Path: utzoo!attcan!uunet!husc6!think!ames!pasteur!agate!garnet!weemba
From: weemba@garnet.berkeley.edu (Obnoxious Math Grad Student)
Newsgroups: news.misc
Subject: Re: Decrypting the P&W Digest Etc.
Message-ID: <14105@agate.BERKELEY.EDU>
Date: 10 Sep 88 11:49:21 GMT
References: <13971@agate.BERKELEY.EDU> <2015@looking.UUCP>
Sender: usenet@agate.BERKELEY.EDU
Reply-To: weemba@garnet.berkeley.edu (Obnoxious Math Grad Student)
Organization: Brahms Gang Posting Central
Lines: 170
In-reply-to: brad@looking.UUCP (Brad Templeton)

Note: I am not sure if I read Brad's article correctly.  If I seem to
be overreacting, I apologize to Brad in advance.  Consider what follows
an argument with a strawman, who may or may not be Brad.

In article <2015@looking.UUCP>, brad@looking (Brad Templeton) writes:
>One should not ever allow automatic decryption of rotated articles.
>That defeats the whole point!

>The whole idea behind this scheme is that people must take a special,
>explicit action to see the article.  Each and every article.

I disagree.  I think this should be up to each and every reader to de-
cide for him/herself.  The default should of course require explicit
action.

>							       If the
>action need only be taken once, you could claim "subscribing to alt.sex"
>was the action, but you would not get very far.

Subscribing to alt.sex will not automatically decrypt the P&W Digest
in Gnews.  The user must install the following hook-kills for alt.sex:

  (pre nil setq article-digest-maybe nil)
  (pre nil gnews-set 'article-header-hook
       '(lambda nil
	  (if (string-match "pwdigest@alembic" (article-field-raw "From"))
	      (progn
		(setq n-reply-digest-mail t)
		(article-digest 'article-digest-maybe nil t))
	    (mapcar 'hook-kill-do article-digest-maybe)
	    (setq n-reply-digest-mail nil)
	    (setq article-digest-maybe nil))))
  (pre nil gnews-set 'group-last-hook
       '(lambda nil
	  (mapcar 'hook-kill-do article-digest-maybe)
	  (setq n-reply-digest-mail nil)
	  (setq article-digest-maybe nil)))
  (post nil mapcar 'hook-kill-do article-digest-maybe)
  (post nil setq n-reply-digest-mail nil)

This is highly non-trivial stuff!  There is no Gnews command to in-
sert the above set of commands automatically, nor will I write one.
(This goes ditto for the TeXhax and Neuron digest.  The number of
subsidiary concerns needed to automate the above seems too large
to plan for ahead.  For one thing, I have no idea of how many digests
are gatewayed into unmoderated newsgroups)  Someone who does the
above has clearly decided that he or she truly wishes to read the
P&W Digest.

Moreover, because of the complexities of dedigestification, it would
require much extra effort on my part to make an explicit "x" rot13ing
as opposed to the automatic rot13ing coded for in that last "t" argu-
ment to the article-digest command above, work smoothly with my Gnews
Digest mode.  If I had to, I'd do the work, but since I don't, I won't,
and I resent anyone suggesting that I "ought" to do this extra work.

Indeed, there was one detail while I was debugging the code that was
most simply solved by asking the moderator to include a trailing "End
of P&W Digest" message at the end of his digest, like all the other
digests use.  I asked the moderator to make this change, and he did
so.  The one question was whether this should be rot13ed or not: I
suggested yes, since this will work best for those without a dedi-
gestifier.  Gnews works with both (or any other message, if needed.)

The Gnews manual gives the explicit code needed for the TeXhax digest.
It mentions the small change that must be made for a rot13-ed digest.
It is left to the individual user to take action.

It is true that the Gnews user does not have to do a lot of typing.
The extended help command, invoked while reading a digest, takes
one to the Info node on digests.  From there, one can quickly find
the TeXhax code and yank it out.  Then one quits out of Info, starts
the Hook Kill Edit mode, in yank back in the TeXhax digest hooks.  Then
one makes a few quick changes, and one is all set up for next time.

