moderated How on earth does Groups.io thread a topic?


Brian Vogel <britechguy@...>
 

I ask this because I'm exasperated and I just can't figure out how certain topic splits are occurring.

Today has been a big "inexplicable topic split" day on the Windows 10 for Screen Reader Users Group.

The topic using the refresh tool from microsoft to reinstall windows 10: is now split into three threads, all with titles that match as far as the human eye can see.  The same is true for help with outlook 2013 please! though it's only split into two separate threads.

All participants are, to my knowledge, using e-mail to interact with the group.  The fact that titles of the topics are precisely the same makes it utterly mystifying to me why they would ever not have all been assigned to a single topic/thread, particularly when they all come in on the same day.

Can someone explain to me how Groups.io threads e-mail messages into a topic and how these two examples came to be separated into multiple topics.  This is far from the first time, and almost certainly not the last, that this will occur.  It seems that manual intervention to merge things will be required, but I'd love to know why the software itself does not recognize what appears to be a single topic as a single topic.

--
Brian

I have made mistakes, but have never made the mistake of claiming I never made one.   

          ~  James G. Bennet


 

There are anomalies for sure. I think this has come up in the past and it has something to with headers, and in some cases, it's just impossible to get it right. In my group, there is one member whose posts always start a new topic, even when she doesn't do anything to change the thread title (she always posts via email, but so do many other members who don't create this problem). I am always having to go in and merge thread "my cat won't eat" with thread "my cat won't eat." Same exact titles. Every single post of this one member's starts a new thread.

This used to happen with more members than just the one, and Mark fixed some of them. Every time, he gets closer. It's like Achilles and the tortoise...

--
J

Messages are the sole opinion of the author. Especially the fishy ones.

I wish I could shut up, but I can't, and I won't. - Desmond Tutu


Brian Vogel <britechguy@...>
 

Well, this is insanity making on several of the groups I read regularly.  It happens with far more than a single user and, in fact, the first topic I made reference to in my initial query is now 4 separate threads.  There is no logical reason for this, as it should be simple to use subject/title and time frame of posts in some coherent way to prevent this.  This isn't rocket science, and anything with the same subject/title in an e-mail message posted within, say, 24 hours of the last thing posted with exactly that same subject/title is virtually certain to be a part of an ongoing topic.



Duane
 

On Tue, Jan 3, 2017 at 03:16 pm, Brian Vogel wrote:


virtually certain
Depends on the group. On a couple of my groups, it's not unusual to get posts with the same subject, but different threads, due to the nature of the groups.

There are several different things that can cause a thread to split. Most common is when someone changes the subject, even if they then change it back. Another common one is when someone copies the subject to a new message, so there's no "in-reply-to" information. One that some of my members run into is starting a reply, saving a draft, and finishing later. Their email program does something that loses the critical info to maintain the thread. All of the details are in the headers.

Duane


 

This happens in my group, with the one user at least (there may still be others, although fewer now then before for sure), with messages posted even within minutes of each. If I didn't go in and merge them together immediately, the threads would proliferate even further. 

J

On Tue, Jan 3, 2017 at 3:16 PM, Brian Vogel <britechguy@...> wrote:

Well, this is insanity making on several of the groups I read regularly.  It happens with far more than a single user and, in fact, the first topic I made reference to in my initial query is now 4 separate threads.  There is no logical reason for this, as it should be simple to use subject/title and time frame of posts in some coherent way to prevent this.  This isn't rocket science, and anything with the same subject/title in an e-mail message posted within, say, 24 hours of the last thing posted with exactly that same subject/title is virtually certain to be a part of an ongoing topic.




--
J

Messages are the sole opinion of the author. Especially the fishy ones.

I wish I could shut up, but I can't, and I won't. - Desmond Tutu


 

Duane,

I don't think this one user in my group is doing any of those things, but I will certainly ask her. I think there's a problem in general with the way her email interacts with the system. For example, the top-posting does not work in her case, either, and I frequently have to go in and trim the prior posts from her messages.

