Backtagging

October 31, 2009 2:51 PM

Ongoing tagging and the backtagging project.

Just curious, if the backtagging project is over or if it's an ongoing project, as it may still be valuable.

posted by mr_crash_davis mark II: Jazz Odyssey to Feature Requests at 2:51 PM (34 comments total)

It's ongoing! Backtaggers still have backtagging rights and I think we've still got a chunk of Meta to do.... "5,420 untagged or untitled posts. | 7,076 completed so far" I'm away for the weekend w/ very limited internet, but if people want to be added as backtaggers, email cortex and he can set you up I'm fairly sure. Or wait til Monday and email me [no, emailing me and saying "hang on to this until monday" will not work] and people can help. But otherwise, yeah it's ongoing but sort of back-burnered. Backtagging MeTa is tougher than the other parts of the site.
posted by jessamyn (staff) at 2:56 PM on October 31, 2009

Thanks, Jess. Maybe this should be more prominently displayed somewhere. I figure if I didn't know about it that means at least one other person didn't either.
posted by mr_crash_davis mark II: Jazz Odyssey at 3:10 PM on October 31, 2009

Announcement of when the original project was finished.

Practically speaking, backtagging only covered posts prior to January 2005 so the bad tags you see since then are completely the fault of the originally posters.
posted by smackfu at 3:20 PM on October 31, 2009

(Maybe we need a retro-retagging project next.)
posted by smackfu at 3:22 PM on October 31, 2009

Oy, I feel guilty for neglecting my duties. Okay. I'll open a beer and get ready for some baseball and do some backtagging.
posted by rtha at 3:33 PM on October 31, 2009

I would like to suggest again that a new flag for posts - "These tags suck! Add to retagging queue," but with a snappier name - would be useful.
posted by nowonmai at 4:19 PM on October 31, 2009

Heh. I think you might be cut out for backtagging, nowonmai, if you're not already on the crew.
posted by cortex (staff) at 5:04 PM on October 31, 2009

Oh lord, do I ever agree with nowonmai. Some of the tags suck, especially on MeTa where the content of the thread is highly highly likely to be different than the stated issue.

If you decide to wade into the tagging morass that is MeTa, I highly recommend looking at the popular tags first, and committing yourself to reading at the very least the beginning quarter and ending quarter of the actual comments. I'm not going to pretend it's for everyone, but you do learn a crapload about metafilter history.
posted by donnagirl at 5:31 PM on October 31, 2009

Is there any argument for backtagging MeTa posts as if you were posting it—that is, without hindsight?
posted by carsonb at 6:50 PM on October 31, 2009

carsonb, that's an interesting suggestion. I guess I was coming at it from the perspective of findability, so if people ask "when did that awesome foo injoke arise?" there will be a thread marked "awesomefoo' and "injokes". I have a vague memory of discussing this in the earlier backtagging threads, but I can't find it at the moment.

In a practical sense, it probably just means our backtagged posts are better tagged than our in-the-moment tagged ones. I know posters (and mutual contacts, right?) can tag posts at any time, but I doubt many posts get more tags late in the game.
posted by donnagirl at 7:34 PM on October 31, 2009

As a practical matter, the idea that you need to read the whole MeTa thread to tag it is probably why that site's taking forever and shows no signs of being done soon. It's much, much easier to tag/title things just based on what's on the backtagging page.
posted by smackfu at 7:42 PM on October 31, 2009

I guess I was coming at it from the perspective of findability, so if people ask "when did that awesome foo injoke arise?" there will be a thread marked "awesomefoo' and "injokes".

Things are findable when there's some predictable format for categorization. On the other backtagging projects, for the most part tags were applied as if they were composed with the post. ('metafilterhistory' and 'deadlink' being the exceptions I think of off the top of my head.) It would make sense then to keep that rule for backtagging MetaTalk, and use some other format for keeping track of awesomefoo and other injokes. Like the wiki. Even though it becomes a two-step process, it is a more consistent folksonomy.
posted by carsonb at 9:58 PM on October 31, 2009

Oops, I was in the original project but thought it was all done. I'll get back to it. If we can find a way of reviewing some existing tags, I'd like to help in that too.
posted by paduasoy at 1:51 AM on November 1, 2009

I spent some time playing with term extraction as a means of checking for doubleposts, but I'm realizing its true value is in suggested tagging. I won't repost the comment, but take a look here to see what I mean.
posted by potch at 2:45 AM on November 1, 2009

The backtagging interface is capable of suggesting tags and, yeah, it seems to work pretty well. How would that be implemented on the post page though? Would it have to be done after a preview button press, or could it be done automagically as the post is composed?
posted by carsonb at 5:56 AM on November 1, 2009

This is a good reminder to get back to it; it has somewhat slipped my mind.
posted by never used baby shoes at 9:27 AM on November 1, 2009

If I remember right, the suggested tags were actually powered by a Yahoo! API that has since been shut down. But it's a testament to how little backtagging I've done in the last while that I can't be sure that's the case.

potch, the idea of a roll-our-own term extractor is one I've liked for a while now, though it keeps not being something that gets to the top of my priority list. I did some work on the idea a couple years ago when I was doing the Word Cloud stuff (may it rest in limbo).

