IE8 loads pages faster, but not fast enough, in Microsoft test recreation

Microsoft suggested that its reader use what it described as comScore's Top 25 Web sites; actually, they're selections from 25 of the Top 50 Properties. We used as many of those sites as we could -- specifically, 23 of them -- although two would not work for purposes of our tests today. Blogger.com was generally down, and MySpace uses JavaScript in such a way as to kick out our JavaScript-based stopwatch. So in their place, we used two other selections from comScore's Top 50: TV.com (from CBS Interactive) and Weather.com.

The other 23 sites are as follows: Google (both US and China), Yahoo, Windows Live, MSN, YouTube, Microsoft.com (sort of a theme here), Wikipedia, Facebook, QQ.com, Baidu.com, WordPress, eBay, Sina.com.cn, Mozilla.com, Adobe.com, AOL, Amazon.com, Apple.com, soso.com, xunlei.com, 163.com, and Ask.com. We also threw in a "#26" Web site just for the hell of it: betanews.com.

For the Internet Explorer 7 figures, since IE7 and IE8 cannot co-exist on the same machine, we set up IE7 on a separate, similarly configured Vista-based system. And for each site we loaded, we made sure to clean the browser cache each time, so the page contents loaded completely from the Web, not from memory.

Betanews tests Wednesday of Microsoft Internet Explorer 8 versus Mozilla Firefox 3.0.7 and all of the major experimental Web browsers from Mozilla and others, using browsers run on a clean Windows Vista-based Virtual PC, and borrowing some but not all the methodology suggested by Microsoft, reveal IE8 to load popular Web sites' home pages 73% faster than IE7. But against Firefox 3.0.7, IE8 was 5.19% slower in loading our battery of 26 sites.

But that's not to say IE8 was "dog slow." In fact, the slowest browser overall in Betanews tests -- believe it or not -- was the same Apple Safari 4 that blew away both IE and Firefox in JavaScript tests a few weeks ago. Safari 4 was nearly 49% slower at loading Web pages than IE8, and Opera 10 Alpha was just over 23% slower than IE8.

Beating IE8 handily, however, were Firefox 3.1 Beta 3 with 23.83% more speed than IE8, and the clear champion of this event, the new Beta 2 edition of Google Chrome, with more than double IE8's speed: 107.27% more speed, as it turns out.

If we were to use an index method for gauging relative load times, using IE7 as the basis (1.00), then IE8 would have an index score of 1.73 -- which translates into, 73% faster than IE7. Firefox 3.0.7 scores a 1.98 on that scale, which is higher than Safari 4 Beta's 1.84 and Opera 10 Alpha's 1.70. But then Firefox 3.1 Beta 3 scores an impressive 2.27, and Google Chrome 2 Beta takes the win with a colossal 3.33.

We've seen this happen with many a high school thesis: The author has a brilliant theme in mind, but in researching the subject matter, he ends up uncovering facts that contradict the theme or make it seem out of place. Microsoft's white paper started out by getting the reader in the mood to feel something powerful, but ended up with as close to flat as you can get in terms of comparison.

So to mask that little hole in the logic, the white paper concluded with more of a warning to readers than an invitation: "As this analysis has illustrated, there are many factors involved in benchmarking and many steps involved in attaining valid results. This document should provide the information needed to perform benchmarking tests and isolate extraneous factors to allow for a true analysis of browser performance." Well, maybe we don't all have the patience to go setting up video cameras and time-indexing 150 different snapshots...which is why we opted to use computing tools instead. So although Microsoft's hidden message may be, "Kids, don't try this at home," in the real world, users typically find a way.