J

On Tue, Jan 3, 2017 at 3:38 PM, Duane <txpigeon@...> wrote:
On Tue, Jan  3, 2017 at 03:16 pm, Brian Vogel wrote:

>
> virtually certain
>

Depends on the group.  On a couple of my groups, it's not unusual to get posts with the same subject, but different threads, due to the nature of the groups.

There are several different things that can cause a thread to split.  Most common is when someone changes the subject, even if they then change it back.  Another common one is when someone copies the subject to a new message, so there's no "in-reply-to" information.  One that some of my members run into is starting a reply, saving a draft, and finishing later.  Their email program does something that loses the critical info to maintain the thread.  All of the details are in the headers.

Duane





--
J

Messages are the sole opinion of the author. Especially the fishy ones.

I wish I could shut up, but I can't, and I won't. - Desmond Tutu


Brian Vogel <britechguy@...>
 

On Tue, Jan 3, 2017 at 03:38 pm, Duane wrote:
On Tue, Jan 3, 2017 at 03:16 pm, Brian Vogel wrote:

virtually certain
Depends on the group. On a couple of my groups, it's not unusual to get posts with the same subject, but different threads, due to the nature of the groups.

 virtually (adj) - slightly short of or not quite accomplished; `near' is sometimes used informally for `nearly' and `most' is sometimes used informally for `almost' 

I realize there are cases where this might not be true, but I'd be willing to bet substantial sums of money that you'd be doing far less thread splitting were things to be lumped together strictly on subject and proximity in time than one now has to do merging because the threading is just so flaky when you have a subject match that was posted in rapid succession after another with the same subject.

I'll err on the side of accidentally keeping a thread as one that the user intended to be split (but didn't change the subject) than accidentally splitting a thread that is actually meant to be a single one when the subject matches, it comes on the heels of others, and other header information is missing.  Fuzzy logic is necessary here, and could be applied.

Life is the art of drawing sufficient conclusions from insufficient premises. ~ Samuel Butler, 1835-1902

The same is true of deciding appropriate threading when there are "insufficient premises."




 

Brian,

I ask this because I'm exasperated and I just can't figure out how
certain topic splits are occurring.
It depends on more than just the Subject. Here's the implementation:
https://groups.io/g/beta/message/5772

All participants are, to my knowledge, using e-mail to interact with
the group. The fact that titles of the topics are precisely the same
makes it utterly mystifying to me why they would ever not have all
been assigned to a single topic/thread, particularly when they all
come in on the same day.
The inference is that the later messages had a In-Reply-To (or References) header that cited a message other than one that matched the subject.

This can happen if members are in the habit of clicking Reply on an arbitrary message from the group, then replacing the Subject to match a thread in the group.

As a mod of the group (or a member receiving individual messages) one can use View Source to examine the Subject and any In-Reply-To or References fields in detail. This usually tells the tale.

Shal
https://groups.io/g/Group_Help
https://groups.io/g/GroupManagersForum


toki
 

On 01/03/2017 10:42 PM, Brian Vogel wrote:

I ask this because I'm exasperated and I just can't figure out how certain topic splits are occurring.
Look at the email client that is being used.

Most of the major email clients manage to break the RFCs so badly, they
shouldn't be called email clients, but rather
user-hostile-anti--communication-software-specifically-designed
to-ensure-that-communication-is-impossible-clients.



jonathon


 

On Tue, Jan 3, 2017 at 11:42 pm, Shal Farley wrote:
This usually tells the tale.

The keyword being "usually." I remember the contortions that even you (Shal) and Mark (offlist, through support) went through in trying to make the threading happen correctly in several cases in my group. Some cases, at least at that point, were very intractable. 
--
J

Messages are the sole opinion of the author. Especially the fishy ones.

I wish I could shut up, but I can't, and I won't. - Desmond Tutu


 

J,

The keyword being "usually." I remember the contortions that even you
(Shal) and Mark (offlist, through support) went through in trying to
make the threading happen correctly in several cases in my group.
I don't recall at the moment.

