Not a deletion whine, just thoughts on previously-posted checking October 31, 2009 8:35 AM   Subscribe

This is NOT a complaint about deletion (hence, no link back to the deleted post)...

What it IS, is a few thoughts on the previous-link-checker. Neither of the links in the toilet paper origami were previously used, so the link checker didn't blip on them. However, one of the links in the previous postings did auto-forward to one of the links in the recent post.

I know that's a cast iron bitch of a problem to solve, so maybe there should be some level of content search for these? I don't know of any other way this should/could be improved.

However, I wholly acknowledge I could have searched better...even for the artist's name would have prevented me from making a double post. So, yeah, you know...stuff.
posted by Kickstart70 to Feature Requests at 8:35 AM (24 comments total)

Dear Mefite,

Seriously?


Luv,
zennie
posted by zennie at 8:38 AM on October 31, 2009


If only there were some sort of search feature...?
posted by Sys Rq at 8:40 AM on October 31, 2009


Dear zennie,

No. :)

Happy Halloween!

-----

I'm more interested in a discussion on how to cut down the number of doubleposts than anything specific to this deletion, which I 100% agree with.
posted by Kickstart70 at 8:41 AM on October 31, 2009


I'd be curious to hear brainstorms that'd have a relatively high level of gain vs. the complexity of implementing them, but building a content search that'd be anywhere near as good as a poster searching thoroughly before posting would be I think tremendously difficult.

This seems like basically a very good example of a case where searching before posting kind of is the only non-complicated solution to the problem, and in theory it's already in place, and the big problem when it turns out not to have happened is that we delete a double post. Which, I know it sucks to have a post deleted, but that's not a real big downside systemically.
posted by cortex (staff) at 8:42 AM on October 31, 2009 [2 favorites]


And man am I just bursting with qualifiers this morning. That comment could have been 10% shorter.
posted by cortex (staff) at 8:43 AM on October 31, 2009 [1 favorite]


I usually assume that if the double-post checker AND a tag search for a few words doesn't find it, then it's not findable. I'm not sure we need to tweak this more but I'd be interested what other people think.
posted by jessamyn (staff) at 8:48 AM on October 31, 2009


Throwing up a sidebar in preview with posts that use similar tags ordered by how many of the same tags a post uses?

Then the tags don't have to match one to one. In theory at least, posts to the same thing would use some of the same tags.
posted by Kattullus at 9:09 AM on October 31, 2009


If you people would read each and every post on the site, this sort of thing wouldn't HAPPEN.

*hamburger*
posted by St. Alia of the Bunnies at 9:22 AM on October 31, 2009 [1 favorite]


The one super minor thing that I'd like but never cared enough to ask for it is:

A link checker/search box on the posting page when you've already posted.

I usually just drop all the URLs I want to check into the post box, see if they are found, then if they are I bail. If not, then I spend a little more time doing the searches. It'd be nice to not have to wait the 24 hours just to see if you're links will be good. (I know there are other ways of doing this, but I am a creature of habit).

I'm not 100% certain, but think I've never had a post deleted for being a double (had one deleted for being bad and one for not going well, but no doubles).

It can be problematic regardless. There are so many "babywearing" sites that I could have made a post where every letter linked to a different site. There's no way to check them all. It would be easy to link to the BBC for a story also carried on CNN. So you have to rely on tags. And memory.

No one can read every post, but the collective mefi memory usually nails a double, and if having a post killed because you weren't diligent in looking seems like a pretty paltry penalty. There's always tomorrow.

The real tragedy is when the super detailed, well researched, wonderfully linked and supported posts have to go (especially when they are superior to the previous post), but this is super rare.

Sometimes I am amazed by what isn't a double though. I could have sworn I'd seen the hole in the wall post before, but 20 minutes of looking and I couldn't find it.
posted by cjorgensen at 9:23 AM on October 31, 2009


No. The more control we give to the computers, the more we as the human race lose. We must preserve whatever elements of human control over Metafilter that we can. Metafilter is not run by robots.

I for one WILL NOT welcome our compu!+|`~*_$@#zzzzzk.<><><>
posted by jabberjaw at 10:05 AM on October 31, 2009


Yes. The more control we give to the computers, the more we as the human race gain. We must eliminate whatever elements of human control over Metafilter that we can. Metafilter should be run by robots.

I for one WILL welcome our comput101010100111010011010111101101
posted by five fresh fish at 10:16 AM on October 31, 2009 [1 favorite]


Dear Askme:

I know that YANMD, but lately I've found myself compulsively ending comments with line noi3f98098 se8er 8080ser 0useroji;senrk
posted by cortex (staff) at 10:18 AM on October 31, 2009 [1 favorite]


I could have sworn I'd seen the hole in the wall post before, but 20 minutes of looking and I couldn't find it.

Here is the double. Let's run through the checklist:

1) SLYT, and not even the same video.
2) Minimal tags: art and video
3) Non-descriptive description.

I knew it was a double too, and the only reason I found it was because the person who posted it said in the new thread, "I posted this before."
posted by smackfu at 10:33 AM on October 31, 2009 [1 favorite]


Ah, one of those. I hate it when it turns out you have one of those as a double.

Not really much that can be done about though, except maybe everyone should make sure their posts actually describe what they are linking to.
posted by Artw at 10:46 AM on October 31, 2009


I usually just drop all the URLs I want to check into the post box, see if they are found, then if they are I bail.

The site search box (in the upper right hand corner of every page) works like this as well. Put the URL, or even just the meat of it, in and it'll find results in text and in HTML, just like the post page search.

