March 29, 2003

ieSpell

A handy Internet Explorer plug-in that I frequently use is ieSpell, which will spell check text that you type into forms. It's great for help ensuring that this blog comes to you type-free.

Posted by Andrew at 02:24 PM

March 27, 2003

Mozilla build comments

MozillaNews hosts a page where users comment on the quality of Mozilla nightly builds.

This page is also useful if you've been waiting for a particular feature to land.

Posted by Andrew at 09:46 PM

March 18, 2003

What if Internet Explorer would have lost?

ZDNet has a great articles this week "What if Netscape had won?"

Charles Cooper fantasizes about a world were Netscape would have won the browser war and reflects on the current, one browser state of the Web, where Microsoft has had little incentive to innovate since 1999.

In addition to the article, there is a lively "Talkback" section where users get to voice their opinions on the Cooper's thoughts.

Thanks to Andy King's WebReference Update for highlighting this article.

Posted by Andrew at 09:02 PM

March 13, 2003

Opera 7.03 is out

Opera releases version 7.03 of its browser.

This fixes some security issues..

Posted by Andrew at 10:59 PM

Mozilla 1.3 final is out

Mozilla released version 1.3 of its browser today.

Among the new features in 1.3 are Mozilla Midas (rich text editing controls), improved spam filtering, and the ability to dynamically switch profiles.

Posted by Andrew at 10:06 PM

The Google Dance

This past weekend many website owners have sat in front of their computers and pressed their browser's refresh button repeatedly as Google has begun its monthly update. The update, now widely know as the "Google Dance," has become a monthly online festivity where webmasters closely watch as Google's new index gradually comes online and either rejoice or despair over their new rankings.

Why is it called the "Dance"?

The update period, which typically lasts a few days each month, is known as the "Dance," because the result pages at the main Google page and its two test domains (www2.google.com and www3.google.com) frequently fluctuate as the new rankings gradually come online. Until the dance begins, the results from the main domain and the test domains are mostly stable.

What's happening during the Dance

The Dance is when Google adds a new index with the results from the month's earlier "deep crawl" of the Web. I'll say a little more about the "deep crawl" a little later. During the Dance, Google uses the previously mentioned www2 and www3 domains to test the new index for any abnormalities. Once the new index has been tested, Google will steadily update each of their data centers with this new index. While this is happening, search results from the main Google domain can change on a minute-by-minute basis, since some searches will pull its results from an updated datacenter and others will pull from one that still has the old index.

Deep Crawl vs. Fresh Crawl

The Dance solely concerns itself with integrating the latest monthly deep crawl into the Google index. However, if you frequently search Google using the same keywords, you may notice that the search results fluctuate more than once a month. This is due to a relatively new feature of Google known as the "Fresh Crawl." The fresh crawl occurs almost continuously to spot frequently updated sites and to add the new content to the engine's index. This frequent fluctuation in the engine's search results has come to be known as "Everflux." Due to these slight variances, the accepted method to detect the beginning of the dance is to look for differences in the number of backlinks to major sites (such as Yahoo) between the main site and the test domains, and not by looking for changes in search results.

You can easily spot pages that have been fresh crawled by examining Google search results pages and looking for pages that have a date in the last line of their entry (between the page URL and the "cached" link). The fresh crawl uses a different spider, known as "freshbot," than the deep crawl, which uses "deepbot." These two spiders use two different blocks of IPs and can easily be differentiated, with deepbot using IPs that start with 216. and freshbot using IPs that start with 64.

With the emergence of freshbot, does the deep crawl still matter? Yes, it still matters quite a bit. The "fresh" results are not stable, with fresh pages popping in and dropping out of the results pages on a daily basis. The stable, persistent rankings are based solely on the deep crawl. Also, Google uses the deep crawl results to calculate PageRank, that magical numeric value representing a page's importance, which is displayed on the Google Toolbar via a little green bar.

Watching the Dance

As stated earlier, the surest way to know that the Dance has started is to check for differences in the number of backlinks between the main Google site and the test sites. You can do this by searching for "link:www.yahoo.com" at the main Google site and www2.google.com and www3.google.com and watching for any variance between the results. The Google Dance Tool streamlines this process by allowing you to search all three domains simultaneously, presenting all three results in a frameset. This tool also allows you to simultaneously search each of Google's datacenters.

During the Dance, you can use this tool to get a sneak peek at the new search rankings, watch for new page's inclusion into the index, and to check on the progress of the dance as the new results propagate to the separate datacenters. You can also get a sneak peek into a page's new PageRank by changing the IP to which the Toolbar points.

Further Reading:

Posted by Andrew at 05:43 PM | Comments (2)

March 10, 2003

Andy King interview

Last month I mentioned and recommended Andy King's new book, Speed Up Your Site: Web Site Optimization. Today Adrian Holovaty has posted an interview with Andy.

Adrain and Andy discuss web optimization within the context of news sites, but Andy's advice holds true for almost any other type of site.

Topics covered include:

  • Decreased performance caused by the overuse of online ads
  • Chunking of long stories into smaller, more quickly downloaded sections.
  • The percentage of the online population still using modems
  • Optimal download times and page sizes to aim for
  • Proper use PDFs, images, and other girthy file type
  • Using "high-bandwidth" and "low-bandwidth" versions of sites
  • Semantic markup


Posted by Andrew at 10:13 PM

March 04, 2003

A silly site

Figured that I'd pass along this silly site.

Enjoy :)

grudge-match.com/

Posted by Andrew at 10:36 PM

March 02, 2003

New Alexandria library wants it all

The NY Times reports in "Online Library Wants It All, Every Book", that a new library in Alexandria, Egypt has set its aim to be much like the fabled library of ancient Alexandria, namely to have a copy every book know to existence.

Posted by Andrew at 02:02 AM

March 01, 2003

ODP public forum

I've just registered at the ODP Public Forum, a place where the public can interact with ODP editors.

Posted by Andrew at 09:39 PM
a