How can an implementation of NIP-17 by different developers be so different to the end user?
I understand it hasn't be widely implemented yet, but clients that have implemented it have issues.
Genuinely curious.

- #Amethyst and #0xchat seem to be the closest in similarity - They both communicate (mostly) fine between each other, and chats are created as they should. I think one thing in need of cleanup is that some messages on Amethyst are received as "template via 0xchat".

- #Coracle almost works, but for some DMs it creates a completely separate group chat when viewed on 0xchat. Further, some DMs are sent unencrypted with no option for encryption (if this has anything to do with inbox relays I'm unsure how when NIP-17 messages can be sent in other clients to/from the same users).

- Not sure what the deal with #Iris is. The Desktop app(hasn't been updated in a while) can only do groups, but these groups can't be joined by other clients. Web is missing this functionality entirely, including beta.iris.

I'd love to onboard other people and suggest "use this on your phone, and this on desktop/web", but we can't communicate.

Not sure if I'm missing something... Or if it's just too soon, even though what's there seems like it's supposed to work.
 
#AskNostr 

 I would not worry to much about the state of DMs with NIP-17. They're all going to be outdated and deprecated when NIP-104 is merged and the new standard. 

 I was aware of Whitenoise as a reference client with MLS - Should've realized JeffG would propose a NIP itself.
Thanks for the pointer, hopefully it sees faster/better adoption. 

 I wouldn't use this logic. There is always something better. I am sure after MLS gets coded and merged a new DM protocol will emerge and then nobody is going to code anything because we are all waiting for the perfect spec.  

 Yup, and you define for everyone what better means. MLS might only be "better" for a very small sunset of users. It certainly won't be for me. 

With good enough encryption (NIP-44), new npubs and relays (private or specialized) you can go a looooong way. Without having to wait for specs and while retaining operability with most apps.  

 I hope this isn't the case. This is why NIP-17 never saw large adoption because it wasn't perfect. It was better, but not the unicorn that people wanted. 

 There is never going to be a unicorn. MLS is good but there will definitely be better things. 

 Which is why I hope that developers continue upgrading and implementing these new specs as they get built and released. Otherwise we'll have 20 clients with 10 different DM methods, making a huge headache for users. 

 My point is that we shouldn't just dismiss the state of DMs as you suggested in the first reply. Devs should implement NIP-17 as it stands today even though NIP-104 is in the horizon. 

Considering that NIP-17 took about 1.2 years to be developed, audited, implemented by 2+ clients and merged, MLS will take another year to be ready.  

 It is worth noting that even with MLS coming out, NIP-17 may still be the preferred DM type because it supports device synchronization, whereas MLS does not.