Folks who've mailed me about joining the crew, we'll get you added in the next day or so.
posted by cortex (staff) at 10:29 AM on November 1, 2009

Hey, I was tagging and found a thread that was an awesome example of what I mean when i say reading might matter more.

Here, the stated problem is a garden-variety comment spammer. But, in a move that cortex described as "Bam! Haughey with the sucker punch!", Matt used the thread to announce a test of the front page tab design. People chat about that, and have some conversations about the various uses of favorites on the way.

It's not Metafilter History Writ Large, but tagging it "spammer" and moving on seems like something of a loss.

Sorry to keep banging this drum, just thought it might help clarify what I was talking about.
posted by donnagirl at 6:22 PM on November 2, 2009

Or it's an argument they shouldn't announce changes in random MeTa threads.
posted by smackfu at 5:25 AM on November 3, 2009

Oh man. If I can find a hotel with decent internet and don't sleep tonight I think I have a solid shot of upping my performance over my previous top ten showing. Yall been warned.
posted by allkindsoftime at 5:47 AM on November 3, 2009

I've added everyone who has emailed me, if anyone else wants to help, please drop me a note through regular email.

Or it's an argument they shouldn't announce changes in random MeTa threads.

There are always wonderfully convincing arguments that things shouldn't happen the way they happen, otherwise we really wouldn't need mods at all. This is sort of a "some tags are better than no tags" situation, so do your best and thanks for helping.
posted by jessamyn (staff) at 6:28 AM on November 3, 2009

Oh, definitely agree. Especially since I really doubt people go back and add summarizing tags to their own posts after the discussion is complete.
posted by smackfu at 6:42 AM on November 3, 2009

We've gotten past the 9,000 MeTa threads tagged/titled tonight; less than 3,500 to go!
posted by Monday, stony Monday at 8:07 PM on November 8, 2009

Fewer than 3000 left, and dropping fast! There must be several people tagging simultaneously.
posted by donnagirl at 6:17 PM on November 10, 2009

Yay everyone, this is so cool.
posted by jessamyn (staff) at 7:24 PM on November 10, 2009

Only 1759 left at last count; we've cut down two thirds of it since the thread was posted.
posted by Monday, stony Monday at 2:04 PM on November 13, 2009

We're under 1000 now!
posted by donnagirl at 8:22 PM on November 16, 2009

499!
posted by Monday, stony Monday at 4:14 PM on November 17, 2009

295! We could finish tonight. Or, someone or some group that does not include me could finish tonight, because I'm going to bed.
posted by donnagirl at 8:27 PM on November 17, 2009

I'm going to bed too. The remaining posts tend to be quite hard to tag/title, so I often "pass the buck", only to see the same posts pop up again. (less than 250 left, which means we're 95% of the way there, compared to where we were when this was posted)
posted by Monday, stony Monday at 8:54 PM on November 17, 2009

Heh, interesting. I wonder if we should make note of the last posts to get tagged so we can look at them as potential case studies in untaggability.
posted by cortex (staff) at 9:06 PM on November 17, 2009

This project has made me wonder about how the system chooses 10 posts to serve up for tagging. I seem to see the ones I pass on in cycles - they'll come up in nearly every set, for a bunch of sets, and then disappear. But I see them again much later, so it's not that someone else tagged them. I'm getting some now that I know I've never seen before, because they're easy, like meetups, and I wouldn't have passed on them. That wouldn't be remarkable, except that there are some posts I'm sure I've seen more than 25 times.
posted by donnagirl at 5:30 AM on November 18, 2009

I'll go in and bat a little clean-up.
posted by jessamyn (staff) at 6:24 AM on November 18, 2009

DONE!!! Oh sweet Jesus, those last ones were painful. I won't make any claims of them being done well, but they're done. Can we have a party now? Backtaggers seem like they might be particularly intense partiers.
posted by donnagirl at 9:15 AM on November 18, 2009