Integrated crypto sounds useful, but it's fragile and ultimately a lose

Tim May tcmay at got.net
Sun Nov 3 12:41:11 PST 2002


On Sunday, November 3, 2002, at 12:19  PM, Tim May wrote:
> As with the situation a decade ago, there are:
>
> * several OSes in use (2-3 in Wintel world, 2 in Mac world, plus
> outliers)
> * various release versions of each
> * about 5-8 major mail programs covering these platforms
> * about 3-5 major newsreader programs

And I forgot to mention Linux...

Anyway, this is "tower of Babel" situation we have always faced with 
trying to tightly integrate crypto with apps and OSes.

> Several times over the past decade I have heard people urge others to
> change their mailer to one that is supported.
>
> This is even worse than "not one-click operation," as it asks users to
> abandon programs and OSes they like or need in order to obtain a
> marginal gain of sending a receiving encrypted messages with one click.


To expand on this point a bit, I suspect one of the main reasons people 
who once used PGP stop using it, either privately or at corporations 
(as we have heard folks here testify about), is because something 
changes and things "break."

They upgrade their OS, they get a new release of a mailer, and things 
break. And they don't have the time, energy, or inclination to track 
down all of the little gotchas that may have cause things to break. I 
know this happened to me several times over the years with various 
versions of PGP, Eudora, and Mac OS 7, 8, and 9.

And I expect that if and when I upgrade my OS, or Mail program, and PGP 
breaks, I'll be without PGP until it all becomes straightforward again.

Expecting people to use mailer and OSes other than the ones they 
already use just so they can get a bunch of "Hey, isn't this PGP rilly 
cool?" unsolicited messages is silly.

(Which is why, to harp on it, I would favor a very clean text-only 
approach. Then there would be a slight amount more work needed, but not 
the breakage we see.)





More information about the cypherpunks-legacy mailing list