LJ: tags

Nov. 24th, 2005 01:42 pm
cellio: (avatar)
[personal profile] cellio
Tags: so close and yet so far. Sigh.

I really like being able to tag journal entries. This should, in principle, make it much easier to find specific entries in the future -- the "gee, I know I posted about $SUBJECT sometime last year" factor. Yes, I know tags can be used two ways, for retrieval and for entry synopsis; I care deeply about the former and not at all about the latter. If my subject lines, opening paragraphs, and cut-tag text fail to give you an idea reasonably quickly about what the entry is about, I have likely failed as a writer.

Ok, so retrieval is critical. LJ documentation says that you can only retrieve the 100 most recent entries with a given tag, which sounds like a lot but isn't if you've had your journal a while. Worse, it was only some time later, after I'd tagged a bunch of entries, that I discovered that the limit is really 70 (or 75?), not 100. At least in my journal's style. I think this is probably a bug in the style and I did report it a while back; as I recall, it disappeared into the land of "we'll investigate". But that was at least thee months ago, so maybe they really consider it a featue.

This limitation might be fine if you could query tags using boolean expressions. It would let me have fairly general buckets while still being able to find, say, my entries about SCA cooking ("sca" and "food"). Boolean expressions have been requested, but I don't think they're working on it.

So, erratically, I've been refactoring some of my tags that turned out to be too general. While a few of these were in the "what was I thinking?" category, like my "judaism" tag, some are ones I fully expected to stay under 100 when I created them. And they did -- but they went over 75. Or they're in danger of doing so.

There will come a point in the future when the limits on entries will make tags useless. I've had this journal for over four years and wouldn't be surprised if I still have it four years from now, after all. Eventually they're going to have to either boost the limit or implement boolean expressions. (And maybe hierarchical tags.)

So please excuse the taggy clutter in my journal; this is a gradual process. (And I've got a lot of untagged entries, too.) If you notice suboptimal tags on my entries, particularly older ones, feel free to drop me email or a comment about it.

(no subject)

Date: 2005-11-24 07:28 pm (UTC)
From: [identity profile] cahwyguy.livejournal.com
...or they could just provide the ability to search within *your* journal. That would be really useful.

(no subject)

Date: 2005-11-24 07:32 pm (UTC)
From: [identity profile] sdorn.livejournal.com
This sounds incredibly kludgy, but you could tag your LJ entries through del.icio.us (http://del.icio.us/).

(no subject)

Date: 2005-11-24 10:45 pm (UTC)
From: [identity profile] nsingman.livejournal.com
I wonder if just tagging this entry with "lj," rather than "lj" and "tags," perhaps, is suboptimal? :-)

Happy Thanksgiving, Monica!

(no subject)

Date: 2005-11-25 02:48 pm (UTC)
From: [identity profile] patsmor.livejournal.com
Perhaps you could discuss the limits with the folks who run the place?

I've tried to make my tags more specific, but am failing, so I don't know what I'll do when I get to the limit. Alas!

(no subject)

Date: 2005-11-25 08:27 pm (UTC)
From: [identity profile] nickjong.livejournal.com
I've also been thinking about how to tag my journal, and I reached the same conclusion that the tags' purpose should be to enable retrieval of previous entries on a topic, not for mere summarization. I'm still trying to figure out how to generate new tags. A tag is only useful if it applies to more than one entry, but does that mean I shouldn't create the tag until I have two entries to which to apply it? This method seems susceptible to false negatives, wherein I forget that I've already written about a topic or when. The alternative is to propose new tags for each new topic of each entry about which I might potentially write again. Later you could prune unduplicated tags by sorted tags by usage. But this method seems inelegant and potentially misleading to the reader. Hmm, I suppose a mixed approach may be best in practice.

Expand Cut Tags

No cut tags