Just a thought experiment:
if you could/had to make something like a NNTP 2.0 (with no need for
backwards compatibility) and server and client software for it today, what
would it be like?
In terms of specifications, technologies used, user interface, etc.

George Musk wrote:
>
> Just a thought experiment:
> if you could/had to make something like a NNTP 2.0 (with no need for
> backwards compatibility) and server and client software for it today, what
> would it be like?
> In terms of specifications, technologies used, user interface, etc.

Some of the problems with the current Usenet:

1) Transport protocol
NNTP (RFC 3977) does not support notification if something interesting
becomes available.

2) Transported data
The article format (RFC 5536) is very complicated.
But it supports embedding audio and video (with MIME).
And flowed text is possible (RFC 3676).

3) User interface
User interfaces for mobile devices are rare.

4) General
Moderation is absent or complicated.

Because progress with anything related to Usenet is extremely slow,
it is likely not a good idea to come up with something incompatible.

Incompatible protocols like ActivityPub already exist.
They have not replaced Usenet (and likely will not do in the future).

[Xpost reduced]

On Wed, 7 Aug 2024 14:32:01 -0000 (UTC), George Musk <grgmusk@skiff.com> wrote:
>Just a thought experiment:
>if you could/had to make something like a NNTP 2.0 (with no need for
>backwards compatibility) and server and client software for it today, what
>would it be like?
>In terms of specifications, technologies used, user interface, etc.

don't know? this seems like a question for nntp server administrators,
nntp software programmers, professional and amateur usenet veterans...

https://www.eyrie.org/~eagle/software/inn/
>...
>news.software.nntp

i'm still using 40tude dialog with hamster scoring; this oldie still
works better than anything else out there; e.g., scoring cross-posts:

[*]
=-100 Xpost %>1
=-200 Xpost %>2
=-300 Xpost %>3
= etc.

most newsreaders provide some means of scoring or filtering articles
in newsgroups by analyzing message header fields, e.g. from, subject,
message-id, references, injection-info, xpost, path, etc., and where
text strings are found matching that newsreader's user-defined rules,
scoring and actions are applied; 40tude dialog's help "scoring rules":

40tude Dialog > Help[F1] > Index > Scoring Rules > Scoring/Actions Syntax
>Scoring and actions is one of Dialog's most powerful features. The syntax
>and actually most of the code is from Jurgen Haible's excellent local mail-
>and newsserver Hamster.
>What is Scoring?
>Scoring is the process of assigning a number between -9999 and +9999 to a
>message by applying scoring rules to the message. A scoring rule usually
>analyzes one header field of the message and if it this header field matches
>a certain text, a score value is assigned. For example, you can create a
>scoring rule that assign the highest score value +9999 to all messages that
>have your email address in the From header field, so that your messages
>always receive a score of +9999.
>The score value is shown in the header list in green for positive scores
>and in red for negative scores:
>Note that the list of headers can be sorted by score (with or without
>threading), so scoring can be used to visually organize and separate
>important from unimportant messages.
>...
>Usenet articles are scored twice in Dialog. When you get headers in a
>group the scoring rules are applied to the available, limited number of
>headers, however when you retrieve the complete body of the message, the
>message is scored again and this time all headers can be scored.
>The scoring and action rules are stored together in one file, the score
>file, which you can access by selecting Settings, Scoring and actions
>from Dialog's main menu.
[end quoted excerpt]

40tude Dialog v2.0.15.1 (2005-2-7):
https://www.barghahn-online.de/4td_faq/download.php
https://www.barghahn-online.de/4td_faq/download/4d2b38.exe

In comp.misc George Musk <grgmusk@skiff.com> wrote:
> Just a thought experiment:
> if you could/had to make something like a NNTP 2.0 (with no need for
> backwards compatibility) and server and client software for it today, what
> would it be like?

NNTP has compression support now, though I'm not using it because
the server I use doesn't support it. That's about the only thing
I'd want, and I don't actually care much that I've almost got it.
All my Usenet experience is really missing is activity in all the
forgotten groups I'm subscribed to, but I can't see a new protocol
bringing that back if it maintained many features I like from NNTP.
Protocols/platforms that don't have those features are already
plentiful and successful, my only ambition would be to bridge to
them (or the vaguely similar ones, like Web forums) over the
existing NNTP protocol.

> In terms of specifications, technologies used, user interface, etc.

This may be extremely pessimistic, but I now automatically expect
any new discussion protocol/platform to be worse than NNTP/Usenet
and not worth looking at. I'd just fear that any fragmentation due
to an attempted incompatible NNTP replacement would only kill off
the little activity still going on with Usenet.

--
__ __
#_ < |\| |< _#

On Wed, 7 Aug 2024 14:32:01 -0000 (UTC), George Musk wrote:

> if you could/had to make something like a NNTP 2.0 (with no need for
> backwards compatibility) and server and client software for it today, what
> would it be like?

