Tag Guidelines

November 13, 2006 7:40 AM

Tags: I hate how they have to be crunched into one word. What's the policy/ettiquette about using underscores? Does it break the database? Or should I avoid them just because they're not convention, therefore no one will know to search for them?
posted by Eideteker to Etiquette/Policy at 7:40 AM (53 comments total)

You should avoid them because no one will search for them. People are going to be searching for WongKarWai not Wong_Kar_Wai.
posted by chunking express at 7:47 AM on November 13, 2006

A tagging system that doesn't allow for spaces in tags is just an amusing exercise for the developer. It's not really meant to be used.
posted by Armitage Shanks at 7:55 AM on November 13, 2006

I really see no reason they can't be separated by commas. LiveJournal's had a four-word, separated-by-commas interest system for more than half a decade. The four word limit keeps it from getting out of hand, and helps encourage consistent tagging. Even better would be separated by semicolons, so I can tag something "Good Math, Bad Math", for example.
posted by Eideteker at 8:10 AM on November 13, 2006

Separatingbycommasisdumb whatweneedis "quoting support" alaflickr.
posted by Plutor at 8:24 AM on November 13, 2006

The disadvantage of using something other than spaces as a delimiter is that you are thereby requiring all users to know the delimiter. Whatever disadvantages space-as-delimiter carries, it is the most intuitive solution available.
posted by cortex at 8:25 AM on November 13, 2006

Flickr's use of quotes to include spaces (which I like), ends up with the tag "a tag with spaces in" being the same as the tag "atagwithspacesin" (i.e. they both point to flickr.com/tags/atagwithspacesin or whatever), which I think is a good way to handle it. This would also avoid the issue of stupid_tags_with_underscores_in.
posted by EndsOfInvention at 8:28 AM on November 13, 2006

Flickr has a weird hybrid solution. You can use spaces in tags, but only if you put the phrase in quotes. And the photos with the spaced tag are grouped with the unspaced ones.
posted by smackfu at 8:28 AM on November 13, 2006

Wow simulpost.
posted by smackfu at 8:29 AM on November 13, 2006

Sounds like Flickr understands user interface, whereas many do not.

That's how it should be.
posted by twiggy at 8:29 AM on November 13, 2006

The disadvantage of using something other than spaces as a delimiter is that you are thereby requiring all users to know the delimiter.

Which you describe right next to the tag entry field, as Matt does now. That's not the problem

Whatever disadvantages space-as-delimiter carries, it is the most intuitive solution available.

Which, of course, explains why there are currently 12 posts tagged with "john." No, cortex, using commas or quotes is much more intuitive, as they allow for multi-word tags without requiring the singularly *un*intuitive multi-word smush.
posted by mediareport at 9:18 AM on November 13, 2006

Yeah, I should redo it so that it's like flickr, displaying multiple words with a space as long as you bound it with quotes. Underscores are lame.
posted by mathowie (staff) at 9:20 AM on November 13, 2006

Thanks, Matt. I agree that underscores are lame.
posted by Eideteker at 9:26 AM on November 13, 2006

manyProgrammersPreferStudlyCapsWhenDealingWithThingsLikeThis. itIsAnEasyTaskToTeaseApartWordsWithSuchASystemAsWell. underscoresAreHeckOfLameInAnyCaseAndShouldBeDiscouraged.
posted by boo_radley at 9:32 AM on November 13, 2006

Thanks, Matt. While you're thinking about tags, would it be possible to add a simple plurals feature to the tag search, so that "magazine" and "magazines" each pull up posts tagged with both categories? There are a number of tags that are heavily split like that (movie/movies, game/games, etc), and it would be great if you could find a way to automatically return both results when a member searches for one.
posted by mediareport at 9:33 AM on November 13, 2006

mediareport, I think you're misunderstanding me. commas and quotes as elegant delimiters for partitioning multi-word tags may be intuitive to folks accustomed to tagging systems, but to the neophyte who doesn't think twice or even read the explanatory text, neither is as intuitive as a space for delimiting single tokens.

I'm not defending spaces as the Right Thing, I'm saying that, to the naive user, spaces, which are the fundamental delimiter of English and a whole lot of other natural languages, are more intuitive than Given Delimiter X. Intuitive != "intuitive from an informed design perspective". There are folks here who are unfamiliar with basic HTML.
posted by cortex at 9:34 AM on November 13, 2006

I'm not defending spaces as the Right Thing

Great, good. Then we don't have an argument.
posted by mediareport at 9:40 AM on November 13, 2006

Well, not that argument, anyhow.

