Answer highlighting in search results

Consider the example, [empire state height]. With today's improvements, the answer —1250 ft, or 381 m — is highlighted right in the search result:

This might be cool for Wikipedia, but it totally screws every other website owner.

If you have a website that deals with anything Google's algorithm has decided is a "fact" then watch out, your traffic is about to drop. Google's destroyed a lot of niche online markets but this is a broad attack against all sites.

The future of this feature is that you never have to leave Google at all.

Today, after receiving a report that the service was unavailable last night, I took another look at it. Still not sure why it went down, but while I was working on it I've added some basic logging (for troubleshooting purposes) and looked around the internet to see if people are using the service.

Since logging was never included in the first place, the only way to surface a reasonable count of URLs was to count the cache files. Since cache files are programatically deleted, I was pretty shocked to find over 11,000 files in that directory. Now that I have actually logging in place, the site displays a counter. Today's count? 11,411 URLs expanded!

Of course, people only know about the service because of the folks who have helped spread the word about it! Thank you to all these folks for mentioning expandUrl:

• Blog Brasil Academico "If you prefer a cleaner and more free advertising try expandurl.com."
• Troy Sabin "The Yang to the URL shortener Yin - URL expanders"
• Free Nuts #2 in their top ten URL expanders list!
• Borrowed Code Mentioned as part of a blog entry about fighting Twitter spam.
• Benny Chen "After failing to find a good solution, a short address before the humble beginning to run out to restore the service; Can only say that a group of prescient people, ah!"
• Kate Morris "There is another useful tool available - http://www.expandurl.com - it shows you not only where it goes, but how many times it is redirected and how. :) Not my tool, but a friend built it."

Finally, I discovered a Google Code project that integrates the expandurl API into Google Chrome. This is very cool to me, as it means the service will be used by lots of folks (even if they don't know they're using it!). :)

Not too shabby for 221 days in. ;)

Bing, ba-bing, bing bing, ba-bing!

Fruitless perhaps, but worth the effort nonetheless. ;)

Could your company pull up a backup of anything from seven years ago? Let alone make it functional in the same context it once was? I surprise people whenever they learn I keep emails for as long as I do.

Anyone who didn't believe they are keeping a Time Machine-esque index of every version of any page ever crawled should be convinced now. Just like a lot of folks want BackRub, I bet a lot of folks would love access to a searchable and far more comprehensive internet archive than the, well, actual Internet Archive.

Things I've learned about my sites from playing around with this?

• I did not rank quite as well as I remember ranking.
• I abused duplicate title tags, a lot.
• My general design style has evolved but is still recognizable (probably a bad thing!).
• I was reminded of some of our old domains and the odd subdirectory structure of the different web projects.

What's interesting about this is that "Special K" is slang for Ketamine a "general dissociative anesthetic for human and veterinary use" that is used as a recreational drug. Why would Kellogg's risk associating their cereal with recreational drug use by inviting their customers to a page which is purported to be an unbiased representation of a search term?

A quick visit to the Yahoo page shows absolutely no results related to Ketamine or recreational drug use. However, a Google search for special k devotes positions 4, 5, 6, 7, and 10 to pages related to Ketamine plus 1 of 7 related searches include a drug reference.

To give perspective, a Google search for "special k" (in quotes) gives 1.33 million results. Removing any references to the cereal using the search "special k" -kellogg -cereal yields 1.17 million results. If 88% of web pages on the internet which mention "special k" are talking about the drug, how can none of Yahoo's top results for the term mention the drug?

This is an example of the culture difference between these two engines that yields a dramatically different result in user loyalty.

Edit: Upon further research, it looks like someone already noted a difference in the search results. (Kocchi just misinterpreted the direction of slant.)

So, a visitor searching for custom t-shirts is going to see a page like this (I've highlighted where the organic SERPs show up):

I prefer a search engine result page that is simple and puts the results first. Here's the site search at GetYourShirts.com for the same t-shirt query. Of course, we won't get into a discussion comparing the quality of the results. ;)

Thankfully, Google Patent Search solves the most glaring of flaws in the official search. The interface for a "casual patent searcher" is a breath of fresh air although Bill Slawski's analysis of the patent search indicates that Google has a long way to go before capturing the hearts and minds of professional patent searchers, such as patent attorneys.