At some point prior to the message I cited Mark had added the "one-week" rule for messages without either form of citation field (In-Reply-To or References). I think that was to work-around various email clients, especially mobile clients, which fail to use or provide those either of those header fields.

Shal
https://groups.io/g/Group_Help
https://groups.io/g/GroupManagersForum


 

Brian,

On Tue, Jan 3, 2017 at 2:42 PM, Brian Vogel <britechguy@...> wrote:

I ask this because I'm exasperated and I just can't figure out how certain topic splits are occurring.

Today has been a big "inexplicable topic split" day on the Windows 10 for Screen Reader Users Group.

The topic using the refresh tool from microsoft to reinstall windows 10: is now split into three threads, all with titles that match as far as the human eye can see.  The same is true for help with outlook 2013 please! though it's only split into two separate threads.


I've traced things and figured out what happened. The system worked as designed, but perhaps the algorithm needs a tweak. We do threading two ways. One is by looking for the threading headers in the message (In-Reply-To and References), and use those to figure out which thread a message belongs to. These are standard headers that all mail clients should use, but alas some don't. 

If a message does not have those headers, we then look for a message with the same subject within the last week. If there is one, we assume this is a reply to that message and thread it.

Now, here's the rub. We only look for matching subjects when the incoming message subject starts with 'Re:'. Because we don't want to accidentally thread messages that should be in different threads. In your thread's case, one message came through without threading headers and without a subject that started with 'Re:', which caused a new thread to start.

So, should we remove the 'Re:' restriction? I could make the argument that the person that sent the reply message, without any threading headers and with a subject line that did not start with 'Re:' needs to upgrade his email client, because that's severely broken behavior.

Thanks,
Mark


 

Hi Mark,

Do we have an idea as to what that client might be? If so, I’ll let affected groups know.

Cheers,

Josephy

 

From: beta@groups.io [mailto:beta@groups.io] On Behalf Of Mark Fletcher
Sent: Friday, January 6, 2017 9:48 AM
To: beta@groups.io
Subject: Re: [beta] How on earth does Groups.io thread a topic?

 

Brian,

 

On Tue, Jan 3, 2017 at 2:42 PM, Brian Vogel <britechguy@...> wrote:

I ask this because I'm exasperated and I just can't figure out how certain topic splits are occurring.

Today has been a big "inexplicable topic split" day on the Windows 10 for Screen Reader Users Group.

The topic using the refresh tool from microsoft to reinstall windows 10: is now split into three threads, all with titles that match as far as the human eye can see.  The same is true for help with outlook 2013 please! though it's only split into two separate threads.

 

I've traced things and figured out what happened. The system worked as designed, but perhaps the algorithm needs a tweak. We do threading two ways. One is by looking for the threading headers in the message (In-Reply-To and References), and use those to figure out which thread a message belongs to. These are standard headers that all mail clients should use, but alas some don't. 

 

If a message does not have those headers, we then look for a message with the same subject within the last week. If there is one, we assume this is a reply to that message and thread it.

 

Now, here's the rub. We only look for matching subjects when the incoming message subject starts with 'Re:'. Because we don't want to accidentally thread messages that should be in different threads. In your thread's case, one message came through without threading headers and without a subject that started with 'Re:', which caused a new thread to start.

 

So, should we remove the 'Re:' restriction? I could make the argument that the person that sent the reply message, without any threading headers and with a subject line that did not start with 'Re:' needs to upgrade his email client, because that's severely broken behavior.

 

Thanks,

Mark


 

On Fri, Jan 6, 2017 at 9:51 AM, Joseph Lee <joseph.lee22590@...> wrote:

Hi Mark,

Do we have an idea as to what that client might be? If so, I’ll let affected groups know.



It's a Gmail user that uses their own email client instead of relying on the Gmail web interface. There are no distinguishing headers, but the message does have the sig 'Sent from Mail for Windows 10', so I guess that's what we're dealing with. I did a quick Google and didn't find any reports of this particular issue, but I did find a report of that mail client not respecting Reply-To headers, which would also cause (different) issues. Sigh, Microsoft.