*highfive*
posted by cortex at 9:43 AM on November 13, 2006

Tagging is the law of the land; Tag as thou wilt.
posted by loquacious at 9:49 AM on November 13, 2006

Taxonarchy?
posted by cortex at 9:55 AM on November 13, 2006

mathowie: if you're considering mediareport's idea of collapsing variations on a word (e.g. report/reports/reporting, etc), please take a look at the Porter Stemming Algorithm, which makes that sort of thing easier. Doing so would be awesome.
Seriously Awesome, dude.*nods*
posted by boo_radley at 9:56 AM on November 13, 2006 [1 favorite]

Well, not that argument, anyhow.

The other one's not worth having, really. I thought by "the most intuitive solution available" you meant "the best solution for MeFi." You didn't, it isn't, end of story. :)

*highfive*
posted by mediareport at 10:15 AM on November 13, 2006

I more or less agree, vis-a-vis the importance of the argument, and believe you have characterized well the fulcrum of our disagreement!

*extremely manly, swaggery highfive*
posted by cortex at 10:28 AM on November 13, 2006

I'm not defending spaces as the Right Thing, I'm saying that, to the naive user, spaces, which are the fundamental delimiter of English and a whole lot of other natural languages, are more intuitive than Given Delimiter X. Intuitive != "intuitive from an informed design perspective". There are folks here who are unfamiliar with basic HTML.

I'd actually disagree here. In English, it's pretty normal for lists of things to be separated by commas. For the purposes of making a list of descriptive terms, commas are probably a more natural delimiter than spaces alone.
posted by Karmakaze at 10:45 AM on November 13, 2006

Writing is sufficiently artificial that discussing "natural delimiters" would be a less-than-productive topic. Looking at most writing that comes across my desk, it seems that bullets are the "most natural" way to represent a list. Pick a camp and pay your dues, really.
posted by boo_radley at 11:01 AM on November 13, 2006

Really, this is an inherent flaw of tags, and confuses casual users of them (have a look at Youtube's tag system sometime). It's like in the early days of wikis, when page links were created with CamelCase. Silly system, designed by nerds.

Quotes and commas are both fine solutions. I'd say that commas is more intuitive -- as long as you write "separate tags with commas" somewhere, I don't think anyone's going to get it wrong and use spaces instead.

Of course, the usefulness of multi-word tags is another issue entirely.
posted by reklaw at 11:05 AM on November 13, 2006

Could we get a textbox on the tag list screen so we can type in a tag directly and have the site go directly to that tag?
posted by blue_beetle at 11:14 AM on November 13, 2006

I know this is at this point a silly and moot argument. I'm just keeping it up for the sake of, well, interesting argument. So:

Writing systems are inherently artificial, yes, but we make them a natural part of our daily lives.

Try writing a paragraph quickly and naturally while omitting any spaces between words. Tell me then that spaces aren't the closest thing, for a literate person, to a natural, instinctive delimiter.

By the same token (ha!), we don't use commas in major search engine strings. Commas may be common list delimiters in some contexts, but that clearly doesn't apply to all common computer-related tasks that Joe User undertakes.

Similarly, many, many folks apparently don't understand that you can (and, depending on your goal, need to) use quotation marks to delimit a literal fixed string rather than just page-level co-occurrance of a handful of search terms. So while quotes are an elegant and useful tool, the likely aren't in Joe User's working vocabulary.

Again, I'm not at all defending spaces as an elegant solution. I'm defending them as the simplest possible solution that gets X% of the job done without requiring any instructions for naive users. If quotes and commas are basic hand-carved tools, then space-as-delimiter is the very act of breathing.
posted by cortex at 11:24 AM on November 13, 2006

I'm defending them as the simplest possible solution that gets X% of the job done without requiring any instructions for naive users.

Er, no. The number of botched multi-word tags is a significant problem for the site's tagging system. Folks find multi-word tags useful; hell, they *are* useful. A system that doesn't accomodate them naturally is a system that could use a fix. Have you not noticed how common botched multi-word tags are around here? I don't know why you're being kind of obtuse on this one. When you look at how folks actually use the site, the problems with multi-word tags caused by using spaces as a delimiter far outweigh whatever minor learning curve the use of commas requires.

