Although I've been writing about Internet software for years, I've never done a formal review of a Web browser. Why not? The reason are myriad. For one thing, with a piece of software, unlike a book or a movie, an individual impression isn't particularly relevant. I can like a movie for intensely personal reasons, for instance, and not worry about the fact that no one else may share my opinion. However, with a software review, I try to come to a relatively objective conclusion that I expect at least some other people to share.
But Web browsers are screwy. They're all screwy, and it's driven me crazy over the last few years. Whenever I talk to people about which Web browser they use and why, they inevitably say something different than the previous person I spoke with. The only consensus seems to come from the fact that there are a limited number of Web browsers, so by default, a certain number of people must agree in the end. If we had more Web browsers, opinions would vary even more widely.
Here's the problem. If I were to review a Web browser, I'd be concerned with three primary issues: features, performance, and stability. Normally, I don't think stability should be a major part of a software review, but I also expect most software to be relatively stable, which hasn't been true of many versions of the common Web browsers. But how do you evaluate stability fairly? Everyone seems to have a different experience. Perhaps Internet Explorer 3.01 was rock solid for this person, but Netscape Navigator 4.04 crashed constantly. And the next person found Netscape Navigator 4.05 fairly reliable, whereas Internet Explorer 4.0 bombed out of sight all the time. I even still hear from people who swear by much older versions of the main Web browsers. Of course, perceived stability is also related to hardware, extensions, and even the type of Internet connection. In short, I've never been able to make any reasonable generalizations about Web browser stability.
So what about performance? Consider all the variables involved in serious performance testing. You must test different types of pages repeatedly, then test different types of Internet connections, clocking both initial load times and loads from cache. Of course, specific settings and system configurations make a difference, and testing anything on the Internet brings in variables related to the performance of your machine, your connection, your ISP's connection, the Internet backbone, the remote server's Internet connection, the load on the remote server, and more. Frankly, it's a nightmare. I've tried to do performance testing in the past, but I gave up when I realized my test methodologies had so many holes that they were practically transparent.
Okay, so what about features? We're definitely on firmer ground here, but even still, one person's feature can be another's bug. For instance, I've become quite fond of Internet Explorer's AutoComplete feature when typing URLs in the Address field. It takes a little getting used to before you figure out what it's likely to do with any given text string, but I've been able to internalize its quirks. TidBITS Technical Editor Geoff Duncan, on the other hand, hates that feature and turned it off instantly. Similarly, I like the fact that Internet Explorer provides a history feature that works across windows and sessions, in contrast to Netscape Navigator's history feature, which is window-specific (it goes away when you close a window) and session-specific (it goes away when you quit). Geoff, ever the contrarian and in this case hailing from one of the outer planets, somehow claims that window-specific history is better. Put it this way. Arguing with Geoff is almost as annoying as arguing with me, but luckily, I seldom have to argue with myself.
My Current Browser -- Here, then, is what I'm going to do. I'm going to tell you what I use for my main Web browser at the moment, and why. If you disagree with me because my choice crashes constantly on your system, or the alternative is faster in your experience, I won't argue with you about it.
That said, my current favorite browser is the just-released Internet Explorer 4.01 that we wrote about last week in TidBITS-430. I've been playing with early releases for a while now, and version 4.01 can be compared to the previous 4.0 release with three words: "Faster, More Stable." Oh, there's also support for Apple's ColorSync technology in specially created JPEG images, but frankly, I don't much care about that, although graphic designers concerned about how their images look on the Web might be.
In the testing Tonya and I did in January when she wrote about Internet Explorer 4.0 for MacWEEK, there was no question that Internet Explorer 4.0 was slower than Netscape Navigator (most any version). The reason was quite simple - Internet Explorer added DHTML (dynamic HTML) to its page rendering engine, but the developers didn't have time to tweak the code. Thus, every page loaded slower than it should have (and slower even than in Internet Explorer PR1, a preview release that lacked DHTML). And, although I found that both Internet Explorer and Netscape Navigator crash with roughly equivalent frequencies on my Mac, that was still far too often, especially since there were certain actions which were guaranteed to destroy Internet Explorer 4.0 (try searching for "Smith" at Amazon.com).
With Internet Explorer 4.01, though, performance has improved significantly. For the reasons mentioned above, I haven't tried to do any performance testing, but almost everything about using it feels faster, especially returning to previously loaded pages. It's certainly more comparable to Netscape Navigator now, which is welcome. Even more important, in several weeks of using the beta builds, I've had only two crashes, and I was able to exit gracefully out of both using MacsBug.
