In 2006 I wrote a blog post about how CSS sucks. The post was popular and somewhat controversial. It’s been 15 years and the core of the problem remains. CSS has certainly improved but it’s still holding back designers and engineers.
I’ve reposted the blog post with the original comments from blogger as the original site was taken down.
Sept 25th, 2006 – CSS Sucks
CSS is certainly an improvement on plain old HTML but its limitations are staggering and the lack of industry support will continue to hold back designers for many years to come unless we begin to build and design something better.
- For all that CSS has been able to do it’s a technological failure. CSS just doesn’t work as expected. How can I say it’s a failure when millions of sites use it? CSS can be used to style basic text attributes but browsers aren’t consistent in how they use this technology. Even though there is a “standard” and some browsers partially adhere to the standard to truly be a useful standard you need two things: Predictability and Consistency. CSS has neither. Any designer who has tried to create a large and complex site using CSS will tell you that all popular browsers interpret the standard differently.
- CSS is ‘markup centric’ not ‘design centric.’ I have this idea that designers should spend more time designing great looking sites and less time fiddling around with markup tags and browser compatibility. When I say ‘markup centric’ I mean that every CSS design tool forces users to go into source code mode to create an attractive modern site. Many designers take pride in hand coding CSS. Tools for designers should be design centric. PDF/postscript is a good example of a design centric markup , (unfortunately not very suitable for the web.) Designers don’t argue about how to create semantically correct postscript tags they just create great designs using great design tools. CSS sucks because it forces designers to think about how to make it work technically rather than how to make it work from a design perspective.
- Why on earth do we think that cascading is a useful feature? The way that styles cascade from one level of layout to a deeper layout makes it difficult to figure out why a particular item is styled in a certain way. By contrast non-cascading style sheets would be equally powerful and more predictable. The cascading makes it harder to interpret the page for both the designer as well as the web-browser. In fact the complexities in cascading is one of the reasons why so many browsers screw up the standard. In theory cascading could save bandwidth but in practice it creates bloated documents to get around the cascading issues.
- The box-model is too simplistic. The high level idea of CSS is that you can create attractive pages using margin, border, padding and content attributes. While this is a nice theory, it’s primitive in its understanding of both layout problems and design. Highly developed design tools have layout engines that offer multiple layouts, non-rectangular margins, proportional layouts, dock-able layouts, table layouts, column layouts, etc. etc. It’ll be years before these features make it to CSS and many more before browsers implement them with any consistency. If browsers keep spending so much time on CSS they’ll have a well polished turd. Tools like Aldus Page Maker had better design tools, font tools and layout capabilities 10 years ago. This is because good design tools start with the design, not the markup.
- When writing software you learn what works and what doesn’t. You get new and better ideas and you throw away the old ones. This process of starting fresh is absent from the current CSS way of thinking. Each version of CSS builds on the previous one without acknowledging any fundamental flaws. CSS and its HTML sibling are the ultimate designs by committee. Any enhancements to CSS/HTML are piled on top of the old standards. This makes it progressively harder to create powerful, compatible and consistent browsers. This also makes it harder for designers to create sites that target the new platform because they are constantly trying to satisfy the compatibility with older browsers. Version compatibility has to be all or nothing. If you support V3 it has to be 100% supported and tested. Supporting some of the features actually makes things worse.
- There shouldn’t be multiple right answers for a visual design. The way CSS works there can be many ways to do the same thing. In fact there seem to be endless debates about the proper way to hack together trivial things like rounded corners. Rounded Corners? I mean really! Again I refer you to Aldus and even MS Word circa 1997. These features are not that hard to develop but getting them to work in a “standard way” seems to be all but impossible.
CSS captures styles not semantics or design intention. A design intention would be something like: “I want to balance these two columns” or perhaps “This text should line up with the logo image in the first column.” When designers do things like this:
They are capturing the style specifics not the design intention. Why 32 pixels? Why 40%? Perhaps the logo is 32px tall? Perhaps the other column is 60% wide? When the logo changes size or placement how will you know what styles to touch? There is a basic concept called parametric design that can be used to specify the parameters of the design. This concept helps embody the design intention as a set of rules that can then be preserved as the design changes. Even a very simple parametric design allows you to preserve design intention rather then hard coding sizes and dimensions.
- Design should be declarative not interpreted. Again CSS has to process a large number of rules before it can figure out where things are supposed to go. After these rules are interpreted this data is thrown out and each and every browser that opens up the web-page has to re-interpret the data. This is incredibly inefficient. First of it makes web-pages load very slowly. Even when you’re on a fast connection the browser can’t figure out where to place objects until the entire web-page has finished loading. Secondly this interpretation is very prone to errors. A declarative design isn’t open to as much interpretation allowing it both render quickly and consistently.
- CSS is a pain to work with. Take a look at some of the designs over at CSSZenGarden. The designs are both attractive and sophisticated. A good designer could take these designs and mock up similar designs in PhotoShop or Illustrator in a matter of hours but take the same designs and ask for it in CSS and it may take a couple days. Each time you make an edit to your CSS you have to refresh your browser to see what it’s actually going to look like. Then after you get one browser working you need to double back and get the other browsers working.
- If you can’t get consistency across browsers then you can’t rely on CSS to accurately and properly design your site. If you can’t get the site to look exactly the way you want on every single browser then how can you claim that CSS is a good design tool or even a success? The fact that there is no alternative to create attractive websites doesn’t make CSS a good tool. There are two ways to solve the problem. The first is to continue to hammer on standards and CSS asking for a better solution. This has been happening for the last 10 years and it just doesn’t work. The alternative is to realize that CSS is flawed in it’s intrinsic design and begin to ask the questions of how could you do it better?
Archived comments from the original posting