If GMail is smart, how can it be this dumb?
October 9, 2010 1:22 PM Subscribe
Why does GMail (mistakenly) append new messages to older threads with identical subject lines?
In my GMail inbox, occasionally a new message will show up appended to an old thread. The condition for this seems to be the sharing of a subject line, usually something generic like "Re: Hey!" or "Re: Checking In." The old thread to which the new message is appended can be *quite* old sometimes.
GMail seems awfully (algorithmically) smart about so many things. Why is it having trouble telling threads apart? Can it be that it only matches subject lines, not message headers, recipients, etc.?
(And why would I trust a "Smart Inbox" if they can't get my dumb inbox right?)
In my GMail inbox, occasionally a new message will show up appended to an old thread. The condition for this seems to be the sharing of a subject line, usually something generic like "Re: Hey!" or "Re: Checking In." The old thread to which the new message is appended can be *quite* old sometimes.
GMail seems awfully (algorithmically) smart about so many things. Why is it having trouble telling threads apart? Can it be that it only matches subject lines, not message headers, recipients, etc.?
(And why would I trust a "Smart Inbox" if they can't get my dumb inbox right?)
Sometimes people reply to an old message and then delete everything rather than using contacts or address books, without realizing that deleting the body doesn't also delete the In-Reply-To header that points to the old message as parent, so that threaded email programs will still show it as a reply to that message. Usually this is noticed when a seemingly unrelated message with a different subject line shows up attached to an existing thread, but it could be happening in your case if they aren't changing the subject.
posted by Rhomboid at 1:54 PM on October 9, 2010 [3 favorites]
posted by Rhomboid at 1:54 PM on October 9, 2010 [3 favorites]
How do you expect Gmail to do this without basing it on the Subject line? Gmail is overlaying threading on top of existing email protocols which don't have a concept of threads. The subject, to:, cc: etc are the only ways to do threading. Google is smart but they're not magical.
posted by bitdamaged at 1:56 PM on October 9, 2010 [4 favorites]
posted by bitdamaged at 1:56 PM on October 9, 2010 [4 favorites]
This has happened to me but only very, very rarely. I'm talking, like, maybe 5 times tops, none recently, and I've been using Gmail since 2004.
I think it definitely must have something to do with the email's subject, but I have sent/received hundreds of emails with the subject lines "hello," "sup," "hey," "question," etc. and they all sort just fine.
If I remember correctly (there's no possible way I could locate it), at least one of the times this happened was under the following conditions: the email had the exact same subject, was sent to/from the exact same address, and had a similar subject matter (contained a unique name of a third party we were talking about).
If this is happening to you a lot, I have no idea what the cause could be; definitely send Gmail feedback on it.
posted by phunniemee at 2:01 PM on October 9, 2010
I think it definitely must have something to do with the email's subject, but I have sent/received hundreds of emails with the subject lines "hello," "sup," "hey," "question," etc. and they all sort just fine.
If I remember correctly (there's no possible way I could locate it), at least one of the times this happened was under the following conditions: the email had the exact same subject, was sent to/from the exact same address, and had a similar subject matter (contained a unique name of a third party we were talking about).
If this is happening to you a lot, I have no idea what the cause could be; definitely send Gmail feedback on it.
posted by phunniemee at 2:01 PM on October 9, 2010
This is a good incentive for always using better subject lines in emails. This practice will save your ass in the future when you need to search for information. If you always send emails like "my itenerary" you will find it really annoying to search for a specific one in the future. My Itenerary Chicago May 2010 will make it very easy to find.
Sadly, you have very little control over other people's subject lines.
I suspect I have largely avoided this problem in gmail by deleting most emails and only keeping ones that contain information that I can forsee the possibility of needing.
If you are worried about the priority inbox I suggest you try it. It doesn't really change anything and you can switch back and forth between the priority inbox and regular inbox views at will. All it does is try to learn what you read right away. It doesn't make anything go away and is more like an additional gmail filter/folder.
posted by srboisvert at 2:02 PM on October 9, 2010
Sadly, you have very little control over other people's subject lines.
