) !
I'm convinced that there is lots of fertile ground for very profitable businesses in this new world but people need to stop thinking in terms of capturing users and using outdated models.
> I think an expiration flag on blossom blobs would be a good addition to that protocol
nostr:npub1ye5ptcxfyyxl5vjvdjar2ua3f0hynkjzpx552mu5snj3qmx5pzjscpknpr what do you think?
It would be useful, but unfortunately there isn't an "expiration" header in the HTTP spec. so we would need to make our own
Added blossom expiration issue for y’all’s consideration.
And that’s a problem why…?? 😅
> Digital bearer tokens make non-interactive micro-payments possible.
nostr:npub1zuuajd7u3sx8xu92yav9jwxpr839cs0kc3q6t56vd5u9q033xmhsk6c2uc what does this flow looks like for your typical whatsapp user?
For example:
1) Have / obtain sats
2) pay per message to relay?
All that free shit is just app communism. Good riddance to bad rubbish.
If you can't build and maintain freedomtech for pennies on the dollar, it isn't freedomtech.
Keet.io to the rescue! 🙌
Yay! Another closed-source "trust me bro" P2P app... 🤔
Fair point, they of course do have to make it truly open source first..
But so far I still think they will do so eventually the already have the modules on github I th:nk there is hope 🤔
They have no intention of doing so.
Why not? I thought they wanted to just get it working properly first
But I guess we'll have to just wait and see
It would be pretty stupid if they didn't opensource it though.. No one would trust it so what would be in it for them?
You seem to be quite technical, have you looked at their stuff (docs.pears.com holepunch building blocks like hypercore)?
Do you think this has potential to be used for P2P apps and if not why?
I'm not very familiar with javascript and nodejs thus unable to fully grasp how it works but I get the underlying principle and it sounds like it should work and be a great addition to other aproaches like nostr 🤔
I haven't looked deeply at it but my take from skimming (which has been confirmed by several other engineers that have played with it more deeply) is that it's way too complex to ever actually get much use from the community. Because of this, it's very likely that it'll remain a niche tool only used by a handful of people.
In case you didn't know... White Noise doesn't require any infrastructure beyond standard Nostr relays.
No worries, CIA and NSA will keep paying it 👀
Is your client testable yet? Looking forward to it...