The reason I put up with Internet Explorer 4.0's mediocre performance and stability before was its feature set, which was sufficiently different from Netscape Navigator's so as to make the speed hit worthwhile. My favorite features include the following.
The feature I use the most is a shortcut that enables me to open a link in a new window by Command-clicking the link. Both Internet Explorer and Netscape Navigator provide an Open Link in New Window command in the pop-up menu that appears if you click and hold on a link, but Command-clicking is much faster. In part, this feature works well for me because I use two large monitors and Internet Explorer opens the new windows on my other monitor, enabling me to keep reading the original page while the new ones load. Netscape Navigator tends to open new windows on top of my existing one, making it harder to keep reading while pages load. I especially like opening multiple windows when working with search results pages. Ideally, I'd use Internet Explorer's Search tab, but since it retrieves its list of search engines from Microsoft's generally overloaded Web site, it's too slow to load. And, when it does load, it doesn't include AltaVista, the search engine I prefer.
I've become fond of Internet Explorer's History feature because it can track the last 1,000 pages I've visited (1,000 pages generally covers about two weeks of normal usage for me). You can open the History window and search for text in page names and URLs, although I'd like to see Microsoft take the feature a step further and search the full text of cached pages. Connectix's SurfExpress does this, and it can be quite handy. Netscape Navigator's History feature frustrates me because as I said above, it's window and session specific, which means it seldom contains a site I want to revisit.
Related to the History feature is Internet Explorer's AutoComplete feature, which guesses at URLs or page names as you type. I type a lot of URLs, since I generally know where I want to go, and I've found that the utility of AutoComplete requires that I watch the Address field as I type. If it starts guessing wrong, I just keep typing. For instance, say I want to go to Apple's home page. I prefer to type "apple" and have Internet Explorer figure out "www.apple.com". However, if I've been to the Apple Software Updates page recently, it will instead guess that page. So, I type "apple", notice that Internet Explorer has guessed wrong, then press Delete to delete the rest of the guess and Return to have it find and load Apple's home page. Geoff complains rightly that Internet Explorer overrides proper name resolution as well as Open Transport's search domains. So, when Geoff types "www", he expects his Web browser to use the search domain first and fill in "quibble.com" at the end. Because Internet Explorer ignores the search domains, it instead tries to load "www.www.com."
I generally like Internet Explorer's Download Manager window, which makes it easy to see what I've downloaded and how large it was (since people always ask me to include download sizes in TidBITS articles). Unfortunately, the silly thing won't close when the download is done, so I'm constantly closing it manually or hiding it behind my main browser window. I'd like to see Microsoft close that window when the download is done, perhaps providing an option to alert me via a sound that I could set to a voice alert (no modal dialogs, please!).
Although I keep relatively few bookmarks, I need to visit a few Web pages frequently, and I appreciate the fact that Internet Explorer lets me place those bookmarks on the Favorites Bar for easy clicking, rather than choosing them from the Favorites menu. Various utilities exist for modifying Netscape Navigator's Directory Buttons, although this sort of customization should be available in the program. Skylar Stein's $5 shareware Navigator Button Editor 1.3.4 and Joel Reed's freeware ResEdit instructions claim to work with the latest versions of Netscape Navigator and Communicator.
<ftp://ftphqx.info.apple.com/Apple.Support.Area /Apple_SW_Updates/US/Macintosh/Utilities /ResEdit_2.1.3.sea.hqx>
Finally, as part of our editorial process surrounding the creation of TidBITS Updates, one of us generally writes the update and saves the file in a folder served by Personal Web Sharing. Then that person calls another editor and asks them to read the update. It turns out that editing a paragraph of text in a Web browser is best done with a large font size, and the Larger and Smaller buttons that can appear in Internet Explorer's toolbar simplify this process immensely (make sure to use the Compatible Plus toolbar setting in the Browser Display pane of Internet Explorer's Preferences dialog). Although Netscape Navigator provides control over font sizes in the Fonts pane of its Preferences dialog, there's no quick way to change the size for a single page.
I've seen little brand loyalty with Web browsers, and I certainly don't have much. If the next version of Netscape Navigator were to offer these features and more, I'd switch in a second. I keep both Web browsers available anyway, and I periodically use both while testing Web page designs and whatnot. What I'm waiting for though, is to see what independent developers do with the Netscape Navigator source code. I assume that everyone has a favorite feature they'd like to see implemented in a Web browser, and Netscape's move of opening up the source code for Netscape Navigator might prove tremendously interesting as a way of obtaining an innovative new feature set, plus potentially improved performance and stability.
In short, what I want from a Web browser is performance, stability, and features that make my use of the Web more efficient. For the moment and for my uses, that means using Internet Explorer 4.01.