I suspect I have largely avoided this problem in gmail by deleting most emails and only keeping ones that contain information that I can forsee the possibility of needing.
If you are worried about the priority inbox I suggest you try it. It doesn't really change anything and you can switch back and forth between the priority inbox and regular inbox views at will. All it does is try to learn what you read right away. It doesn't make anything go away and is more like an additional gmail filter/folder.
posted by srboisvert at 2:02 PM on October 9, 2010
I assume you already know about conversation view, and you think it's being applied incorrectly.
But are you sure the message wasn't an actual reply to an old message? It might seem weird, but here's how some people use email: they'll think, "Gee, I want to send fourcheesemac a message. How do I do that? I don't know his email address offhand. I guess I'll find the last email exchange I had with fourcheesemac, and reply to it." Now, that might not be the greatest way to do email. But the weirdness could be coming from the person sending you an email, not Gmail.
posted by John Cohen at 2:11 PM on October 9, 2010 [1 favorite]
But are you sure the message wasn't an actual reply to an old message? It might seem weird, but here's how some people use email: they'll think, "Gee, I want to send fourcheesemac a message. How do I do that? I don't know his email address offhand. I guess I'll find the last email exchange I had with fourcheesemac, and reply to it." Now, that might not be the greatest way to do email. But the weirdness could be coming from the person sending you an email, not Gmail.
posted by John Cohen at 2:11 PM on October 9, 2010 [1 favorite]
I don't have an answer, but this kind of thing gets on my nerves too. I suggested to Gmail that they allow us to manually split conversations into multiple threads. The more people that suggest it, the more likely they are to add it, if it's technically feasible (which it seems to me it would be).
This obviously won't fix the initial annoyance but at least it would let you separate them for posterity's sake (and for the replies to the incorrectly filed message).
posted by SuperNova at 2:42 PM on October 9, 2010 [1 favorite]
This obviously won't fix the initial annoyance but at least it would let you separate them for posterity's sake (and for the replies to the incorrectly filed message).
posted by SuperNova at 2:42 PM on October 9, 2010 [1 favorite]
Sometimes people reply to an old message and then delete everything rather than using contacts or address books, without realizing that deleting the body doesn't also delete the In-Reply-To header that points to the old message as parent
You can see this header in GMail by clicking on the arrow next to Reply and choosing Show Original. If a "new" message has that line in the header, it's not really a new message.
posted by smackfu at 2:43 PM on October 9, 2010 [1 favorite]
You can see this header in GMail by clicking on the arrow next to Reply and choosing Show Original. If a "new" message has that line in the header, it's not really a new message.
posted by smackfu at 2:43 PM on October 9, 2010 [1 favorite]
The subject, to:, cc: etc are the only ways to do threading
This is simply false. The In-Reply-To and References headers have been specified and in wide use since RFC 822 (1982). If you go to any site that archives technical mailing lists (like gmane.org, mail-archive.com, or marc.info) you will see plenty of threads where the subject was changed in the middle to reflect the new direction of the discussion but the threading structure remains. You will also see parent/child tree hierarchies that cannot be reconstructed from subject and to/cc, only by References (example).
posted by Rhomboid at 2:44 PM on October 9, 2010 [2 favorites]
This is simply false. The In-Reply-To and References headers have been specified and in wide use since RFC 822 (1982). If you go to any site that archives technical mailing lists (like gmane.org, mail-archive.com, or marc.info) you will see plenty of threads where the subject was changed in the middle to reflect the new direction of the discussion but the threading structure remains. You will also see parent/child tree hierarchies that cannot be reconstructed from subject and to/cc, only by References (example).
posted by Rhomboid at 2:44 PM on October 9, 2010 [2 favorites]
It is true though that GMail will use the subject and sender to group messages into threads where they aren't really replies. Like if the same person replies to a post twice on Facebook, both of those notifications will be grouped together even though there is no re: or In-Reply-To. So it's not impossible that GMail is just screwing up.
posted by smackfu at 2:52 PM on October 9, 2010
posted by smackfu at 2:52 PM on October 9, 2010
How do you expect Gmail to do this without basing it on the Subject line? Gmail is overlaying threading on top of existing email protocols which don't have a concept of threads.
