In my Yahoo My Web: An eBay For Knowledge article out today, I cover the promise that Yahoo My Web has for potentially improving search results through trust
networks. As for the promise of tagging to
improve results, I find myself just as dubious as I’ve been about tagging.
Where to begin? I could let Gary run lose with a litany of complaints. In fact, he probably will share his own perspectives in the near future. But I’ll dive in on the problems as I
Most important, the tags are in the impossible position — one that even Yahoo admits when I talked with them about this — of trying to do two things at once that aren’t
compatible. They are:
- Trying to show the freshest content on a topic
- Trying to show the best content on a topic
At the moment, neither situation is happening. Freshness is determined by when a document was saved. So if I save the iPod home page
under the tag of ipod (link viewable only to My Web 2.0 users), there’s nothing "fresh" about it
except that I’ve just added it. The page has been around for ages. The mere act of saving it didn’t make it fresh.
In contrast, the new
iTunes 4.9 software with support for podcasting is new. If someone adds that, it’s a nice way to alert others monitoring this
topic to the latest about iPods.
Fresh Versus Best
Tagging at a specialized search engine like Technorati doesn’t face the double-duty challenge. Technorati is dealing primarily with feed and blog
content. That content by its very nature is fresh in some way. In other words, no one is blogging, "Hey, here’s the iPod web site" and feeding it as fresh news via
ipod tag. People are for the most part — aside from spammers — saying something new or offering a fresh opinion about iPods
and things related to iPods.
As a result, if you want to tune into the latest stuff about iPods, the relatively specialized and fresh content that Technorati gathers can be found via tags.
The new Live 8 area is a good example of this.
In contrast, if you want to find a general good resource about a subject, the tags at Technorati suck. Where’s the official Live 8
web site? It’s not at the top of the recent blog posts for the Live 8 page. The only reason it’s on that tag page at all is because
Technorati made a customized, special page for the event. For a regular page, go back to the ipod tag page and try to find the official iPod home page. You
Directories Were For Categorized Best Stuff
Showing a list of the best content on a categorized topic — as opposed to the freshest content — is the role traditionally filled by directories such as Yahoo’s own
Yahoo Directory. Look at the
Live 8 category there. It’s sparse, surprisingly so (or perhaps not given Yahoo’s general
abandonment of its directory, but at least it has something at all, unlike the Open Directory). But nonetheless, you have no problem finding the official site and top resources
about the event, including Technorati’s page!
iPod? When I looked at the Technorati tag page for this, one of the top things listed was someone spamming to sell me sunglasses using a gibberish page which was tagged as being about iPods.
Meanwhile, Yahoo’s iPod category shows the official
site first along with a bunch of resources that look good and are focused broadly about iPods.
Well what about del.icio.us? People are bookmarking general information over there, right, not just fresh stuff! Are they? Whenever I look, it seems like people are busy bookmarking
a lot of new stuff.
Looking at the google tag today, I saw bookmarks about the new Google Earth service or the new Google Maps
API. How about ipod? Some new stuff, some old stuff — and the same result you get with Yahoo. Stuff that’s "fresh" isn’t
necessarily so, while the popular view shows me only "recently" popular stuff rather than what I’d call "always
popular" such as the iPod home page.
Tagging In The Verticals
How about Yahoo-owned Flickr? Yahoo talked to me this week about how 70 percent of all items on Flickr are tagged, but then it
immediately qualified without prompting that because Flickr is a photo service, tagging is much
Indeed — if you don’t tag a picture, you pretty much have no good way of finding it. Tagging makes much, much sense in a photo space. And I love photo tagging. Check out
my Photo Search: Google Picasa
2 Vs. Adobe Photoshop Album 2 article from earlier this year. I tag like a madman with Photoshop Album. I live to tag!
You know what? I’m weird. And people tagging on Flickr? They’re weird as well. Weird in a good, organized way. Go talk to people you know who have digital cameras —
not your net happy friends but relatively ordinary people or don’t work in some net-related industry. They aren’t tagging, not on their computers and not with Flickr. Maybe
they will eventually, but it’s far more likely it will only happen among the masses in areas where tagging is really useful and essential. For general web search, it’s not.