But the user still has to take this effort.

And if the P&W Digest moderator wants to write an Emacs Lisp command
to insert the correct code automatically, and include it with his
monthly-or-so rot13 information, well, why not?  The Gnews user will
still have to move the cursor to the right spot in the article, eval
the defun, and then run the command.

(OK, if the moderator further puts a ^L right after the defun, then
Gnews will only display up to the formfeed, sparing the user from
having to do the first step above.  Wow wow wow.)

>						  If you don't have to take
>it at all, it's the same as calling the data compression done during the
>transmission "encryption."

Huh?  Please note the following points:

(*) rot13ing is not *really* encryption in the first place.  At least,
    not in any classical sense of the word.
(*) Compression is compression.  Calling it encryption won't make it
    so.  (Ditto rot13.)
(*) Compression/decompression of news batching serves an entirely diff-
    erent purpose: to save money and time.

Rot13ing exists for one reason only: to reduce the chances of someone
being accidently offended.

>Somebody once told me that their newsreader did an automatic decrypt on
>any article with "rot13" in the title,

I agree with you that no newsreader should ever do this BY DEFAULT.  If
this is all you had argued against, I would have merely posted a brief
article saying that you misunderstood my cryptic description of what
Gnews provides.

>				        and so would I please put that in
>the title of encrypted rec.humor.funny postings.

This somebody has crappy software, if they cannot specify the automatic
decrypt on the "Keywords: rot13" field that you *do* include!  In Gnews
2.0, hooks for autodecryption is fairly trivial to implement.  The only
question would be what would trigger it.  I will leave that for the user
to decide.

>My response was not what he wished.  I now delete rot13 if it's in the
>title.  If somebody starts putting in automatic rot13 detectors, I will
>have to switch away from rot13, no matter what the complaints.

[This is the part where I'm not sure where you stand: are you saying
auto-rot13-ing should not become the default for a newsreader, or that
it should not exist period?  Everyone please remember that I am arguing
with someone who believes the latter, and that this someone may or may
not be Brad.]

I was about to use your newsgroup an example of the new autodecryption
software hooks!  Oh well...  Now I'll have to think about whether to put
them in as what *ought* to be the canonical example, and then include
a mini-editorial about certain moderators.  I hope I'm misreading you!

Anyway, you do know that it is fairly trivial to implement a chi-square
test for English letter frequencies: if chi-square comes out abnormal,
apply whatever brad13 technique is necessary.

And if you use a permutation algorithm, even something as simple as
printing reversed lines, that too, can be tested for rather rapidly
and accurately.  You aren't going to keep your method a secret, are
you???

The main driving philosophy behind Gnews has always been that the user
should be able to customize *extensively* and, whenever possible, *very
simply*.  As he/she *wishes* to read news.  This is very easy to do in
Emacs: all internal variables are accessible to the user, so by merely
following good programming practice and not using embedded constants,
feature after feature is trivially customizable.  The current draft of
the manual has about 35 pages devoted to examples and explanations of
customization, and more are forthcoming.

Moderators whose attitude is that his readers are not able to make such
customization decisions for themselves about *HOW* they want to read his
newsgroup make me sick.  I didn't do all this customization effort be-
cause I believe I know the holy truth about how people want to or ought
to read news.  Not even close.

Perhaps it's time for rec.humor.funny.funnier, whereby weemba culls the
best from rec.humor.funny and reposts them?  I guarantee that I would
repost every last brad13ed joke you post, in rot13ed format, with ROT13
in the "Subject:" line.  Then the somebody who complained would be able
to read through my group instead.  I would also include an extra header
that mentions the Message-ID of the articles I repost, and write extra
software to set up a KILL for that Message-ID, so that if the user reads
r.h.f.f well before reading r.h.f, then the user will not be bothered
with your brad13ing efforts in the least.

ucbvax!garnet!weemba	Matthew P Wiener/Brahms Gang/Berkeley CA 94720