This is factually incorrect. Existing email protocols do have a concept of threading in the form of the In-Reply-To header. This ensure that messages can be threaded, even if the subject changes and that only messages that are actually replies end up in the thread.
Unfortunately this is another example of existing email standards that Google chooses to ignore or override. In stead, in this instance, it favours of its own algorithmic interpretation of what it thinks threading should be over what actually is in a thread.
Can it be that it only matches subject lines, not message headers, recipients, etc.?
It seems to match subject and sender AND IIRC correctly mostly ignore the in-reply-to header, so it is likely not people hitting reply and changing the subject. In my experience it will happily decide that messages sent months apart with different contents belong in the same conversation.
N.B. Turning off conversation solves this problem only in as much as it turns off threading altogether.
(And why would I trust a "Smart Inbox" if they can't get my dumb inbox right?)
While I would not deny Google that Google product can be very useful, I think you would be foolish to assume that they have your interests at heart, rather than their own commercial ones, and that the latter do not take precedence at all times. As had been said before on here if you are getting it for free you are the product.
posted by tallus at 3:26 PM on October 9, 2010
This is factually incorrect. Existing email protocols do have a concept of threading in the form of the In-Reply-To header. This ensure that messages can be threaded, even if the subject changes and that only messages that are actually replies end up in the thread.
Unfortunately this is another example of existing email standards that Google chooses to ignore or override. In stead, in this instance, it favours of its own algorithmic interpretation of what it thinks threading should be over what actually is in a thread.
Can it be that it only matches subject lines, not message headers, recipients, etc.?
It seems to match subject and sender AND IIRC correctly mostly ignore the in-reply-to header, so it is likely not people hitting reply and changing the subject. In my experience it will happily decide that messages sent months apart with different contents belong in the same conversation.
N.B. Turning off conversation solves this problem only in as much as it turns off threading altogether.
(And why would I trust a "Smart Inbox" if they can't get my dumb inbox right?)
While I would not deny Google that Google product can be very useful, I think you would be foolish to assume that they have your interests at heart, rather than their own commercial ones, and that the latter do not take precedence at all times. As had been said before on here if you are getting it for free you are the product.
posted by tallus at 3:26 PM on October 9, 2010
On one of the mailing lists I administer, this was a huge topic for debate (along with the default gmail behavior of hiding your own replies to lists you are subscribed to.) The consensus then was that Gmail completely ignored the In-Reply-To header (I do not have a gmail account, so I can only report anecdata.)
All I can say for sure is that my gmail-using friends hate it when I change the subject of a message mid-thread, because it breaks threading for them (but not me - mutt FTW)
posted by namewithoutwords at 3:58 PM on October 9, 2010
All I can say for sure is that my gmail-using friends hate it when I change the subject of a message mid-thread, because it breaks threading for them (but not me - mutt FTW)
posted by namewithoutwords at 3:58 PM on October 9, 2010
Response by poster: Thanks, all. Weird. Happens rarely though.
posted by fourcheesemac at 5:22 PM on October 9, 2010
posted by fourcheesemac at 5:22 PM on October 9, 2010
I'm with Rhomboid on this one. I have never seen it happen, and I've been using Gmail as my primary account for both personal and professional email (fulltime freelancer) for, like... a really long time. Five years? Six? How old IS Gmail anyway?
And in theory, I should see this a LOT. Not only am I inclined to keep reusing the same subject line ("Argh" and "Bleh" being common) but my friends are, too.
Off the top of my head, I can think of three non-savvy users who habitually call up old emails, delete the contents, and reply to them. Because they can't figure out how to use the address book, and they don't remember everyone's email address. Sad but true.
I bet if you look closely you will find that this mysterious occurrence is happening just as Rhomboid says. The next time you get one, ask the sender how they sent you the message.
