“Gnomedex 2002” On Google Shows Again The Need For Webmaster Control Over Titles & Descriptions

060609-gg-t.jpg

We’ve written before about the need for search engines to give webmasters more
control over their titles and descriptions. Today, I came across another good
example illustrating why this is needed — Google telling me that Chris Pirillo’s popular
Gnomedex event is happening in 2002, as
you see in the screenshot above.

I’m usually not one to do much with screenshots, but I’m jumping into them
big time with this post to help illustrate why this problem is happening.
Gnomedex is listed in the Open Directory’s fairly well abandoned

Cyberspace Events
category. Here’s how it looks there:

060609-odp.jpg

Ages ago, some human editor there wrote up a title and description for the
then-Gnomedex 2002 event. Now when you search on Google, Google decides to use
that title and description rather than pulling information from the Gnomedex
home page itself.

This won’t always happen. For example, look at this screenshot:

060609-gg-t2.jpg

Now the title is different. What happened? Google dynamically decided that
using the title from the HTML title tag of the Gnomedex home page was more
relevant than using the Open Directory’s title of this page. Notice that
previously I searched for just
gnomedex
but in the second screenshot, I looked for
gnomedex tech
conference
. Adding those additional words not only changed the results but
also how Google felt it should describe the page.

Now look here:

060609-gg-t3.jpg

This time, BOTH the title and description are drawn from the page itself,
rather than the Open Directory. What happened? My query changed to

gnomedex tech conference enthusiasts
. Google again examines the Open
Directory’s title and description for the page to see if it contains my search
words or would be relevant to show. It decides against that and instead turns to
using content from the page itself.

There are very good reasons for descriptions to dynamically change.
Extracting text from a page that matches what you’ve searched for can help you
know if that’s a good page to click through to. That’s the primarily reason
Google did dynamic snippets/descriptions ages ago.

Nor do you want to always depend on the meta description tag. As I’ll show in
a future post, many authoring programs can insert this tag (along with title
tags) with no descriptions or with no information. That’s bad — and that’s one
reason why Google depends on the Open Directory information in part.

But enough is enough. MSN Search
just introduced
a way for site owners to just say no to having the Open Directory information be
used to describe their pages. I want the rest of the search engines to climb on
board with it. Using that would eliminate the problem above, just as it would
eliminate the case we’ve already
written about,
where Google is making it seem as an Alaskan gubernatorial candidate has already
been elected.

Since I’m doing screenshots, let me spin that one out, as well:

060609-gg-a.jpg

As you can see, a search on
alaska governor
shows you both a fact result at the top of the page (and it’s correct — Frank
Murkowksi is the governor) plus a link to the
official site
for Murkowski.

Now Tony Knowles used to be governor of Alaska. The former two-termer wants
the job again, so he’s running
against Murkowksi. He’s even got a campaign web site, as you can see in a search
for his name:

060609-gg-a2.jpg

In fact, you can also see in the screenshot that the Knowles campaign has
been so successful that he’s already been declared governor of Alaska by Google
in the same search results. Look at the second listing, and you’ll see the title
for the official Alaska governor web site is "Governor Tony Knowles."

What’s going on? And why’s the description mention Murkowski as governor?
Again, it’s an Open Directory issue, where Google has decided to use just the
outdated title from over there.


This Open Directory
category
, until recently, used that title for the governor’s web
site. The ODP has since fixed it after Threadwatch
brought attention to it, but
it’s still floating around in the data stream. For example, here’s a search at
the ODP showing it:

 060609-odp-a.jpg

FYI, this isn’t just a Google issue. Over at Ask for
tony knowles:

060609-ask-a.jpg

You can see in the third listing that Ask does exactly what Google is doing.
But also look at the candidate web site listed first. Isn’t he running for
governor? Yes, but the OPD still has him listed

here
as running for US Senate. Ask used both the ODP title and description,
making this seem out of date.

And here’s MSN, birthplace of the
noodp
meta tag you can use to prevent this type of stuff happening. For a search on
tony
knowles
:

060609-msn-a.jpg

Oops — Tony is both running Alaska as governor currently and also has gone
back in time to run for US Senate. The good news is, if he can go back and get
elected, then maybe he’ll meet with Sergey Brin last Tuesday when other senators
refused to at
the last minute and help lobby his counterparts in the US House Of
Representative not to vote against net neutrality, as they did
yesterday. Time
travel — magic!

Seriously, while it’s great to have the opt-out, that’s only going to help if
people proactively use it. Most will not do so until after some big problem is
flashed in their face. A better solution these days is probably to stop using
the ODP as a data source at all unless people specifically opt-in for using its
titles and descriptions.

Well, at least Yahoo doesn’t have these types of problems. After all, Yahoo
uses stuff from its own Yahoo Directory. So in a search on
tony knowles:

060609-yh-a.jpg

Man, can’t the ODP get a break? Sure — this time, it’s the Yahoo Directory
category that is out of date. Cruise

over there
, and you’ll see that tonyknowles.com is given a description by
Yahoo about his senate attempt. That was correct at the time, but since then,
Knowles has changed the web site over for his gubernatorial attempt.

FYI, sometimes Yahoo DOES use the OPD as well as its own directory. So a
noodp tag would be useful to help things there — plus perhaps a noyahoo tag as
well.

Let’s go back to Gnomedex, and I’ll finish off with a few things. Here’s what
Ask is currently showing:

060609-ask-g.jpg

First, the Smart Answer on the top from Wikipedia is pretty nice, I think.
But how about the listing of the site itself. What’s with this "Gnomedex 6.0"
stuff?

That was the title tag on the site the last time Ask was there, which
according to the

cached page
was May 22. Since the title tag has changed since then, the two
don’t match. That will get corrected the next time Ask goes back — assuming it
doesn’t decide to use the ODP listing, for some reason.

As I said, a good first move would be for all the other search engines to get
behind the noodp tag that MSN introduced. Honestly, a better solution would be
to stop using the ODP information altogether, so this type of stuff stops
happening.

FYI, for more background,
Proposed
Search Engine Standard For Titles & Descriptions
on our Search Engine Watch
Forums covers a long discussion we had on the topic last year. That led to a
live forum discussion on the topic involving search engine reps that’s covered
here:
Session Five: Day Two: Indexing Summit 2: Redirects, Titles & Descriptions
.

I’ve been terribly, terribly remiss in not doing the last follow through
myself on what was discussed, including the idea of a "I really mean it — I’m
not just making a mistake" meta tag that tells the search engines to only use
material from your web page. I did a short summary of the results

here
, and next week, I’ll finally make time to get things finished up on
that. The time is finally right for change, I suspect.

Want to comment or discuss? Start a thread in the
Search
Engine Optimization
section of our Search Engine Watch Forums, and we’ll
come back and link to it from here.