Obviously, having proper firewall and anti-virus protections in place is a must, even with the most sophisticated of internet browsers, however, completely ignoring the security issues of certain software seems obtuse when the last statistic I heard (this is purely anecdotal, so you science purist, cover your ears) was that a fresh install of Windows without any updates (I want to say it was SP1), no firewall or antivirus protection, connected to the internet with the integrated, non-updated IE was completely incapacitated due to malicious code within minutes. Again, completely anecdotal and heresay, but it’s a stat that I wouldn’t be surprised at, were it to be true.

I’ve seen some interesting blogs concerning security amongst browsers, but being the psuedo-scientist that I am, I don’t have links for you. However, my point is not to bash this browser or that on security, but to raise the issue of, where’s the benchmark for that. It would be interesting to see a similar benchmark as is exhibited here, but include metrics for malicious code. I don’t know how the various security suites determine the malicious code counts that I’ve seen them use for marketing purposes, but using similar methodology, you could design a benchmark that tests real-world browsing experience, adding in the various arbitrary compliance tests, adding in security metrics and then throw all the statistical values into a blender, puree for 30 seconds, pour over hot-headed fanatics and let stew for several months. See what happens. My 60 Russian kopeks.

The overall idea here, as I understand it, was to perform a set of benchmarks on loading times for common/popular pages on common/popular, fresh, default browsers because there are so few “real world” benchmarks out there. It was NOT, from my understanding, stating that the data from the the said benchmarks is the END all and BE all of browser speed. I use Opera@USB on MANY different computers with MANY different connection types. I use portable firefox with a max of 62 connections per site. I can understand most of how this benchmark works but not all. Is the default # of CPP (connections per page) the same on all browsers? What kind of bandwidth do you have available? Thus, I would have to agree with some of the arguments on here.

This data and Benchmarking Scheme is a GREAT way for obtaining generalized load time on “default browsers”. Especially for web developers to get an idea of how fast their page will load on 10% of the web browsers out there without 6 tool bars and 30 add-ons.

Now a days you can so heavily modify most browsers, I just don’t know. For example my results would be heavily skewed because of the changes I’ve made to my browsers and the fact I have 2X(21MB)/sec cable connections bridged, one for up and one for down with strict load balancing.

I wouldn’t say your data/results are irrelevent because there are those out there that run a clean, mostly add-on free browser(s). These figures matter in that sense.

But if you where to tell a general user that, “this browser is faster at loading these sites.” I can guarantee you it won’t be for long, and if they install ANY new browser it will be faster then the one they “were” using.

“Opera is the winner, but not by far. Interestingly, the browsers all perform differently on different websites.”

THE BROWSERS ALL PERFORM DIFFERENTLY ON DIFFERENT WEBSITES.

This will always be the case and one browser will never truly be faster then another. We should stop looking at the results as “what browser you should use” and start looking at the results as what can you change to make the browser you like, are developing, are coding your page for, faster.

Love the work and this post was VERY will done. There are just SO many variables out there.

Every user would have to install and use (on a daily basis) each browser, as they normally would, THEN run the benchmark. Take those benchmark results and run them through and averager ( I know it’s not a word), then you might find the fastest browser, for YOU.

My suggestion would be to make an add-on for the popular browsers that saves the load time statistical values (ms to load, amount of data downloaded, speed of download/upload, flash blocker? Ad blocker? that is less data to download) and upload said data anonymously to an averager, then output real time averages to the add-on visually. That would give you a better idea of “real world” browser speed. And the geeks here know how much general users LOVE add-ons and junk. You could probably get a pretty awesome following with some great statistical data.

Anyways, I’m only really posting this to practice for my online class discussions for school. So what I say here really is worth 2 cents, maybe even a hay penny.

Keep up the good work!

BTW. I love opera and it is my default, but for some reason the built in spell checker considers “a” in the middle of “a” sentence as mis-spelled. It may be faster, but that doesn’t mean it’s better.

I lost my motivation to complete it because it really only serves to appease the statistics fanatics, and I earned only ~50 cents in advertisements even there have been 23k visitors to this page to date.

If you have little need for the standard deviation and other statistics, there isn’t a problem with adapting this one.

It would be a pretty cool tool for web developers who are working on optimizing a site’s performance/loading time.

-Steve

Good article,
\o/

I included Havenworks because I thought it would be a good stress test, and some websites do happen to look like that. Amazingly the Myspace page is not too far off in terms of loading times, so it is not totally unreasonable.

So for instance for IE using:
run IEXLORE.EXE -extoff

Else thrid party addons like flash or java could influence the test.

It is very strange that you included the havenworks page as it is a both a terrible webpage and it has a ridiculous impact on your teststats.