However, I have had it break the message thread when someone changes the subject line. I think it does that on purpose, assuming that if the subject changes, so does the conversation.
posted by ErikaB at 6:37 PM on October 9, 2010
And in theory, I should see this a LOT. Not only am I inclined to keep reusing the same subject line ("Argh" and "Bleh" being common) but my friends are, too.
Off the top of my head, I can think of three non-savvy users who habitually call up old emails, delete the contents, and reply to them. Because they can't figure out how to use the address book, and they don't remember everyone's email address. Sad but true.
I bet if you look closely you will find that this mysterious occurrence is happening just as Rhomboid says. The next time you get one, ask the sender how they sent you the message.
However, I have had it break the message thread when someone changes the subject line. I think it does that on purpose, assuming that if the subject changes, so does the conversation.
posted by ErikaB at 6:37 PM on October 9, 2010
Response by poster: Tallus, I wasn't asking about GMail's privacy issues and your response seems a red herring to me. Of course GMail tries to deliver a better service, however you define their "product," and there is no better web-based mail platform out there, as millions of others would seem to agree. I'm not unsympathetic to the critique, but I'm not naive, as you suggest. I'm not asking whether I can trust Google with the digital contents of my life, to which I have given informed consent; rather, I am asking whether I can trust their technology to automatically manage my digital content in ways I have hitherto managed manually. The question is practical, not ethical. Hope that clarifies things a bit.
To all, thanks. Email technology is a blind spot for me, and I tend to think of it as old and simple. I understand algorithmic programming conceptually and have been curious to see the development of what is effectively low level (and increasingly not so low level) AI technology at the level of end-user interfaces like Amazon's or Google's. I find it fascinating in particular how older informational architecture principles come into conflict with newer ones (I work extensively with archives of recorded sound, where the same issues are common, in different form.)
posted by fourcheesemac at 7:22 AM on October 10, 2010
To all, thanks. Email technology is a blind spot for me, and I tend to think of it as old and simple. I understand algorithmic programming conceptually and have been curious to see the development of what is effectively low level (and increasingly not so low level) AI technology at the level of end-user interfaces like Amazon's or Google's. I find it fascinating in particular how older informational architecture principles come into conflict with newer ones (I work extensively with archives of recorded sound, where the same issues are common, in different form.)
posted by fourcheesemac at 7:22 AM on October 10, 2010
Response by poster: ErikaB, next time it happens I'll examine the thread more carefully to see if you've nailed it.
I too have lots of "Arrrrrgh"s in my subject lines.
posted by fourcheesemac at 7:24 AM on October 10, 2010
I too have lots of "Arrrrrgh"s in my subject lines.
posted by fourcheesemac at 7:24 AM on October 10, 2010
Response by poster: Oh, and I should credit Rhomboid, 2d answer above, for the same explanation, put differently. I'll bet you guys are right. As I said, GMail is so well engineered in other ways that this seemed weird, like a broken window on a nice house. I half expect GMail to develop a feature that answers all my email for me with responses that sound just like me and are personalized for the recipient.
A boy can dream. I started emailing in 1990. If I had only known what a burden it would become.
posted by fourcheesemac at 7:34 AM on October 10, 2010
A boy can dream. I started emailing in 1990. If I had only known what a burden it would become.
posted by fourcheesemac at 7:34 AM on October 10, 2010
In stead, in this instance, it favours of its own algorithmic interpretation of what it thinks threading should be over what actually is in a thread.
Threading in the real world is harder than you think, as this write-up from JWZ shows. This was for Netscape 3.0, so it certainly predates Google and GMail.
posted by smackfu at 6:18 PM on October 10, 2010
Threading in the real world is harder than you think, as this write-up from JWZ shows. This was for Netscape 3.0, so it certainly predates Google and GMail.
posted by smackfu at 6:18 PM on October 10, 2010
This thread is closed to new comments.
posted by Knappster at 1:34 PM on October 9, 2010