Bonum Certa Men Certa

1996 article that aged well: "This page optimized for" (arguing with customers)

posted by Roy Schestowitz on Oct 06, 2023,
updated Oct 06, 2023

Editorial note: This article was written a long time ago. It is merely being republished here. Sadly the article has become relevant again lately.

by Jahn Rentmeister


This page dates back from September 1996, and a lot has happened since then. Some changes have been made to this text, especially in the section of examples, where I tried to reflect changes on the example sites. From my point of view, the state of the Web has improved a lot since I wrote this article, so you might find the discussion exaggerated or unnecessary (although I think it still isn't).



You probably have seen it hundreds of times: You follow a link and and arrive at a site's opening page. The people who designed that page were probably paid for doing so, and one should expect them to try to raise your interest in the site and gain it as wide a readership as possible.

And the first thing that renders on your screen is "Best viewed with Netscape Navigator and 800x600", in big, bright yellow letters on a white background. You can't read it, because you have disabled image loading, so you manually load images to read the message.

The first time this happens, you actually believe it: "There must be a reason why Netscape Navigator on 800x600 pixels is best for this site", you think, resizing your window. Usually, if you know a bit about HTML, the second time, you don't believe it anymore.

There are some reasons why you shouldn't optimize your site for a particular setup, let alone demand that your visitors adjust their setups.

Table of contents:

Why this is a bad idea

There is a saying in marketing (at least in Germany): No one has ever won an argument with a customer. Visitors to your site are your customers, and, if you tell them their configurations are wrong, you are arguing.

Basically, you are telling people "You are using the wrong system". This is rude. People are using the systems they use for a good reason. If you demand the newest Netscape version, you are telling NextStep users to go play elsewhere. If you demand certain plug-ins, you are telling all people who aren't using Microsoft Windows to bugger off. If you use extensive graphics without ALT texts, your site isn't a pleasure at all to people connected via modem. Can you imagine what your boss would say if you were being rude to customers? Most of the time, the site owner is the customer of the site designer, and visitors are customers of the site owner, not of the site designer. Thus, there is no real need for the site designer to care for the visitors: He argues with other people's customers. For the customers, however, it looks like the site owner is arguing. And, as customers usually do, they will react accordingly.

Most of the time, the whole warning message isn't really necessary. It seems as if so-called Web designers think they aren't being trendy if they don't use the latest extensions, and, to show how cool they are, they put this warning on the front page. Some even put it on every page of the site. On many of these sites, there is nothing that couldn't be viewed with any other graphical browser. Others could even be viewed with Lynx, if only the "designer" hadn't been too lazy to include ALT texts. To state this clearly: If you say "This site optimized for XYZ", this doesn't say "I am hip and use the latest in technology"; it says "I am either too lazy or too incompetent to design a proper site, but I think it looks cool with XYZ using my setup". I have seen lots of sites. Probably, you have seen lots of them, too. I have seen not a single site that says "for XYZ browsers only" or "You need XYZ for this site" where this made sense. Have you seen such a site? There are sites that say "If you want to hear the audio (or view the video), you need XYZ", which makes sense, but sites that simply claim "You need" either aren't worth visiting at all or could quite easily be changed to be enjoyable with other browsers.

Even if a site can be viewed only with a certain browser: The only thing this "XYZ only" warning can accomplish is to keep people off your site. This isn't what you want, is it? If so, just shut down your site. Usually, people won't install the software that you demand just to view your site. If I visit your site, I won't even adjust my font size. You can argue that most other people aren't like me, but you can't prove that, and, even more important, you are arguing with a potential customer.

Last but not least: If your site cannot be viewed with text-only browsers, this means blind people cannot read it. There are pages that simply cannot be prepared for text-only browsers. There is also information that cannot be prepared for text-only browsers (a city plan, for example). But I bet your information isn't of that type.

A real-world browser configuration

If you "optimize" your site for a particular browser, this really means it is "optimized" for a particular browser with a particular setup. In order to give you an idea of how diverse users and their setups out there are, I will describe some of my own browser configurations here. If you think you already know that this is a wide world, just skip it.

