Re: Address munging

Dmitry writes:

> Adam Sjøgren <[email protected]> wrote:

>> Dmitry writes:

>>> These include some pretty old and pretty alive mailing lists, which
>>> have paradoxically been hanging in a state that is worse than being
>>> completely absent from Gmane.
>>
>> Except that you can't reply directly to authors
>
> List addresses are affected as well, do not they?

All text that looks like an email address is, as far as I know - what
difference does it make if the list address is?

You don't want to write to the mailing list via Gmane?

>> the groups are perfectly usable, right?
>
> By posting via Gmane NNTP gateway, you mean?

Yes. That's one of the main features of Gmane, I think?

> Unless you are (1) use a domain name you control for your mail address
> and (2) either (a) old-fashioned enough to configure you mail server
> not to sign outgoing mail or (b) tech-savvy enough to implement
> client-side DKIM for your NNTP client (I am not aware of any, that
> have that feature out of a box), whatever you post will systematically
> tend to end up in junk folders of subscribers, if not rejected
> outright.

Uhm, how is this related to Gmane's address encryption? SPF will fail
either way.

(If the adress encryption used a gmane.io-address, it would even _help_,
because then SPF wouldn't fail. Alas, it doesn't.)

> I believe, you are aware of that sma-all refinement of ‘perfectly
> usable’, since you yourself seem to belong to the category 2b, do not
> you? :-).

(I just set up Gnus to add DKIM-signature:-headers to my outgoing news
articles the other day; I don't see how any NNTP client could do that
out of the box, as you need to add a public key to DNS for DKIM to
work.)

I fail to see how address encryption makes the whole SPF/DKIM/DMARC
thing any worse, regardless of which of your categories one falls in.

And I don't see how being able to read the mailing list with encrypted
addresses is worse than not having the mailing list at all.

>> I believe it was because of worries that spammers would be
>> harvesting email-addresses in a central place.
>
> It was, sure.  But is it still a relevant worry?

I think it was a relevant worry, and I now think Gmane would have been
better off not carrying lists that wanted address encryption, rather
than implementing it - but I do not think that was obvious at the time
it was done.

> Do I recall correctly, that due to the very same concern even bodies
> of messages was munged once, but finally that have been lifted?

Do you mean that bodies used to be encrypted? I don't remember seeing
that. How would that work on clients?

I'm think the address encryption is everywhere in the article? An
example: gmane.comp.boot-loaders.grub.user:5256 which is quite new.

>> Some mailing lists would say no to Gmane carrying their list without
>> the encryption.
>
>> If the mailing list owner has agreed to have the list on Gmane on
>> the basis of encryption being on, it would be rude of Gmane to just
>> turn it off, right?
>
> That is, there are such lists?

It is impossible to automatically tell whether the list owner accepted
Gmane's subscription because encryption was turned on, or if the list
owner would be fine with it turned off.

And when you don't know that, you can't just turn it off and expect no
list owner to be unhappy - and rightfully so.

> However, many (most?) of groups in question are processed for the
> reasons that for sure have nothing to do with the wishes of their
> listmasters, since they are publicly archived elsewhere, and first of
> all at the place of their origin.

If you say so.

I can't read minds, so I don't know if list owners that are fine with
having unencrypted email-addresses in one archive are also happy to have
them unencrypted on Gmane.

Sometimes people are more lax about things they control themselves.

> E. g., full and uncrippled archives of gnu.org, nongnu.org and
> libreplanet.org are easily available via HTTP at [1], yet, for
> example, [email protected], [email protected] and
> [email protected] are affected.

Sure. And you can tell from that what the list owners think?

> I guess, theyʼd picked up a ‘conceal addresses’ flag only because few
> years ago a subscription form featured it turned on by default,

Looking at the page in the Wayback Machine the default seemed to be no
encryption until somewhere between August 10 and August 14, 2006 and it
stayed like that until at least July 8, 2016.

Anyway, we can't know if it was a conscious decision on the subscribers
part, and we can't know if it factored into the list owners decision.


  Best regards,

    Adam

-- 
 "There's no such thing as paranoia                         Adam Sjøgren
  It's always twice as bad as you fear"                [email protected]