Here's a document which explains why it's not a trivial problem to
solve, and suggests a way of solving it:
https://wiki.mozilla.org/Discussion_Forums/Proposal
Comments very welcome.
Gerv
Us folks that have always accessed the groups through NNTP. how does
this plan work for us? Seems like I remember having to request to sign
up for groups back when when we started with Netscape Navigator 3.0.1.,
at netscape.
oh BTW: I notice the date of the wiki is 00:30 July 15, 2010. WE have
time traveled into the future. <grin>
--
Phillip M. Jones, C.E.T. "If it's Fixed, Don't Break it"
http://www.phillipmjones.net mailto:pjo...@kimbanet.com
I'd just say the John Resig's comments aren't really applicable to the
mozilla groups I read. I moderate the Firebug google group and while
John's comments are correct, they only really bad if you have high
traffic. Sure they are annoying but the spam on the mozilla newsgroups
is as bad or worse. At the traffic rate you have I don't think it would
be so hard to use moderated groups.
On the other hand, google groups are horribly slow to read compared to
Thunderbird with newsgroups. So even if Resig didn't write one of the
great diatribes of all time, I would still hope you don't use Google groups.
jjb
From that page:
Groups which, by their nature, have a regular influx of new posters
(e.g. support groups) might have trouble. What we do is an open
question - we might use good moderator coverage, or we might exclude
them from this (trading off spam for easy access).
I'd think that blocking new (legit) posters in any of the forums might
be a problem. How would the case-by-case whitelisting happen -- would
somebody have to monitor spamassassin's logs picking out false
positives?
I assume that spamassassin could be configured to let everything
posted via NNTP and ML through, only taking a hard look at the stuff
that comes from Google, since that's the source of all the spam.
Would it be possible to set the proposed system up on some groups and
temporarily configure spamassassin to pass everything through but
generate logs about what would have been blocked?
--
»Q« /"\
ASCII Ribbon Campaign \ /
against html e-mail X
<http://www.asciiribbon.org/> / \
That's true on one hand, and on the other, anyone making his
communication dependent on Google is doomed.
Robert Kaiser
--
Note that any statements of mine - no matter how passionate - are never
meant to be offensive but very often as food for thought or possible
arguments that we as a community needs answers to. And most of the time,
I even appreciate irony and fun! :)
Add "moderator time" to the list of required resources?
BTW, the way W3C deals with this is by sending a message to
every new poster explaining that their post is about to be
publicly archived. They have the choice of approving that
particular post, approving all future posts from that email
address, or letting the post expire and finding some other
route for feedback. Asking permission like this has the
nice side-effect of blocking most spam from the W3C lists.
~fantasai
You will need a one-time approval to post, then things will continue
much as before.
Gerv
I naturally look at it from the perspective of support newsgroups, which
is a different animal than the rest of the groups:
* Almost everyone who starts a new thread is a first time poster, and
will usually not take part in other threads.
* Because there are other (better) web-based support forums
(support.mozilla.com, getsatisfaction, mozillazine, etc.), I don't see
web-access as mandatory for the support newsgroups.
So my initial thought is that making the Google Groups end read-only
would be a better option for the support newsgroups.
However,
We still get the odd spam posted via NNTP (Subject includes "IVÁN"). And
there are instances where mailing list members respond to posts that
have been removed from the news server.
So your proposal may be the better solution. We would need a team of
moderators to make sure new questions get posted with as little latency
as possible.
When the lists were first set up, I had to deal with a lot of users
thinking the list address was a private support address. They would post
messages without subscribing to the list. The solution was to
auto-reject messages from non-members, with rejection message explaining
what the list was for. If the newsgroup is moderated and going through
mailman, auto-rejecting messages from non-members would have to be
turned off. Is there a way to make messages sent to the list from
non-members automatically rejected /and/ messages from the news feed
held for moderation?
My rough comments from my read when this was first posted.
We should allow _all_ posts through after being hit against SpamAssasin
checks/bounds. As we currently do not have any such "whitelist"
restriction on the groups as they stand now, we can probably tweak the
settings to try and get the lowest possible false-positive.
Alternatively, we can use the whitelist approach in conjunction with the
above, and let the whitelist have the lowest-possible-spam-assasin
scoring, while anyone who did not elicit themselves to be white-listed
will get a slightly more aggressive check.
This accounts for:
"Anyone can post" (NNTP, today).
-- currently lists need a subscription first.
"No barrier for entry" (Support groups)
-- Support group users need not jump through hoops to get their post
read/seen, provided their post doesn't look like spam; (We can probably
tweak spamassasin settings to also auto-reply, "We have detected your
post as potentially being spam, if you are not a spammer, visit: <url>
to get your message marked for priority approval. If you intend to do
frequent posting to our lists you can also <link>file a bug</link> to
get added to our whitelist" or some such.