The configuration that many so-called "Web designers" assume is: latest version Netscape on a 800x600 true-colour screen under Windows 95 with ShockWave, TrueAudio, Video for Windows and Quicktime VR installed, font size 12 for Times New Roman and 10 for Courier New, image loading enabled, accepting cookies, connected via a T1 line. What they usually say is "This site best viewed with Netscape and 800x600". Note that my first configuration fits this description.

My first configuration is: Netscape 2.02 on a Windows 3.1 system on a 256-colour 800x600 screen, with the browser window being about 400x600 pixels, because I have it cover only half of the screen. Image loading is disabled, and, most of the time, document foreground and background colours are overridden by my default (black on grey with blue links). Java and JavaScript are disabled because they aren't secure in this browser. The font size is 16, both for fixed (Courier New) and proportional (Times New Roman) fonts. The browser accepts cookies. There are no plug-ins installed, and I won't install any. The computer is connected to the Internet via a LAN, which is connected via a laser line to a machine that is connected to the Internet. Thus, available bandwidth depends on the weather: With fog or a dove sitting at the wrong place, there might be no bandwidth at all. (I still can send e-mail, though.)

My second configuration is: Lynx 2.4.2 on an AIX rs/6000 system, connected via vt220 emulation, black text on white background, vi keys enabled.

My third configuration is: Netscape 3.0 on a Sun Solaris system, connected via a low-memory X-terminal (will have to boot if things get too colour-intensive), Java enabled, JavaScript disabled. I don't know the resolution, and the browser window covers the right half of the screen. Font size is 18, image loading is usually on.

I don't install browser software myself on any of those systems, so I simply cannot "Download a REAL browser [tm]". There are sites where I do load images and accept their colour settings, but I have seen sites where you'd better not do that (see examples section).

Several ways for users to make your optimizing pointless

Font size

Like me, many people customize their font sizes to fit their needs. Usually, the standard font sizes are just too small. Therefore, saying a page is "optimized for 800x600" is absolutely pointless if the page consists mostly of text, or if image loading is disabled.

People either change their font sizes to be legible in a fully maximized window, or they change it to be legible in their normal window size, which is 400x600 for me. Your page can look totally different with different font sizes and different window sizes. Your information can become illegible.

Customizing font sizes will especially lead to problems if you use frames or tables with absolute width or height: If these contain text, it won't be legible. There are sites that have frames with absolute dimensions that aren't scrollable and contain text. Now, if my standard text size is bigger than assumed, I won't be able to read that text. This leads to the conclusion that you should specify at most one dimension in terms of pixels, if the table or the frame contains text. Every frame that contains text should be scrollable. Even then, long words lead to problems.

Also, not knowing the font size of your target system leads to problems with images that consist of text. Many "Web designers" seem to think that text has to look exactly as they think it should. Of course, then, on high-end systems with screen resolutions beyond 1600x1200, the text inside of the images is illegible. This can become a real problem for people who browse on X Window System terminals. To cite a victim:

"Many sites will provide buttons and other simple text 'widgets' as images. So, while I can scale up my fonts if I have trouble reading the text, I'm still stuck with the small text in the images. It's really annoying, since they're always text, just dressed up to look more 'professional'."
-- Ian Tester (iteste01@postoffice.csu.edu.au, 94024831@postoffice.csu.edu.au before 2. Feb. 97), who browses using a 1600x1200 screen.

Image loading disabled

Images take time to load. Most images on the WWW aren't worth it. Therefore, many users disable image autoloading. (If they want to wait, they can go to the bus stop, instead of your site, since waiting for the bus doesn't cost them money.) Your wonderful-looking site may look like crap, then. Or it may still look good. It depends on you, and there is no reason to stick with the crap. To start with, it is usually a bad idea to render text as images. "index page" is only a few bytes, but, if you put it into an image, it's a few KB, plus a few bytes for the IMG tag. If you choose to do it in spite of my advice, at least include the ALT attribute. It's easy, and maybe you will make some of your potential customers happy by so doing.