But again, why are you continuing this? Matt's already agreed with the comma crowd.
posted by mediareport at 11:50 AM on November 13, 2006

Personally, I think ♫ is the most intuitive delimiter.

tags♫argument♫batshitinsane

See?
posted by languagehat at 11:56 AM on November 13, 2006

As James Burke noted, languagehat, "The only thing you can say about a man who thinks he's an egg salad sandwich is that he's in the minority."
posted by boo_radley at 12:00 PM on November 13, 2006

But again, why are you continuing this?

For Christ's sake, I find the conversation interesting. I like challenging and having challenged the respective arguments in play and understand more fully what contexts folks are arguing from.

For example, it's clear to me now that I should have been explicit about the domain of my argument—I'm not arguing about Metafilter's potential tagging needs so much as potential tagging situations on the Internet at large. When I say "gets X% of the job done", I'm talking about the general job of having usergroup x tag stuff, not about deploying tags ideally on Metafilter.

So in the general case, where you want to ask random soccer moms and grandpas and webTV users to apply tags to things, single-word tokens delimited by spaces is a simple, minimal-learning solution that may sufficiently meet folksonomical needs. That's my thesis. I'm not saying it's right for Metafilter and I'm not trying to convince Matt; I'm just talking ideas.

Matt's already agreed with the comma crowd.

Which is super, because I wasn't trying to convince anyone otherwise.
posted by cortex at 12:47 PM on November 13, 2006

and understand more fully

that should be understanding—god help me should anyone think the hubris a literal reading implies was intended
posted by cortex at 12:50 PM on November 13, 2006

I also resent the use of 'lame' as an insult. Not saying it's a complement or anything, but can't you use crap instead?
*hipturd voice*: "oh the underline is just so downs_syndrome/spastic/congenital_dislocation_of the_hip". See? It hurts and is very cleft palette.
posted by econous at 1:06 PM on November 13, 2006

I resent your use of the word complement.
posted by smackfu at 2:13 PM on November 13, 2006

single-word tokens delimited by spaces is a simple, minimal-learning solution that may sufficiently meet folksonomical needs.

Again, even just "talking ideas" about "potential tagging situations on the Internet at large," I think you're clearly wrong. Commas are at least as intuitive as the space, are just as "minimal learning," *and* take care of the multi-word problem, which you consistently fail to address.

/anal ex-technical writer
posted by mediareport at 2:24 PM on November 13, 2006

As a true believing vegan I resent random motherfuckers telling me I shouldn't teach my grandmother to suck eggs.
posted by econous at 2:34 PM on November 13, 2006

I haven't failed to address the multi-word problem, I'm suggesting that there are possible contexts in which multi-word space-delimited tokens are not a required feature. That what X% is getting at: we can imagine an ideal set of features for something and also imagine a reduced but functional featureset that is acceptable for some purposes.

By analogy, regular expressions are not Turing-complete, but they still do enough of the job that they're useful.

As for commas vs. spaces, I just absolutely can't agree with you. Commas would not be hard to teach, but they'd require at least a little bit of teaching: you'd have to successfully instruct your userbase to deploy them. You'll never, ever have to teach someone to put spaces between their words. That is the very soul of intuitiveness.
posted by cortex at 3:02 PM on November 13, 2006

It's just a question of values, cortex. To me, who dealt with this kind of thing as a technical writer, the issue of the regular creation of useless broken tags is paramount over the - very minor, really - leap to clearly explained comma use. That is all.
posted by mediareport at 3:12 PM on November 13, 2006

It's just a question of values, cortex.

Agreed. Just wanted to be clear enough about where I'm coming from that you don't think I'm nuts. I am all for commas and quotes around here.
posted by cortex at 3:21 PM on November 13, 2006

You're just bored on a Monday and chose the wrong issue for a thoughtful stand. :P
posted by mediareport at 3:29 PM on November 13, 2006

Oh, I don't know that it was the wrong issue.

*contends restlessly*
posted by cortex at 3:40 PM on November 13, 2006

You'll never, ever have to teach someone to put spaces between their words. That is the very soul of intuitiveness.

Indeed, cortex, but it makes categorization more difficult. Take the following for example:

good humor man

Should this match on "good", "good humor" and "man", "good humor man", "good" and "humor man", "good humor man," or all of the above? For the person tagging, natural intuitiveness is their own construct, it is not universal. One person's sense of totally obvious is another's good... humor, man!

