PGP, Mail, Devices, P2P, Bans, Utility [was: UK Crypto]

[grarpamp]

> Its easier to make a slick UI than convince people to do work. Is it
> so much to ask that people who make software try to make life easy for
> their users?

Apple understands and successfully applies that. They put their UX
and code camps together. It's something open source coders should
solicit more into their fold... open source UX'ers.
Assuming for example that Apple's phone crypto really is as good
as their word, then the only real weakness is whatever cheesy
screenpass it takes to login, yet so long as you *can* enter a
strong 128 bit, insert a cryptokey, whatever... then it services
everyone well enough.

> You can drive a car mostly successfully without too much information but
> if it stops at the side of the road and the limited info from your dials
> doesn't tell you what's wrong then a bit more knowledge might just get
> you home.
> If anything, this matches the Enigmail model more than the invisible
> crypto model.

Sure, level based access to more details and knobs is fine.
Dummy Mode, Medium Mucking, and Advanced Footshooting

> - Your key gets signed by your friends, so now your friend network public
> - Emails with PGP are provably from you, in a way that can be traced
> back to physically witnessed government ID.

Then don't have your key signed, and or be a nym and sign those of
nyms you trust based on context, not on govts. Use of WOT is an
option there if you want it, not every use case of PGP needs it.

> PGP as a *product* it was a complete failure. I mean, it
> doesn't even protect metadata.

No, PGP is excellent, at what it does.  Like any other tool, it
fails when people foolishly try to make it do what it can't in
situations it shouldn't.  There is no blame on the tool there, only
upon the people stubbornly bashing their head trying to make it fit.

Continuing to talk about traditional SMTP transport services as
fixable by bolting on whatever... is futile, ignorant, and old.
All the people circlejerking on Metzdowd about designing crypto
fixes into existing SMTP models, and all the newfangled encrypted
webmail providers fall into that category.

> the metadata issue a big problem for everyone who connects to a
> server that isn't owned by them  and I suspect really requires a new mail
> protocol to resolve.

Owning both ends hides the non-body elements, but you're not
hiding dns lookups, sizes, the fact that you hit send, etc.
It's not a new mail protocol that's needed, it's a whole
new messaging network.

> the only option is to make GPG transparent by getting the
> email providers to automatically create key pairs and automatically
> handle signing and encryption by integrating their mail services with
> GPG behind the scenes.

No, this is a non option.
- Free providers are out for their own, like govts, they have no
real interest in their users going dark.
- Paid providers are a bit more loyal, but building that
and making statements about it entails risk of suit for failure
such that they're not really inclined. Most of them can't
even put postfix, dovecot and openldap together reliably and
scalably, let alone functional spam control and peering.
- You're still trusting your keys to the provider, and that the
provider won't be forced to jack you through your browser or
whatever cool software they give you. In today's world, that is
an abject failure.

As before:
Traditional centralized messaging services are a failure.  There
is NOTHING you can do to fix them but to replace them.  That is
where you have the chance to integrate crypto and defenses.  And the
only valid thing to replace them with now against the risk of centralization
and all but godlike attack is a new anonymous encrypted P2P messaging
transport system that scales [1].  You could create a new UI on all
the platforms, and for faster adoption encapsulate existing message
format within and extend existing clients to deal with the cryptographic
addressbooking and keep it Unix and business needs pipeliney. You
could include storage and nym registration. But at minimum you need
to shuffle messages between cryptographic endpoints within the net.

Anyways, you're wasting your time talking about PGPifying webmail,
desnooping and delogging SMTP, and poor little Johnny.
Forget about that tired old refit shit and build something new.
Put a little bling and sound effects in it, call it an IM, it worked
for ICQ and Candy Crush. The users will come, trust me.

Do it soon so you can try to head off bans with your newfound
embedded ubiquity.


[1] We have a handful of researched and implemented systems now.
But don't yet have enough knowledge, experience and wisdom
to determine what really should go into one that will remain valid for
the next 30 years of relatively stable internet API as we know it [2].
There's a good rate of people deploying new p2p, crypto, anti-metadata,
attack resistant, network tech such that by 2020 we may have explored
enough areas to begin to feel confident in assembling a handful of
them into such a system... even into a number of systems... messaging,
storage, info / data distribution, payments. But it will take all
of us trying them out and contributing to do it.

[2] Packet switched IPv6 routed global end to end connectivity.
https://en.wikipedia.org/wiki/Internet
https://en.wikipedia.org/wiki/History_of_the_Internet