If you omit ALT attributes for images, you are hurting not only Lynx users who cannot display inline images, but also yourself: Those who have image loading disabled will now need to load the image, because there is no ALT attribute. The time they spend waiting could be better spent looking at the information that you provide. This is OK if the image is part of that information, but, if the image is only your logo or some text put into GIF format, an ALT attribute would have made better use of customer time.

In addition to the ALT text for users who have image loading disabled, you should also specify a background colour in your body tag that is close to the colour of your background image (if any).

Override document colours

Yes, there are users out there who stick to their own colours for viewing your page. And yes, they do have good reasons to do so: They might be using a B/W screen, or they might be frequently visiting sites that use blue colour for normal text (you don't, do you?), or they might be colour blind, e.g., they cannot distinguish green from red, or they are tired of background images that make the text hard to read (as I am). They are still your visitors, your customers; and you should make your pages delightful for them, because you want their attention, their money, their feedback. This means:

Disabling Java/JavaScript

If your site relies on Java or JavaScript for essential purposes, you have to be aware that many users disable Java and JavaScript for security reasons. By now, there are no known security risks in the newest Java implementations, but many people use older browsers, don't trust their up-to-date browsers, or don't want to spend valuable processing time for scrolling tickers and animated advertising.

As an additional note, if you use JavaScript, make sure your code is hidden in comments and doesn't contain "--" or ">", so older browsers won't display the code. It isn't a good idea to make your page illegible for some users just so that other users can have a "cool" scrolling ticker in their status bar that they don't want. Most JavaScript that you find on the Web isn't worth the time it takes to download.

Not accepting cookies

This isn't a matter of display, but of functionality: There exist not only browsers that don't support cookies, but also users with browsers that do support cookies who deliberately disable them. There are several ways to do so. One way is (for Netscape under UNIX) to link the cookies file to /dev/null. Users might also configure their proxy servers to strip all cookies. So, if your shopping cart application depends on cookies, perhaps I should switch to your competitor, who uses CGI with GET methods, validates his HTML according to the HTML 3.2 DTD, and supports even Lynx. [Note: as of 1998-04-24, there's HTML 4.0. The argument still holds. -jr]

What people will tell you

OK, so you have listened to me, and you are nearly convinced, but the people who have designed that site (or your site) keep telling you it really should be that way, because:

"95% of our hits are from Netscape users, anyway."

Are they? Let's for the moment assume that server logs indicate this.

Now, of these "hits", how many have image loading turned on? Those users who don't turn it on will generate fewer hits, because they don't "hit" your server for the images. To have meaningful numbers, you need numbers about users, not hits. You aren't trying to sell your products to hits, are you?

Of those users, how many are using Netscape Navigator, and how many are using Microsoft Internet Explorer (which claims to be "Mozilla", too, just as Netscape does)?

Even if you count users, and even if you carefully distinguish browsers, you cannot trust your stats: Of all your users, how many are using proxy servers? If a hundred people using the old AOL browser (which has no table support) look at your page in one day, you will only have one hit for all of them. If every single one of 4,000 students with Internet access at the University of Münster visits your site, there will be only one hit, too. You have no chance to reliably count your users, unless you fiddle with HTTP headers to "prohibit" caching. If you do so, your pages will take a lot longer to load, and you can be absolutely sure that not all of those 4,000 students will visit your site.

Now, let's assume you took all of this into account, and, still, 95% of your users are using Netscape Navigator or Microsoft Internet Explorer. Therefore, you are told, it is perfectly OK to optimize the page. Let's for the moment assume that you can afford to lose 5% of your potential customers: After all, there are about 40 million of them out there, so who cares about a few more or less?

These 95% aren't a homogeneous class of users: About 30% of them (this share is indicated by server stats, as stated by several people on comp.infosystems.www.authoring.html for different sites) will disable image loading. You don't understand why they do that, because you paid a lot of money for the images. But people are diverse: I can't understand why the other 70% don't do the same. Actually, some people hit the Stop button on every page to stop image loading, because they don't want to wait that long. So, if you don't want to lose that 30%, you have to make your site legible for Netscape and Internet Explorer even without images. Knowing this, have you ever looked at your site with image autoloading disabled? Or did you browse it only on a demo system, where the pages were read from local disk?