"Ease of understanding"
-- Why didn't my post show up, when I posted 5 hours ago, but
Callek's showed up already posted 10 min ago.
And probably easier on the moderator team too!
Gerv, I brought this up in passing on IRC; but I think I am forgetting
an idea I had then, if you recall it please help refresh my memory :-)
--
~Justin Wood (Callek)
As a side note against the usefulness of poster whitelists, I get tons
of spam on my mozilla accounts claiming to be sent by other mozillians.
I'm not sure if there's that much of a win in terms of spam trapping if
we whitelist the folks we know.
Axel
What I understood by whitelist, is more of a tiered catch-system. Where
whitelist is required for the post to even be considered for posting and
sent off to the main-part of SpamAssasin. I might have read too much
into it though
--
~Justin Wood (Callek)
I understood it differently, with non-whitelisted posts being scanned
by spamassassin and whitelisted posts bypassing that spamassassin, so
that the only problem would occur when a non-whitelisted post also got a
false positive from spamassasin.
A flowchart might help visualize the current proposal(s).
Unfortunately, that means that people who try and post on GG seem to
succeed, but their post gets dropped in the bit-bucket. This is not a
good user experience :-(
> So your proposal may be the better solution. We would need a team of
> moderators to make sure new questions get posted with as little latency
> as possible.
We would. :-|
(This is not a full response to your message, I know.)
Gerv
Dave implies that even so, we would get significant spam if we used this
approach.
> Alternatively, we can use the whitelist approach in conjunction with the
> above, and let the whitelist have the lowest-possible-spam-assasin
> scoring, while anyone who did not elicit themselves to be white-listed
> will get a slightly more aggressive check.
That may well be the way things combine. I have asked him to clarify here.
> -- Support group users need not jump through hoops to get their post
> read/seen, provided their post doesn't look like spam; (We can probably
> tweak spamassasin settings to also auto-reply, "We have detected your
> post as potentially being spam, if you are not a spammer, visit: <url>
> to get your message marked for priority approval. If you intend to do
> frequent posting to our lists you can also <link>file a bug</link> to
> get added to our whitelist" or some such.
Apparently auto-responding to spam can get you labelled as a spammer,
because of joe jobbing (forged From).
Gerv
> I'd think that blocking new (legit) posters in any of the forums might
> be a problem. How would the case-by-case whitelisting happen -- would
> somebody have to monitor spamassassin's logs picking out false
> positives?
Pretty much.
> I assume that spamassassin could be configured to let everything
> posted via NNTP and ML through, only taking a hard look at the stuff
> that comes from Google, since that's the source of all the spam.
You know, I think we could actually do that, that's not a bad idea. I
know the anti-spam stuff in Mailman (the built-in stuff, not what it
shells out to SpamAssassin for) can be configured to hold messages
based on headers, so if we can figure out a header pattern that's
always present in stuff from we could set it to automatically hold for
moderation anything originating from Google Groups, and then leave it
otherwise open.
--
Dave Miller
Systems Administrator, Mozilla Corporation
> What I understood by whitelist, is more of a tiered catch-system. Where
> whitelist is required for the post to even be considered for posting and
> sent off to the main-part of SpamAssasin. I might have read too much
> into it though
Callek is correct here. Everything will go through SpamAssassin, even
if it comes from a known email address. If you're not on the
whitelist, it would just get automatically held for the moderator,
regardless of what SpamAssassin had to say about it.
Actually, people who try to post via Google Groups would not see a way
to post. For example, the mozilla.support.general newsgroup was removed
and is now read-only on Google:
<http://groups.google.com/group/mozilla.support.general>.
Being removed is not the same as being set to read-only.
Unless Dave contradicts me, I am fairly sure I remember him telling me
about this problem with GG. Unless you know of a counter-example (a
group which is still live, but is mirrored read-only to GG, and where
users do not get a "post" UI)?
Gerv
That's how all Mozilla newsgroups looked before posting through Google
Groups was enabled. There was even a bug filed because the posting UI
appeared on a couple of support groups.
https://bugzilla.mozilla.org/show_bug.cgi?id=326634
I wonder, it has only seemed to get worse lately... can we move forward
with *some* form of this?
--
~Justin Wood (Callek)
On support-firefox and support-thunderbird, I've set
generic_nonmember_action to hold rather than reject. I'll see if we are
still getting private support requests. If we're not, I think we try
"spam filter --> auto-approve" for the support newsgroups.
We're still getting private support requests. Many more to the
thunderbird support list, which makes me think it may have something to
do with how they find the list address.
Maybe we can have an auto-response for first-time posters, which
explains that the list is a not a private support address and give other
introductory info like rules of etiquette.
I haven't seen any objections to this plan for the support newsgroups,
so if there are no responses today, I'll post the proposal in the
support newsgroups for feedback.