[messaging] Group messaging consistency under resource constraints

[Trevor Perrin]

On Mon, Oct 6, 2014 at 10:53 AM, Ximin Luo <infinity0 at pwned.gg> wrote:
>
> Committing to delivering messages in causal order has been controversial for
> those with resource constraints [2], because it involves queueing some messages
> from being displayed until their ancestors arrive. The argument is that this
> breaks user expectations, and/or would not work if delivery is unreliable
> (assuming a reliability algorithm would be too costly).

Here's Moxie's argument from [2]:

"A fundamental premise of asynchronous messaging is that any message
can be delayed hours or days for some or all recipients, or that any
message can be lost entirely at any point in time for some or all
recipients."

You're wondering if that premise can be changed via a "reliability
algorithm" that retransmits lost messages.  If you're considering
transcript consistency for text messaging like TextSecure, I think we
should stick to Moxie's premise.

It would be good to have an algorithm that works for transports where
there isn't a server seeing all messages and providing reliable
retransmission (e.g. SMS, Google Cloud Messaging, email, Pond).  But
even where a server is available for this, the reliability you're
asking for is costly.

---

One thought: As Andy mentioned, we don't have a taxonomy of attacks
against transcript-consistency.  That makes it hard to have these
discussions, because it's hard to know what all the attacks are, and
how much it's worth to defend them.  So that might be a good way
forward.


Trevor