In the following paragraphs, I will give some numbers that are rough estimates by me ("wild guesses"). If you think they aren't accurate, I agree: If you can provide more-accurate numbers, I'll be happy to correct mine. I'm sure the revised argument will still hold. Don't quote me on the numbers; quote me on the fact that there are significant shares of users with systems of the various kinds:

Of the 95% you decided to aim for, at least 10% don't have a screen resolution of 800x600 or bigger. Another 5% have that resolution, but don't usually have their browser window set to cover the whole screen. Another ~30% have more than 800x600, and another group have set font sizes so they can read it conveniently, so optimizing for their screen size doesn't work, either. This leaves you with far less than 50% of your "Netscape and compatible"-users, if you optimize for 800x600 screen resolution. You probably cannot afford that.

You are told that you need some nifty graphics and scripting.

As we have decided, 95% of people use Netscape and MSIE anyway, and they both support this cool plug-in for real-time audio and video. You've tested this from a local disk file, and you were quite impressed. But, being suspicious, you want to know how many of your users can actually benefit from this. "No problem, just include a link to the download site for the software, and all will be able to use it."

Well, all that run MS-Windows, actually, because it isn't yet available for other systems. But hey, that's about 90% of the people, so you still have about 80% left. Unfortunately, many of them use modems at speeds of 28,800 bps or even (gasp!) 14,400 bps to connect to the Internet. And no, it doesn't help that you are connected with 45 Mbps -- the whole audio and video has to be sent over the customer's modem connection. Naturally, not all of your potential customers want to hear your company jingle, so many just won't bother, and will simply ignore your pricey site flavoring. I will be among them, because I have no sound card on one computer, no MS-Windows drivers for the card on another, and the speakers connected to the Sun station I occasionally use are located in an entirely different building from my X-terminal's, so I won't hear your company jingle except over the PC speaker: I'd better sing it myself. Maybe some of your customers will actually download your multimedia enhancements. After all, it is essential information -- isn't it? It is probably exactly what you wanted to provide to your customers, and, if the Internet didn't exist, you'd have to send it to them via normal mail, as you did with your catalogs in the past.

Of "your" 95%, some are tired of watching those "Get a REAL browser" messages whenever a new megabyte-munching killer application is released, and therefore use only the latest in technology. Unfortunately, if your pages were optimized to a particular version of a particular browser, they might be unable to view it, because browsers change. Your pages that looked so nice may fail to work on a new release. And your JavaScript for Netscape 2.02 that contained the frame hack gives an error message on Netscape 3.0.

Fortunately, the same people whom you paid to build these pages will gladly fix them, if you provide the necessary funds. They will even re-optimize your site for the new browser "generation" (they prefer to call it "generation", where others simply say "release"), and they will do so at almost no additional charge. Of course, having this much to do writing HTML, they don't find the time to view it with other browsers or setups than their own, let alone validate it against a DTD. But hey, 95% of all people are using that browser, anyway.

Note that we haven't yet discussed the possibility that some of your customers might use an outdated browser version. I for one use Netscape 2.02 (or Lynx 2.4.2, which is equally outdated -- get Lynx 2.6 NOW!). [Note: as of 1998-04-24, Lynx 2.8 is out. -jr]

"Then we would have to provide and maintain a second version."

This isn't true. There are ALT-attributes to the IMG tag, and there are ALT-attributes to client-side image maps. I see only two reasons why someone would want to maintain two versions:

  1. To provide a no-frames version for users who have frame-capable browsers, but don't want to have frames.
  2. To provide a text-only version that displays the same way as the normal version, but loads faster, because there are fewer (or no) IMG tags, no JavaScript scripts, and no FRAMESET tags in it.

If you don't know how to write correct, scalable, working HTML, maybe it is time for you to download a REAL HTML reference, if not for yourself, then for your WWW designers.