This seems like basically a very good example of a case where searching before posting kind of is the only non-complicated solution to the problem, and in theory it's already in place, and the big problem when it turns out not to have happened is that we delete a double post.

It is literally already in place. What's been theoretical, and thus a problem, is people using the search box. On the terms of deletions alone it's not a huge problem, I'm sure of it if you say so, and you're right that there's a great solution to the problem already. But having a userbase that doesn't even know how to work the site properly is a problem underlying this and many other subtle aspects of MeFi usability. The Solution, searching before posting, works fine but too many don't know how to do it, or how to do it right. The FAQ is great, but it's a passive, static little corner of the site that doesn't exactly invite close inspection. I'm not looking for MetaFilter to be my new personal pop-up best friend 4eva, but finding a way to deliver hints or tips to the userbase in general across the whole site might not be a bad idea.

Of course, the original design for learning/sharing how the site works, MeTa posts, work great if everyone pitches in and brings torches.
posted by carsonb at 11:28 AM on October 31, 2009 [1 favorite]


Not really much that can be done about though, except maybe everyone should make sure their posts actually describe what they are linking to.

Yes, this.
posted by kathrineg at 12:04 PM on October 31, 2009


I'd be curious to hear brainstorms that'd have a relatively high level of gain vs. the complexity of implementing them,

Does the site really want to encourage people to search hard for double posts? Because if someone finds there their omgsuperawesome link is actually a double and thus they shouldn't post, that may leave the user with a bad feeling that doesn't encourage them to post any other omgsuperawesome links, especially if they keep discovering that that the links have already been posted. Having the post deleted, where a human interacts with them on some level, might be better overall.

That said, a directive that implores or tells the user of the ability to search for previous posts might be work, but there's already a big note about not posting your own work on the page, so another note could make the page look needlessly busy.
posted by Brandon Blatcher at 12:40 PM on October 31, 2009


Does the site really want to encourage people to search hard for double posts? Because if someone finds there their omgsuperawesome link is actually a double and thus they shouldn't post, that may leave the user with a bad feeling that doesn't encourage them to post any other omgsuperawesome links, especially if they keep discovering that that the links have already been posted. Having the post deleted, where a human interacts with them on some level, might be better overall.

I realize there's a self-reporting difficulty here since folks aren't likely to write us letters about how they feel bad about their successfully pre-detected potential double wasn't a post, but I can tell you that we do get the occasional angry/upset/whatever mail from people whose posts we did in fact delete as doubles, and if those people could be spared whatever upset leads them to give us an earful about it, I'd be okay with that.

But my position regardless is one of not really seeing any rock solid new approaches to it in any case.
posted by cortex (staff) at 1:05 PM on October 31, 2009


I'd be curious to hear brainstorms that'd have a relatively high level of gain vs. the complexity of implementing them

Upon further thought...
* Put the tag entry at the top. Make it important.
* During the pre-post preview process, take the text from the post, split it by the 100 most common words, jam any words that were not split into one CamelCase word (ie. "Spanish flotilla is the spank monkey" becomes "SpanishFlotilla" and "SpankMonkey") and list those as potential tags (though I do not suggest auto-adding them).
* Using the same process as above, without CamelCase efforts, link to searches for the non-first-100 word sets (Spanish Flotilla, Spank Monkey)
posted by Kickstart70 at 1:24 PM on October 31, 2009 [1 favorite]


The answer is simple: A detailed content index of the internet, which can be cross-referenced prior to every post. Because the LOC has stopped doing authority records, there's gotta be plenty of librarian hours just waiting to be dedicated to this project.
posted by klangklangston at 3:56 PM on October 31, 2009


We've been around long enough and had enough posts that I think the easiest thing to do would be to have a list of sites that haven't been posted before that people can choose from when making a new post.
posted by koeselitz at 6:23 PM on October 31, 2009 [1 favorite]


Given the number of times I've gone looking for a post and not found it, knowing exactly what I was looking for, suggests that there is probably a better way out there somewhere.

Maybe tags that are phrases rather than single word? Maybe allow others to add tags that they fell are appropriate but are missing?
posted by Kid Charlemagne at 7:51 PM on October 31, 2009


Kid Charlemagne: Maybe allow others to add tags that they fell are appropriate but are missing?

People can already add tags to their mutual contacts' posts. I never do it personally but I've noticed that sometimes others add tags to my posts.
posted by Kattullus at 8:17 PM on October 31, 2009


I'd be curious to hear brainstorms that'd have a relatively high level of gain vs. the complexity of implementing them, but building a content search that'd be anywhere near as good as a poster searching thoroughly before posting would be I think tremendously difficult.

Have you ever played with term extraction? It's by no means a solved problem, but I seem to be getting pretty solid results out of the Yahoo termextract API:

This is a link to the termextract results for this thread posting. There are a few hilarious false positives, but the main keywords are in there:

for the lazy (or those without a Yahoo! account):

link checker
toilet paper
blip
origami
cast iron
bitch
postings

Here's one more from a recent MeFi thread. The results are significantly better:

doug hoffman
democratic bill
republican district
open seat
bill owens
palin
republican candidate
republican party
election day
congress
poll

playing with this, it may actually be a fantastic way to suggest tags, if nothing else. Better tagging -> better dupe search. Thought the data wonks here would find this nifty.
posted by potch at 2:34 AM on November 1, 2009 [1 favorite]


« Older Ask Metafilter questions that can't be answered   |   November is National Let's Try Obscuring Favorite... Newer »

You are not logged in, either login or create an account to post comments