Mark


 

On Fri, Jan 6, 2017 at 09:48 am, Mark Fletcher wrote:
We only look for matching subjects when the incoming message subject starts with 'Re:'.

Mark,

I looked back through the log to find all the recent threads I'd had to merge (because they were actually the same but had not threaded), and I can see no "Re:" in the titles according to the log. However, I'm pretty sure the "Re's" were there in the actual messages. 
--
J

Messages are the sole opinion of the author. Especially the fishy ones.

I wish I could shut up, but I can't, and I won't. - Desmond Tutu


 

Hi,

Thanks – will send a feedback unless someone posted about this problem in Feedback Hub app.

Cheers,

Joseph

 

From: beta@groups.io [mailto:beta@groups.io] On Behalf Of Mark Fletcher
Sent: Friday, January 6, 2017 10:02 AM
To: beta@groups.io
Subject: Re: [beta] How on earth does Groups.io thread a topic?

 

On Fri, Jan 6, 2017 at 9:51 AM, Joseph Lee <joseph.lee22590@...> wrote:

Hi Mark,

Do we have an idea as to what that client might be? If so, I’ll let affected groups know.

 

 

It's a Gmail user that uses their own email client instead of relying on the Gmail web interface. There are no distinguishing headers, but the message does have the sig 'Sent from Mail for Windows 10', so I guess that's what we're dealing with. I did a quick Google and didn't find any reports of this particular issue, but I did find a report of that mail client not respecting Reply-To headers, which would also cause (different) issues. Sigh, Microsoft.

 

Mark


Brian Vogel <britechguy@...>
 

On Fri, Jan 6, 2017 at 09:48 am, Mark Fletcher wrote:
So, should we remove the 'Re:' restriction?

Mark,

        My opinion is "yes" since you do other context checking that would make mismatching unlikely.  If you want to nail Jello to a tree or push a string up a mountain you'll take the "clients need to be fixed" approach.  It just ain't gonna happen.  It astounds me just how many e-mail clients are out there and just how many people will cling to clients that have not been officially supported for a decade or more.

Brian 


 

One argument against sucking in similar subject lines without a "Re:".

On another site that did associate threads like this, in a family-only group, one person had the habit of using the subject line "Family Update" on all their announcement messages, even though they had nothing to do with other "Family Update" messages. They all got swept into the same thread, which was annoying.

I can imagine similar groups where someone always titles their messages "Today's quote" or "Office Hours" or "Lunch menu".

So it sort of depends on which is more prevalent: Broken mail clients, or people who don't choose subject lines well.

JohnF


Brian Vogel <britechguy@...>
 

John,

          Not that I don't get your point, but how annoying is it to have a thousand-post thread entitled "Today's Quote" or "Lunch Menu" or, if it's so unspecific as to be entitled "Office Hours"?  These things are trivial ephemera by their very nature.

          If someone has something of substance to say, and wants to get it out to a broad readership, it's really incumbent on them to choose a Subject/Title that's adequately descriptive and entices folks to read further.

          There are lots of people who don't choose subject lines well.  For the most part, I doubt that many people will cry a river of tears if their posts get glommed on as the most recent on a "perpetual thread" that's likely trivial.

          So far, at least on Groups.io, I'm encountering a far greater issue with broken threading with subject lines very well chosen than I am with inappropriately threading badly chosen and "generic" subjects.

Brian


 

(and now my signature is back, BTW) I agree with Brian on this. My group, for example, requires everyone starting a thread to include their cat's name, and not just "new member" or "new to the group" or other generic title. You see these in yahoo, and in "see all messages in the thread" you get messages from 10+ years ago. I attribute the fault to the writers of the threads.
--
J

Messages are the sole opinion of the author. Especially the fishy ones.

I wish I could shut up, but I can't, and I won't. - Desmond Tutu