Examples of "optimized" sites

There are two kinds of examples:

1. There are examples of sites that provide useful information, but this information isn't as accessible as it could be. To each of those sites, I have sent e-mail messages stating the problems I see. This isn't a "Netscape Hall of Shame" or something like that. It is a list of sites that I wish had used proper techniques to produce customer delight, because I as a customer would like to be delighted.

2. There are also examples in here of sites that really need to demand a certain browser characteristic to provide their information. This isn't an extensive list: It is only a list of examples that I use to prove my point.

There are no examples in here that demand a special browser in order to be "cool" without the information requiring it. I.e., there are no typical "Netscape Hall of Shame" entries here.

A site that said "best viewed with Netscape".

With some simple changes, it could become a wonderful site for all graphical browsers, even if image-loading is off, and even for Lynx: These changes are to include ALT-attributes and to provide decent link and background colours, and some minor other things on some pages. This site provides real, interesting information. It is http://www.earlyamerica.com/. I don't know whether they now use ALT attributes. If they do, feel free to view it with Lynx; the information is worth it (Imagine: I even enabled image loading for it!), and there is no reason to exclude Lynx users from it. You will miss the audio if you lack the appropriate software.

A site that said "Best viewed with Netscape and 800x600".

This warning was the first thing that rendered of the page. The site used blue for normal text and dark grey for links -- on a light grey background. Thank God I have 256 colours, because on a 16-colour screen I wouldn't have been able to read the text. That would have been a pity, because it was a good laugh. This site was designed by so-called "professionals". That was a good laugh, too. (Note to protect the innocent: Both design and designer have changed since this was written.) From my point of view, it was best viewed with "Lynx -dump > /dev/null", but that's just my personal opinion.

They reacted quite quickly, once notified of the problem. The grey text on grey background was removed. Other things stayed, but it was the site owner's decision, not mine.

Well, maybe you will like to visit http://www.eisenbahn.com/. To be honest, I had hoped this site would stay as it was when I first visited it, because it was such a wonderful example. It didn't, so you won't see what I saw: This is an example of site designers listening to complaints and acting on them. So, if there is something you dislike about a site, tell the people who make it. If they want their information to come through, they'll listen, because you're their target. If you don't tell people what's wrong, they won't be able to fix it.

A page that requires Java for good reasons

This page doesn't bear a big red warning message "You need Java to view this page". It does so only if your browser lacks Java support. This is an interactive city plan. It is real useful information that cannot be provided in HTML in any proper fashion. Though it could be provided using image maps, the performance would be too bad for it to be usable. It is at http://www.koeln.org/mapview/ .

Conclusion

OK, if you aren't convinced yet: Imagine that I am your customer. I have half an hour of time, and I plan on visiting your site, during that. I won't tell you which browser I use and what the preferences are. Are you going to welcome me in a pleasant manner, or are you going to tell me that I am not cool enough to be your customer? Also, I have only $20 left. I can spend them on telephone and online costs, or I can spend them on your products. Which alternative do you prefer?

Even if you don't need me as a customer: If your Internet presence convinces me of the advantages of your product, I might recommend it to others -- but I need to have information for that. I won't recommend your product because your Web site looked so nice. Not even if your product is a software package for creating Web sites.

Other Recent Techrights' Posts

World Wide Slop
If it quacks like a duck...
[Meme] Driver Issues
Where do you want to drive today?
Another Dose of Fake 'Articles' About Linux
Don't give visibility to the nonsense of Microsoft
 