You could just use something like Mastodon, and move to the Fediverse.

On 8/7/24 20:29, Lawrence D'Oliveiro wrote:
> You could just use something like Mastodon, and move to the Fediverse.

Mastodon / Fediverse isn't feature compatible. Not the least of which
is that Usenet servers are local copies of things posted to servers.

Servers exchange articles in a flood-fill style.

The ability to do end-to-end communications without an end-to-end link
is wonderful.

Store-and-forward networking has some advantages that many people don't
take into account.

It's relatively easy to link two disconnected NNTP networks with
something like a mag' tape in a station wagon. You can't do that with
very many networking technologies.

--
Grant. . . .

On Thu, 8 Aug 2024 01:29:26 -0000 (UTC), Lawrence D'Oliveiro <ldo@nz.invalid> wrote:
>You could just use something like Mastodon, and move to the Fediverse.

https://duckduckgo.com/?q=pentagon+fediverse+logo

"Speak. Great power to control... dominate. Speak. Impressive. They can make
planets. Oh, yes. New cities, homes in the country, your woman at your side,
children playing at your feet, and overhead, fluttering in the breeze, the
flag of the Federation."--Kruge, Star Trek III: The Search for Spock(c)1984

On Wed, 7 Aug 2024 21:52:14 -0500, Grant Taylor wrote:

> On 8/7/24 20:29, Lawrence D'Oliveiro wrote:
>
>> You could just use something like Mastodon, and move to the Fediverse.
>
> Mastodon / Fediverse isn't feature compatible.

It isn’t limitation-compatible or bug-compatible either. It's a complete
rethinking of the way distributed social media is supposed to work.

On 2024-08-08, Grant Taylor wrote:
> On 8/7/24 20:29, Lawrence D'Oliveiro wrote:
>> You could just use something like Mastodon, and move to the Fediverse.
>
> Mastodon / Fediverse isn't feature compatible. Not the least of which
> is that Usenet servers are local copies of things posted to servers.

Every time I look at it, I get the feeling that mastodon/fediverse is a
half thought-out "hey wouldn't it be cool if ... " type thing that only
came about because people think "The Web" is synonymous with "The
Internet".

Or maybe it's just the marketing-speak on their websites ...

> [...]
> It's relatively easy to link two disconnected NNTP networks with
> something like a mag' tape in a station wagon. You can't do that with
> very many networking technologies.

Never underestimate the bandwidth of a station wagon full of tape
hurtling down the freeway :)

(or however that went)

--
|_|O|_|
|_|_|O| Github: https://github.com/dpurgert
|O|O|O| PGP: DDAB 23FB 19FA 7D85 1CC1 E067 6D65 70E5 4CE7 2860

On Thu, 8 Aug 2024 07:32:28 -0000 (UTC), Lawrence D'Oliveiro <ldo@nz.invalid> wrote:
>On Wed, 7 Aug 2024 21:52:14 -0500, Grant Taylor wrote:
>> On 8/7/24 20:29, Lawrence D'Oliveiro wrote:
>>> You could just use something like Mastodon, and move to the Fediverse.
>>
>> Mastodon / Fediverse isn't feature compatible.
>
>It isn't limitation-compatible or bug-compatible either. It's a complete
>rethinking of the way distributed social media is supposed to work.

social media is moderated; unmoderated newsgroups are unmoderated

On Thu, 8 Aug 2024 09:39:00 -0000 (UTC), Dan Purgert <dan@djph.net> wrote:
>On 2024-08-08, Grant Taylor wrote:
>> On 8/7/24 20:29, Lawrence D'Oliveiro wrote:
>>> You could just use something like Mastodon, and move to the Fediverse.
>> Mastodon / Fediverse isn't feature compatible. Not the least of which
>> is that Usenet servers are local copies of things posted to servers.
>Every time I look at it, I get the feeling that mastodon/fediverse is a
>half thought-out "hey wouldn't it be cool if ... " type thing that only
>came about because people think "The Web" is synonymous with "The
>Internet".
>Or maybe it's just the marketing-speak on their websites ...
>> [...]
>> It's relatively easy to link two disconnected NNTP networks with
>> something like a mag' tape in a station wagon. You can't do that with
>> very many networking technologies.
>
>Never underestimate the bandwidth of a station wagon full of tape
>hurtling down the freeway :)
>(or however that went)

uncensored plain text communications made public en masse by means
of unmoderated usenet nntp newsgroups is relatively primitive like
a well-seasoned cast iron skillet...if it ain't broke don't fix it

On Thu, 8 Aug 2024 13:41:35 +0100, D wrote:

> On Thu, 8 Aug 2024 07:32:28 -0000 (UTC), Lawrence D'Oliveiro
> <ldo@nz.invalid> wrote:
>
>>On Wed, 7 Aug 2024 21:52:14 -0500, Grant Taylor wrote:
>>> On 8/7/24 20:29, Lawrence D'Oliveiro wrote:
>>>
>>>> You could just use something like Mastodon, and move to the
>>>> Fediverse.
>>>
>>> Mastodon / Fediverse isn't feature compatible.
>>
>>It isn't limitation-compatible or bug-compatible either. It's a complete
>>rethinking of the way distributed social media is supposed to work.
>
> social media is moderated; unmoderated newsgroups are unmoderated

You mean “moderated social media is moderated”?

