GIS data doesn't have a look before it takes on a visual representation. James Fee points out all you need to know in a worthwhile piece here.
For mere historical reasons as it seems, the naked payload data on Shape-files always needs to be accompanied by sidecar files for storing symbology, metadata and other human readable information. As clean it is programmatically to have content and display separated, the more trouble it generates downstream if GIS numerology needs to take on gestalt and visualisation. By the way, Gestalt in German literally means "shape" or "figure".
When we started United Maps, one of the first and naive questions I posted to the ESRI folks was about a CSS-like method of styling large amounts of GIS-data, whole countries on a 1:2000 scale and 21 zoom levels. I could have asked for their latest extraterrestial communication a their eyes went funny. Such a thing would not exist, the answer was, at least not as I envisioned it.
Things may have changed into the right direction: there is a project to convert CSS into Mapnik acting as "an abstraction layer and preprocessor that converts special, CSS-like syntax into Mapnik-compatible style definitions". As James Fee puts it:
CSS is very easy to edit and and work with and Mapnik is great at
rendering cartography (and could easily be supported by desktop
clients). CSS gets stored as a sidecar file to the dataset and the
rich cartography can be read with Mapnik to render the maps as the
author wishes the maps to look.
Pierre from the czech site Tydlevidle comments:
ESRI’s lyr is of course a direct serialization of all the associated
objects, which is after all very flexible and supports symbols,
renderers and objects other which are not yet implemented or come from
third-party developers.
Takes me into a loop - how can we compare the ol' ESRI workhorse approach to the CSS trick pony that Sean Gillies points to here, animating Homer Simpson only using CSS? What's the limits of CSS, how quickly does it render large scale, highly detailed maps, does it work well on mobile browsers? The really long thread of 60 comments beyond the original article carries a lot more.
Jeffrey Warren from the MIT Media Lab's Design Ecology group runs cartagen.org, ...
... a set of tools for mapping, enabling users to view and
configure live streams of geographic data in a dynamic, personally
relevant way. These tools helps users to analyze and view collected and
shared geographic and temporal data from multiple sources. The
framework uses vector-based, context-sensitive drawing methods to
describe data, not merely in terms of lines and polygons, but also with
adaptive use of color, movement, and projection.
As Jeffrey demonstrated at WhereCamp, CSS can be put to work:

All done with GSS-Styles like this:
body: {
fillStyle: "#fff",
lineWidth: 0,
},
node: {
fillStyle: "#ddd",
strokeStyle: "#090",
lineWidth: 0,
radius: 1
},
way: {
strokeStyle: function() {
return "rgba(1,1,1,0.7)"
//return color_from_string(this.user)
},
strokeStyle: "#ccc",
lineWidth: 3
},
leisure: {
fillStyle: "#2a2",
lineWidth: 3,
strokeStyle: "#181"
},
park: {
fillStyle: "#2a2",
lineWidth: 3,
strokeStyle: "#181",
pattern: "/images/pattern-water.gif"
},
Hey, how smart is that!