Tagging — like spontaneity — has time and a place. For some verticals, as I’ve
written, it may
make more sense. That’s especially so for relatively little services that aren’t going to be spam targets. But tagging web listings in general so far makes me think Yahoo’s not going to please anyone.
It gets worse, by the way. Tagging will help you keep all your My Web content you’re saving organized, right? But what happens when you’ve created hundreds of
tags for thousands of pages? Are you going to browse pages? Everyone largely abandoned browsing directory categories ages ago because keyword search was like a warp drive to
zip you to what you wanted, as I’ve explained.
If you really do save thousands of pages over time, you’re not going to want to rely on tagging to locate things. You’ll probably just keyword search. Even more so, that
will be essential, as the tags you initially created probably won’t hold up as things change over time. Do you retag everything? Chances are, you won’t.
Another backwards step example? We’ve had automated clustering technology for ages that will put content into categories, or tag them, if you prefer that term. Check out
Clusty, an example using Vivisimo’s long-developed tech.
Yahoo bought two different search engines — AltaVista and AllTheWeb — that also had clustering that no longer gets offered. Yahoo’s own current technology is even used to
create the Yahoo News Tag Soup "tag cloud" that I wrote about last
month, tech you can now apply to any site or collection of sites you’d like.
Why not use this tech to organized My Web automatically into tags? At the very lest, it would avoid problems like the "important bookmarks" tag being so large in My Web’s
current tag cloud, something that annoys Gary to no end.
It might also help with the short term tag cloud bombing problem I’m sure that’s going to emerge. Look at this:
That’s from the A Search Marketer’s Look At Yahoo My Web 2.0 article I just posted. In
about 15 minutes of work, I popped up "rio karma" and "mp3 player" into the cloud. They won’t last, but neither was I working particularly hard to make it happen. Tag cloud
spamming at the moment seems incredibly easy.
Yahoo says it has defenses in place that will stop this, stuff that will ramp up as needed. We’ll see. But just having just having to have those defenses at all reeks of
another step backwards. Rather than tags solving the search spam problem, an entire new way to eliminate tag spam is going to be developed — just as search spam has had to
endure an arms race of defensive measures.
One more step backwards example. As mentioned, some people are looking to tags to keep up with what’s new. There’s another way to do this. You create keyword-based news
alerts to monitor new stuff.
The problem with the major search engines is that keyword-driven news alerts they offer aren’t tapping into blog and feed content. That could be fixed over night. And news
alerts help ensure that if you’re looking for information on podcasts, you might get it even if someone "tagged" what you wanted in the completely different "podcasting"
I still miss Excite’s awesome NewsTracker service that we had way back in 1997. But there are plenty of good replacements that will automatically scan for stuff on the news
sites, and I covered a few here recently. Hopefully we’ll see the majors come up with ways for you to
flag keywords you wish to monitor in blog and news content, in the way Technorati’s Watchlists work or as PubSub allows, to name only two such
Nice To Have, Just Don’t Expect Much
It’s important to note that the long term plan for Yahoo ISN’T to use tags to refine web results. As my other article out today discusses, Yahoo is depending on
trust data to improve results. That will be applied to
the keyword data primarily inherent in the pages themselves, as well as link data. Tagging will have a role, but not the dominant one. It certainly won’t take over for organizing.
That’s one reason my long term view isn’t to worry about it. Tags are there for those who want them, which is good — very good. They will be useful to some people,
especially so when limited to particular communities. When Yahoo introduces popularity sorting, general tags viewed by everyone might even get better. But as long as they have
to do double-duty, I suspect they still won’t fulfill either role particularly well.
In contrast, an alternative would be for Yahoo to experiment with some type of social compilation of its actual directory, similar to what I
suggested about an Open Directory alternative last month. Let me tag the "best" stuff on a particular topic
separately from something that’s just fresh, new, cool but not the best in the long term. It would be interesting to see how those two different lists developed.