George Musk <grgmusk@skiff.com> writes:
> Just a thought experiment:
> if you could/had to make something like a NNTP 2.0 (with no need for
> backwards compatibility) and server and client software for it today, what
> would it be like?
> In terms of specifications, technologies used, user interface, etc.

* Machine-readable format specification to reduce ambiguity.
* Binary serialization rather than the complex text-based structures we
have in NNTP and the Usenet article format.
* UTF-8 for all text.
* Fixed-sized message IDs, e.g. a 256-bit random value.
* Some kind of early reject for streamed messages, perhaps by sending
metadata separately from context (analogous to separating heade and
body in NNTP) or a more general way to signal rejection at any point
in message transmission. (To minimize transmission cost of spam.)
* Standardized approach to markup and images (with per-group policies on
what’s acceptable).
* Security as standard, meaning, at least...
* Transport is always encrypted (probably meaning TLS)
* All messages signed by author and originating server (supporting
reputation management)
* Hierarchy and group administration cryptographically authenticated
* Submissions to (equivalent of) moderated groups encrypted to group
administration and transmitted over the regular network, i.e. no
dependence on email.

--
https://www.greenend.org.uk/rjk/

Richard Kettlewell <invalid@invalid.invalid> wrote or quoted:
>* Binary serialization rather than the complex text-based structures we
> have in NNTP and the Usenet article format.

When I read through all this stuff, it seems like every change
has just as many pros as cons. Take the point above, for instance.
If I think my newsreader is acting up today or isn't showing a
certain feature (like grabbing stuff via Message-ID), I can just
hop on telnet to the news server and see what's really coming
from the server or call up the missing feature directly. With a
binary format, that wouldn’t be as straightforward!

Richard Kettlewell <invalid@invalid.invalid> wrote or quoted:
>* Machine-readable format specification to reduce ambiguity.

Wikipedia:

|The "second-system effect" or "second-system syndrome" is the
|tendency of small, elegant, and successful systems to be
|succeeded by over-engineered, bloated systems, due to
|inflated expectations and overconfidence.

, see also:

"Things You Should Never Do, Part I". (April 6, 2000) - Joel Spolsky

.

ram@zedat.fu-berlin.de (Stefan Ram) writes:
> Richard Kettlewell <invalid@invalid.invalid> wrote or quoted:
>>* Machine-readable format specification to reduce ambiguity.
>
> Wikipedia:
>
> |The "second-system effect" or "second-system syndrome" is the
> |tendency of small, elegant, and successful systems to be
> |succeeded by over-engineered, bloated systems, due to
> |inflated expectations and overconfidence.
>
> , see also:
>
> "Things You Should Never Do, Part I". (April 6, 2000) - Joel Spolsky

The premise was a new protocol, if you want to critique that then I
suggest responding to the OP instead.

If it’s machine-readable protocol specs you disagree with then you’re
just plain wrong. They’re an extremely successful strategy and ought to
be used even more widely than they already are.

--
https://www.greenend.org.uk/rjk/

Richard Kettlewell <invalid@invalid.invalid> wrote or quoted:
>The premise was a new protocol, if you want to critique that then I
>suggest responding to the OP instead.

I totally get where you're coming from.
I only wrote it like that because I thought you /were/ the OP!

>If it’s machine-readable protocol specs you disagree with then you’re
>just plain wrong. They’re an extremely successful strategy and ought to
>be used even more widely than they already are.

Here's the syntax from RFC 3977. It's already pretty machine-readable
and quite straightforward:

CAPABILITIES [keyword]
MODE READER
QUIT
LISTGROUP [group [range]]
LAST
ARTICLE message-id
ARTICLE number
ARTICLE
HEAD message-id
HEAD number
HEAD
BODY message-id
BODY number
BODY
STAT message-id
STAT number
STAT
POST
IHAVE message-id
DATE
HELP
NEWGROUPS date time [GMT]
NEWNEWS wildmat date time [GMT]
LIST [keyword [wildmat|argument]]
OVER message-id
OVER range
OVER
LIST OVERVIEW.FMT
HDR field message-id
HDR field range
HDR field
LIST HEADERS [MSGID|RANGE]
wildmat = wildmat-pattern *("," ["!"] wildmat-pattern)
wildmat-pattern = 1*wildmat-item
wildmat-item = wildmat-exact / wildmat-wild
wildmat-exact = %x22-29 / %x2B / %x2D-3E / %x40-5A / %x5E-7E /
UTF8-non-ascii ; exclude ! * , ? [ \ ]
wildmat-wild = "*" / "?"

