Tagging is useless

Mark Fletcher on Tagging

I was going to blog something about how tags are bad, evil horrible bad, and highlight the failure of existing search technology, but I couldn’t muster the energy. High level message: tags suck and are unnecessary except in cases where no other textual data exists (like photos, audio or video). Discuss amongst yourselves.

The failure in search technology that tags address is the inability of algorithms to process context and inference. This is likely to be an ongoing problem, as humans have a good deal of difficulty doing this successfully.

Imagine a blog post:

A guy walks into a bar, ouch!

Now I in what circumstances would I be able to find that joke using only the textual information? If I’m searching for guys or bars (slighlty worried about what google will send my way because of that line) you’ll stumple accross it. But when I really want to find this blog entry I’m actually looking for jokes. So I tag it joke and viola now people can find it if they’re looking for jokes.

Say powerline and Daily Kos both wrote articles titled: George Bush is the best President ever. Textually each of those articles would likely be very similiar, but people would probably be unhappy to find the version they weren’t searching for.

Tag one sarcasm and now you (and search engines) can differentiate. (My guess is that sarcasm will be the last line of defense in the turing test.)

If in fact textual information was enough to search accurately then why don’t search engines rely solely on textual matching? Incoming links, stability of the page in question, and other factors aside from the text of the page all affect search engine results. Clearly determining if something matches search criteria depends on more than just the text on the page. And tags can provide another piece of that contextual puzzle. The question is: Will tags become meta tags, marginalized by stuffing?

Community tagging systems will be able to fight this. As Yahoo starts to include My Web tags in their search index I think that they’re search will improve dramatically. This community tagging allows for meta information that is much more difficult to spam and going back to my original examples that kind of information can be very helpful.

High level message: Tagging can provide an excellent complement to textual information, making incredibly useful when searching.

LinkBlog Postings

A VC Posts about daily blogging with del.icio.us but it’s only for typepad users. I do something very similiar but it’s for wordpress.

Save the following script and set up a cron job (unix):

0 4 * * * perl /home/josh/src/perlAutoBlog/autoBlogWordPress.pl

then tag anything you want to show up on your blog as ‘linkblog’ and every day at 4am everything from the day before will be posted as an entry.

LazyWeb

LazyWeb: Harness the power of the internet to answer your questions. Ask a question on your blog, link to that page and it shows up. Think ask Metafilter without posting on Metafilter.

It’s an interesting feed to subscribe too.

Tagging For

Del.icio.us just rolled out tagging for people, so you can tag something specifically for me. Right now I have various had hoc solutions for this, but from now on if you want to send me a link tag it:

for:josh.schulz

josh.schulz is my user name, this is my del.icio.us page.

delicious,tags,internet

Wiki stuff

I didn’t have anything to do with this, except in as much as I sat next to the people who did have a lot to do with it. But it’s still cool so I’m going to point it out: ITtoolbox Wiki went live this evening. It’s a wiki, focused on IT information. Since it’s a wiki, anybody can edit any page. So you can end up with fantastic factual information contributed by God knows who (I know who put the first bit in, but I don’t know who spilled the beans about my Atlantic journey).

It’s exciting to see where this goes. Right now it’s wide open: It could end up being a valuable collection of information or a really stupendus flamewar. But it’s up to the world at large to make it what they want. It’s cool to be able to look at it tonight and have no idea what it will look like next week.

Itunes 4.9

ITunes 4.9 does indeed have podcasting built in.

I just switched from ipodder to odeo. I switched originally because I wanted the podcast in the browser functionality that Odeo offers, but that’s not implemented yet.

But I really enjoy being able to manage my podcasts remotely. If I somebody mentions an interesting podcast I can subscribe to it no matter where I am and it’ll show up on my ipod the next day. If I use Itunes for podcasting I don’t have that luxury.

I’ve been happy with odeo so far: Their tech support is excellent, I’ve had two questions for them and they got back to me quickly and thoroughly. I love being able to grab individual shows from a podcast without subscribing, that way I can check it out before I actually subscribe.

I do have one major beef though: Odeo allows you to tag podcasts, which is great. Ask anybody who has to listen to me talk much and they’ll affirm I’m firmly on the folksonomy bandwagon. However for tagging to be really usefull it has to allow for tag intersection. I want to be able to search for music+hip-hop or news+politics or sports+baseball, not just baseball or just news, or just politics or just hip-hop. I want to be able to drill down in a category by combining multiple tags. Odeo doesn’t let me do it and it drives me nuts, and makes finding things much harder.

Other than that though, it’s good stuff.

An experiment

I wrote a script to pull my itunes playlists into a format tivo can use because I searched around the net and there wasn’t a really good way to do it.

Because I’ve been looking for this for a while and I’ve been reading forum topics where other people ask for this sort of thing so I know its something people will find useful. I want people to find it and use it so I incur a great karmic bonus, but I don’t really want to go searching around for forum topics, registering, and posting ‘look at me!’ messages all over the place. Google will pick it up so eventually people who are searching explicitly for it will be able to find it. I also want a way for people who aren’t activly looking for it, but would find it useful, to stumble accross it.

So I decided to tag it in del.icio.us (itunes ipod tivo mp3 m3u playlist). I’m curious about two things:

1. How many people really do want this functionality.
2. How good is del.icio.us at helping people find new things.

Now I sit back and watch my server logs to see exactly how del.icio.us performs in spreading my little bit of code.

Tivo and Itunes

I love ITunes for organizing my music, especially using smart playlists to build playlists quickly. But when I’m at home I use Tivo Home Media Option to listen to music. Unfortunatly Tivo reads m3u playlists and Itunes doesn’t export them.

Luckily Apple has an SDK for COM access of Itunes available and it turns out it’s pretty easy to write script to replicate all of your itunes playlists as m3u style playlists.

That script is here: batchConvertPlaylists.

Unzip it, then open it up in notepad and change the directory you want the playlists written too (somewhere tivo can read). Then pull up a command window and run:


cscript batchConvertPlaylists.js


The directory will have an m3u copy of each of your playlists (dynamic and static). It works on win2k and Itunes 4.8. It will only work on windows platforms.

I run it as a scheduled task every day at 9am (after IPodder goes and I synch before I leave for work) so every day my Tivo playlists match my Ipod.

Enjoy!

RSS CMS

Rick from 1817 media is actually going to write a RSS only blog platform. I’ve got a lot of ideas about this sort of thing that turned out to be too unweildy to put in a comment on his blog so they end up here.

Roll out!

No day as sweet yet so inevitably likely to cause pain as roll out day. The day your code and ideas and work all get pushed center stage for the world to stare at. If it’s a bug fix, or an add on to an existing product it’s not that exciting, but if it’s something brand new it’s great day.

Today I launched something brand new. For the last two months I’ve been slamming my head against a series of walls, working really hard, and really long, trying to get this project off the ground. It’s that most wonderful of thing: I had a problem and a goal, and the middle was mine. The result is something I’m excited about. I built a system that does cool stuff and has all sorts of wicked implications.

By now you’ve noticed I’m being very vague, and I apologize, but I can’t tell you exactly what I did, or how, or why. But I can tell you what:

Related white papers, visit that link and notice the blue box on the top right. That’s not a random sampling of white papers. That’s a group of papers sifted out of the thousands on our network found to match the content of that blog post. But it’s not just blog posts It’s everywhere on our network, from any point therein I can find related white papers.

A lot of thought and effort went into designing the system that makes that possible. I got to do it from scratch every step of the way, I got a lot of leeway to design the architecture and implementation, and I’m proud of the results.