Once Again Linux Foundation Makes It Clear It's Being 'Absorbed' by Bill Gates
Linux Foundation devotes about 2% of its budget to Linux
IBM is a Boys' Club
If IBM collapsed, the Red Hat engineers who work on GNU and Linux would simply work elsewhere (on the same projects)
The Miserable State of GAFAM
Looking for government handouts
Microsoft is Acting Like a Company That's Running Out of Money (But Still Pretends to be Wealthy in Order to Attract or Retain Shareholders)
Azure has had mass layoffs every year since 2020, yet Microsoft keeps telling shareholders that "clown computing" is growing
Dr. Andy Farnell's Article on Societal Disorganised Attachment and the Role of Social Control Media
The article is quite long and typos were still being fixed as recently as last night
Smear Alert: Linus Torvalds Asking for Better Commit Messages Makes Linus a (Grammar) Nazi
Maybe the "mainstream media" is looking for clickbait or maybe it's actively looking to make a scandal - a phony controversy with which to make the job of coordinating Linux unpleasant
Gemini Links 09/10/2024: Climate Doom and Clagrange
Links for the day
Over at Tux Machines...
GNU/Linux news for the past day
IRC Proceedings: Tuesday, October 08, 2024
IRC logs for Tuesday, October 08, 2024
Dr. Andy Farnell's Article on Why Passwords Still Rock
"Seven for a secret never to be told"
The Problem Isn't That New Cars Use Electricity But That They Use Too Many Bits of Electronics
"...and proprietary software wrapped in proprietary APIs and protocols all without a modicum of compartmentalisation," an associate adds
We're Turning 18 in 30 Days
30 days from now the site turns 18
GNOME Foundation Says It's Nearly Broke (Again), It's Getting Rid of More People (Only Women Get the Boot), and It Will Improve Communications and Transparency Even Though It Secretly Ousts People From the GNOME Foundation Board (for Secret Reasons)
It only talks about this months later (under strict gag orders, only public shaming of a person)
Links 08/10/2024: Australian Fines for Twitter (X), Fake Patent Courts Still Not Scuttled
Links for the day
Gemini Links 08/10/2024: Guilt by Association, Workers vs Owners
Links for the day
Links 08/10/2024: War Updates, Samsung's Layoffs, and Gemini
Links for the day
Links 08/10/2024: Microsoft Deleting Office Documents Instead of Saving Them, "Threads Still Sucks"
Links for the day
gemini.techrights.org and techrights.org (Same Server, Not the Same Protocol)
We're reminding readers that everything in this site is fully accessible via gemini.techrights.org in Gemini Protocol
X Has Axed Itself. This is Great News and Further Affirmation of Everything We've Said About Social Control Media.
Don't waste any more time on social control media
Over at Tux Machines...
GNU/Linux news for the past day
IRC Proceedings: Monday, October 07, 2024
IRC logs for Monday, October 07, 2024
Gemini Links 08/10/2024: Contingency Begets Complexity, Playing With Bezier Curves
Links for the day
Almost Half the Web Users Connecting to Your Site Are Using Linux
almost 1 in 2 Web-connected devices runs Android and about 2% run "proper" GNU/Linux
The Web Has Severe Amnesia Problems, But We Still Remember How Gilberto Gil Promoted Free Software in Brazil
The Digital Tipping Point (DTP) is years behind us now
Synthesised Voices Aren't a New Technology (the Hype Might Be, They Call It "Hey Hi" Now)
I still consider this an extension of the "hey hi" (AI) hype
LLM Hype is Already Descending, Apple Stopped Investing in the Money Furnace
Wall Street is a perverse force in the technology market, incentivising the most harmful (and mostly useless) things
Change Control and What Will Come After Git (If That's Still Possible at All)
It would be wrong to believe (at least misguided) Git can be a "standard" skill 30 or 50 years from now.
On the Web, HTTPS Has Actually Become a Privacy Problem (Broadcasting Usage/Access to the All-Seeing CA Eye). Geminispace Doesn't Have This Problem.
Down to 23 capsules: the rapid demise of Certificate Authority (CA) Let's Encrypt in Geminispace
Links 07/10/2024: Politics, Education, Wars, Financial Crunch
Links for the day
Munich Was Having Real Difficulties Moving From GNU/Linux to Windows
How many are still using GNU/Linux?
Links 07/10/2024:China’s 'Deflation' (Price Decreases), Brazil Still Bars Twitter ("X")
Links for the day
Links 07/10/2024: "Creative Computing" Turns 50, Long War in Middle East Turns 1
Links for the day
Gemini Links 07/10/2024: Luck and Dishonesty, Gaming Getting Worse
Links for the day
Over at Tux Machines...
GNU/Linux news for the past day
IRC Proceedings: Sunday, October 06, 2024
IRC logs for Sunday, October 06, 2024