Niall Kennedy presented, mostly, a basic overview of feeds, defining them for attendees who were completely new to the world of feeds. The depth in which he covered the topic however, was impressive. I think Brett would do well by having Niall kick off any presentation on feeds/content syndication.

There were a few gems (for me) from his presentation. First, a helpful code snippet for linking to alternate language versions of the same document. This code can be read by search engine spiders and by some browsers (like Firefox).

Place in the <head> of the document:
<link title="Arabic" href="http://ar.example.com/" rel="alternate" hreflang="ar" type="text/html" />

Niall reminded the attendees the branding value of including a logo in your feed and the importance of validating your feed using a service like feedvalidator.org or installing a feed validator locally. He provided a list of sites to which you should publish/ping but personally I think ping-o-matic does the trick for everything.

Don't forget to subscribe to your own feed in the major online feed readers out there including My Yahoo!, Google Reader, and Bloglines. Take advantage of any tagging/categorization/rating features of those services to flesh out your own feed.

Niall also made a point of discouraging the creation of new tags, attributes, and categories for your feed. Take advantage of existing standards and extended vocabularies as that increases the chance of broad support in the various readers and aggregators.

Nick Klau, VP of Business Development over at Feedburner gave a presentation on trends they are seeing at feedburner and their own impressive growth.

One of the first, and most important points he makes, is that it is vital to feed owners to make sure that their feeds can be auto-discovered in all the major browsers. IE7 and Firefox 2 (ahem, and Safari) all have built-in readers so it's also important to check your feed in all those apps.

When considering using a third party to host your feed, you should map a subdomain (like feeds.example.com) so that you can retain control over your feed hosting. Although you might expect a feedburner representative to suggest otherwise, Rick made it clear that they do not have any intention of locking in users to their service and they believe firmly that publishers should maintain control regardless of which service they choose.

Owen Byrne, co-founder of Digg, spoke about the state of Digg and a little of their history.

Some of the scalability lessons he has learned with Digg are invaluable and reveal his software engineer background:

• Avoid premature optimization
  Get the code out there then see what needs to be optimized
• Cache, cache, and more cache
  I take this to mean that they do a lot of writing to disk instead of hammering the database for every single page load. He also mentioned memcached in this context
• Hardware is cheap, downtime is not
  Normally this argument goes "Hardware is cheap, developers are expensive" but I prefer the opportunity cost as a comparison against downtime. :)
• Lots of servers - spare, monitoring, testing, developing
  He said (if I recall correctly, I didn't make a note of this) that Digg has something like 90 servers but that many of those are spares or development mirrors of the production servers.

Chris Tolles, VP of marketing at topix.net gave some background about his work at the ODP and motivations in building an algorithmically edited news aggregator.

Topix.net provides over 50,000 feeds and ranks very well for locality + "news" searches. Chris believes it is the "freshness" of their content. Specifically that, even though they are republishing other sites content, they are doing it so quickly and frequently that the search engines love it.

Topix saw somewhat of a stagnation in growth. In response to this stagnation they added a message board for every news item and locality. You can see a U.S. map of forum activity on topix.net here (broken in Safari, use Firefox).

On the subject of getting feed readers, aggregators and search engines to recognize updated content correctly, Chris recommends the appropriate use of the <ins> and <del> tags to mark changes.

Finally, some of the "value added" by topix.net is simply that they are categorizing the content of others. This small bit of difference makes it possible for them to rank as well or better than the original content.

Added: Chris Tolles was written a great article comparing and contrasting Web 2.0 and WebmasterWorld this year.

Other new features include the ability to opt in your site's images into the Google Image Labeler project. Good that you can increase the quality of your images metadata (which hopefully means increased traffic from Google Image Search), not good that you can't see what metadata they end up producing. It'd be a nice feature if Google "donated" back to the image owners what keywords the Google Image Labeler generates for your images. I'm not suggesting that Google publish the metadata for all the images in Image Search -- only that they tell the image owners what metadata was discovered.

The charting looks cool but really only has value when you are experiencing problems -- otherwise it's just for the "ooh and aah" factor.

Altogether a nice clutch of new functionality. Thanks Google Webmasters Team!