So there has to be some unifying constraint that everyone can agree to: a sacrifice of interface for the greater good. Personally, I think the best system would use symantics to break words down to their root ("metafilter", "metafilters" and "metafiltering" would all be equivalent), with optional quotes around phrases. Commas would be treated as phrase separators, the same as spaces, thus avoiding any confusion. Thus:

good, humor, man
good humor man
good, humor man

...would all split into three terms (the last would be the sacrifice of the "ideal" for the greater good), while...

"good humor", man
"good humor" man

...would split into two phrases.
posted by Civil_Disobedient at 7:08 PM on November 13, 2006

Absolutely. The merit of consistent and correct multi-word groupings is not debatable.

On the other hand, in the bottom-barrel space-delimited configuration, a tagging of [good humor man] could still meaningfully retrieve "good humor man" on a search for those tokens. Granted, it would also hit a node tagged with "man gets hit in nuts good humor"—that generality is part of the cost of the simpler system—but it's not as if the input and output of the system would be tragically flawed.

And, for that matter, the search system could give added weight to ordered co-occurance while delivering results that include non-ordered co-occurances as well. Which is what pretty much any search engine worth its salt does. So it's not all baby-with-the-bathwater.
posted by cortex at 8:05 PM on November 13, 2006

How else am I supposed to find out about everything related to the?
posted by knave at 8:23 PM on November 13, 2006

Ha. And of.
posted by mediareport at 9:28 PM on November 13, 2006

Yeah! Or banjo!

Wait. Shit. That doesn't work as well for the 99% of tags that aren't articles and prepositions. Lark it all.
posted by cortex at 10:25 PM on November 13, 2006

Cortex:

Would you write:

"I went to the store and bought grapes tomatoes chicken breasts milk and bread?"

or

"I went to the store and bought grapes, tomatoes, chicken breasts, milk and bread?"

What's more natural?

Then, what's more natural:

grapes, tomatoes, chicken breasts, milk, bread

or

grapes tomatoes chickenbreasts milk bread

or

grapes tomatoes "chicken breasts" milk bread
posted by empath at 11:52 AM on November 14, 2006

And honestly, if someone can't figure out what 'seperate tags with commas' means, what are the odds whatever tags they come up with are worth a damn?
posted by empath at 11:55 AM on November 14, 2006

empath, my point re: spaces is that the one thing I would not write, nor would anyone else, is this:

"I wen tot the store and bought grapestomatoeschickenbreastsmilkandbread."

Note that even the comma separated list is not a CSV—you've gone and inserted a space after each comma.

The example of a written sentence of descriptive English differs from that of tokens typed into an input box, regardless—folks operate in different modes in different contexts. I probably would not type "grapes, tomatoes, chicken breasts, milk, and bread" into some mythical recipe-search box—I'd forgo the commas and quote "chicken breasts", unless otherwise instructed.

In either case, though, there is a commonality—spaces between terms, whether also delimited by commas and quotes or not. That's all. Spaces are more atomic, more instinctive. Doesn't make 'em perfect, by any means.

And honestly, if someone can't figure out what 'seperate tags with commas' means, what are the odds whatever tags they come up with are worth a damn?

Yes, but if someone can't figure out how to install MT or Wordpress or write HTML and perl by hand, what are the odds that they'll start a blog? Etc.

There are broader, simpler, lower-reaching applications for ideas than those that appeal to the tech-literati. It's conceivable that someone would want to garner quantity-over-quality tagging for one purpose or another. Anything that they could do to boost quantity by reducing barriers to participation would be worth considering.
posted by cortex at 12:11 PM on November 14, 2006

"Note that even the comma separated list is not a CSV—you've gone and inserted a space after each comma."

Well, no one suggested a CSV.
posted by empath at 1:57 PM on November 14, 2006

I didn't suggest anyone had. The distinction is that comma-as-intuitive-delimiter is still less fundamental in human-generated contexts than space-as-delimiter, such that when we deploy commas in human-readable lists, they don't supplant the space as they do necessarily in machine-oriented CSV formats. Just a nonce example.
posted by cortex at 2:18 PM on November 14, 2006

*high five*
posted by mediareport at 6:58 PM on November 14, 2006

*throws up horns*
posted by cortex at 7:13 PM on November 14, 2006