PCWorld’s Bogus Browsers Test


PCworld’s Nick Mediati has done one of those flawed  reviews that commits huge errors of omission. This occurs  when comparing the three most popular browsers – Google Chrome 21,  Microsoft Internet Explorer 9, and Mozilla Firefox 15. The review is so incomplete it does browser users a disservice.

If you take a look at the latest browser usage across both PC and mobile users :

Nick gets the top 3 browsers correct – but Apple Safari might be considered as well.

Nick then rates the chosen browser performance and states the case reasonably well  – “But many new Web apps and services rely heavily on HTML5 and JavaScript, so the browser makers have been spending a lot of development time making sure that their programs render such apps and services quickly and efficiently”.

However, then he uses only one JavaScript tests, does not do actual browser function testing,  completely ignores one of the best HTML5 tests in the industry; in short Nick short changes the review. For example, HTML5Test which you can easily run online yourself in 10 seconds by just going to the site at HTML5test.com – one can see the quality of testing done.
As you can see in the above screenshot, Chrome leads the parade, Firefox comes in second and IE9 lags way behind all the other browser vendors. This is typical of Microsoft IE where it has been behind the other browser in both performance  speed and standards support. Just checkout the comparison of CSS2 and CSS3 support at wikipedia where Firefox Gecko leads while Chrome Webkit trail  reasonably closely.

But more revealing is how long it has taken IE to come up to speed in CSS, DOM and JavaScript standards. Bill gates promised the Web community to implement all the HTML, DOM, CSS, and JavaScript standard back in 1998. IE9 is still the worst browser for  web standards compliance 14 years later. If you do any web development, you will know the need to devote 15-30% more time to get your web app working in the IE6 thru IE9 browser set This is important because IE browsers have a PC browser market share  of 56.3% as of 2011 year end – IE9=5.3%, IE8=32.4%,  IE7=7.0, IE6=10.2%.

As for speed performance tests using JavaScript read the controversies over the different tests here. This is why I much prefer live performance tests using the browsers on a local machine or using WebPageTest.org. Here is what a set of tests done locally on my machine showed versus the WebPageTest numbers:

Browser Speed Comparison
Test Google Chrome 21 Mozilla Firefox 15 Microsoft IE 9
Local – website loadtime 7.9 second 8.0 seconds 8.6 seconds
Local – reload page 2.2 seconds 2.2 seconds 2.5 seconds
WebTest – website loadtime 6.9 seconds 7.0 seconds 7.4 seconds
WebTest – reload page 1.9 seconds 1.8 seconds 2.0 seconds
Webtest: Webpagetest.org, five repeats for local tests

These tests do not include  one of the fastest browsers I have found to date – Maxathon. But they do show that IE9 consistently underperforms the other two browsers with Google just nudging out Firefox. This is confirmed in the Tom`s Hardware tests:

Chrome is the notable leader on both Windows 7 and MacOS for composite performance . However, IE did show fastest times for PageLoad in the Tom`s Hardware browser test. Chrome leads in the cumulative JavaScript test while IE9 lagged well behind.  Firefox leads in CSS performance with IE9 again bringing up the rear. IE9 lead in the HTML5 performance test – but this result is skewed because  IE9 is significantly behind in HTML5 features supported and these test only reflect the HTML5 tests IE9 can pass.

But on memory usage, Chrome`s leading  role is reversed. For memory use I defer to the following tests at ITWorld:


Unfortunately the version of the browsers is not the same as done in Nick’s tests; but they are from April of this year and show how memory usage is an important factor. It also reverses the order with Firefox using about 40% less memory than Chrome and IE9 again taking the last position.

Ease of Use

I won’t argue with the Nick’s assessment of basic ease of use – its a near tie. All of the browsers have stepped away from heavily iconbar loaded UIs [but you can add those on if you so desire]. The ability to go into full screen mode with F11 key, the ability to type in a search on the same line as the browser address, tab creation and manipulation – they are pretty close to identically the same with the exception of creating tab groups being unique in Firefox. But positioning and the shape of the refresh and bookmark buttons are needlessly different among all 3 browsers.

However another important aspect of ease of use is the availability of add-ons. And this is where Firefox easily out shines both Chrome and IE. Chrome’s set of add-ons has improved within the last year. And IE has one the best add-on management functions. But Firefox is chock full of useful add-ons so that users can customize Firefox for a range of important tasks. Again, Nick has omitted an important consideration for increasingly web oriented users.

Security and Reliability

Secunia reports the following on yearend 2011 for IE9
Advisories = 6 Criticality = High
Vulnerabilities 32
Secunia report the following on year end 2011 for FireFox
Advisories = 14 Criticality = predominately High
Vulnerabilities 96
Secunia report the following on year end 2011 for Chrome
Advisories = 28 Criticality = predominately High
Vulnerabilities = 321
So as a surprise IE9,  despite the recent 5 extremely critical vulnerabilities, has the best security performance for 2011, according to the latest full year reports available from Secunia. Chrome, to my surrprise, has the worst record. And Firefox is distinctly better than Chrome but still has more than double the rate of problems that IE had in 2011.

However in the 17 test BrowserScope  Security check Chrome leads the pack:

passing 16 of the 17 tests.

However, the security clincher for this user is the fact that Chrome Sandbox security and reliability protection really works. There have been  cases where security or reliability  has brought a tabs   session to an end but not all the other tabs. This happens 10-15 times per month but I run with many tabs open – currently 12. In contrast, I stay away from Safari and IE9 because multiple tabs and security-reliability problems can be complete disasters in these browsers.

Reliability

In his review, Nick did not discuss the issue of reliability of the browsers. Admittedly it is difficult getting data rather than anecdotaly evidence. But a popular test is opening 40tabs in a  browser. Being a heavy user of multi-tabs, I raely cross the 20 tabs line because of memory usage and deteriorating response time. But  anumber of sites report the test results with IE9 failing the test and Firefox and Chrome vying for best results.

Tom`s Hardware has a Page load Reliability Test which is a basic  indicator of reliability :

For this test, Opera emerges as the best browser with  a virtual tie at much worse scores for all 3  main browsers.  On MacOS opera again leads and Chrome is worst. But as a developer, I will tell you that IE9 and Opera lead in terms of number of bugs and workarounds required.

The Standards Conformance Test, again courtesy of Toms Hardware, measures developers frustrations with IE9. Measured against a JavaScript Conformance Test262,  HTML5Test as above, and the CSS3 Test, Chrome scores highest at 82% and IE9 is worst at 50%.

JDPower of car testing fame does a network  telecom service reliability report – but does not have one for browser reliability. So browser reliability tests will remain somewhat  adhoc.

Summary

As can be seen from this PC browser run through, Nick missed a lot of the details. First, the browsers reversed positions occasionally and having the MacOS results are enlightening – on almost every Web browser test  MacOS performed notably worse than Windows 7 . Now in the Tom`s Hardware tests where identically the same hardware was used, just switching the OS, this points to a serious browser failing on MacOS. For a PC World writer this is a big miss.

Just as important, by omitting the Conformance, Reliability, and sufficient Performance tests, Nick misses the sad state that is IE9. – it is way back of the pack  in these important tests. Yes, IE10 is testing out as a hot product but it will be restricted to Windows 8 and it still awaits rigorous test on Windows 8 against the other browsers which have some rumored surprises as well.  Bottom line, Nick should just post a link to Tom`s Hardware  comprehensive set of browser tests for the complete browser picture.