. You can snag the message ID syntax from another RFC, and
filling in the blanks (like "keyword", "range" or "number")
should be a piece of cake.

ram@zedat.fu-berlin.de (Stefan Ram) writes:
> Richard Kettlewell <invalid@invalid.invalid> wrote or quoted:
>>If it’s machine-readable protocol specs you disagree with then you’re
>>just plain wrong. They’re an extremely successful strategy and ought to
>>be used even more widely than they already are.
>
> Here's the syntax from RFC 3977. It's already pretty machine-readable
> and quite straightforward:

I mean something like ASN.1 (although not ASN.1; far more complex than
necessary).

--
https://www.greenend.org.uk/rjk/

On Fri, 9 Aug 2024, Stefan Ram wrote:

> Richard Kettlewell <invalid@invalid.invalid> wrote or quoted:
>> * Binary serialization rather than the complex text-based structures we
>> have in NNTP and the Usenet article format.
>
> When I read through all this stuff, it seems like every change
> has just as many pros as cons. Take the point above, for instance.
> If I think my newsreader is acting up today or isn't showing a
> certain feature (like grabbing stuff via Message-ID), I can just
> hop on telnet to the news server and see what's really coming
> from the server or call up the missing feature directly. With a
> binary format, that wouldn’t be as straightforward!
>

Amen! The power of text!

On Fri, 9 Aug 2024, Stefan Ram wrote:

> Richard Kettlewell <invalid@invalid.invalid> wrote or quoted:
>> * Machine-readable format specification to reduce ambiguity.
>
> Wikipedia:
>
> |The "second-system effect" or "second-system syndrome" is the
> |tendency of small, elegant, and successful systems to be
> |succeeded by over-engineered, bloated systems, due to
> |inflated expectations and overconfidence.
>
> , see also:
>
> "Things You Should Never Do, Part I". (April 6, 2000) - Joel Spolsky
>
> .

Interesting. I read on an encryption mailinglist some critique of crypto
developers. They love to develop new things all the time, instead of
perfecting and learning how to use what we do have.

The argument was to not tinker, and learn how to use, instead of tinkering
and introducing new bugs and stuff.

Right or wrong, it was a nice discussion.

ram@zedat.fu-berlin.de (Stefan Ram) writes:

> If I think my newsreader is acting up today or isn't showing a
> certain feature (like grabbing stuff via Message-ID), I can just
> hop on telnet to the news server and see what's really coming
> from the server or call up the missing feature directly. With a
> binary format, that wouldn’t be as straightforward!

\o/ ____( rlwrap nc newsserver 119 )

--
I do not bite, I just want to play.

On Fri, 09 Aug 2024 19:13:00 +0100, Richard Kettlewell wrote:

> I mean something like ASN.1 ...

Bloody hell.

Where’s the garlic ...

Lawrence D'Oliveiro <ldo@nz.invalid> writes:
> On Fri, 09 Aug 2024 19:13:00 +0100, Richard Kettlewell wrote:
>
>> I mean something like ASN.1 ...
>
> Bloody hell.
>
> Where’s the garlic ...

Gentle reminder of the full text:

| I mean something like ASN.1 (although not ASN.1; far more complex than
^^^^^^^^^^^^^^^^^^
| necessary).

ASN.1 is given as an example because it’s a well-known example of an
interface definition language, not because I’m suggesting using it, as
anyone capable of reading to the end of a sentence can tell.

--
https://www.greenend.org.uk/rjk/

On Sat, 10 Aug 2024 09:14:05 +0100, Richard Kettlewell wrote:

> ASN.1 is given as an example because it’s a well-known example of an
> interface definition language, not because I’m suggesting using it, as
> anyone capable of reading to the end of a sentence can tell.

Maybe if you offered a better example, it would make a more coherent
point.

yeti <yeti@tilde.institute> wrote or quoted:
>ram@zedat.fu-berlin.de (Stefan Ram) writes:
>>If I think my newsreader is acting up today or isn't showing a
>>certain feature (like grabbing stuff via Message-ID), I can just
>>hop on telnet to the news server and see what's really coming
>>from the server or call up the missing feature directly. With a
>>binary format, that wouldn’t be as straightforward!
>\o/ ____( rlwrap nc newsserver 119 )

Thanks for the tip! I hadn't heard of "rlwrap" before.