DDoS attacks are more then just blocking access to a site.
There's two basic kinds of DDoS attacks, one is a flood of network traffic, the other abuses slow heavy resources on the website itself.
If you have a page that is demanding on server resources to generate (heavy SQL queries without caching/etc), then you can send many requests to the page to overwhelm whatever slow backend service it is using - lagging or crashing anything else that relies on that service (locking up the database). This is typically what happens when a site is overwhelmed from "normal" traffic. It can be mitigated with pro-active filtering and caching on the application side.
The much more common form of DDoS is a pure network flood of half open TCP connections or page requests as fast as possible. These types of floods do not wait or care about a server response, and typically will even change their reply route so the source computer doesn't receive the response traffic (Sometimes the source was another target similar to email "Joe Jobbing"/backscatter)
Basically a series of incoming requests that say "HI GIMME THE WEB PAGE" or "HEY I NEED A TCP SOCKET HEY I NEED A TCP SOCKET HEY I NEED A TCP SOCKET" and start asking for the page again before they even get the first response. This behavior is NOTHING like mashing F5 on a website, it's much more damaging. There's various techniques to detect false source addresses and either ignore or "tarpit" the sender so they get stuck on the first request without opening a second.
When the latter form of DDoS attack is used, it has the potential to not just take down the target website, but every computer on its network segment as your upstream gateway router's connection tracking buckets get overwhelmed and it starts dropping any packet routing though. Modern gateways and firewalls are much more robust about handling this (most DDoS attacks these days are fairly hard to maintain beyond a few hours for this reason - Firewall heuristics have come a long way in the last 6 years or so).
ANY DDoS attack at all wastes the time of people at data centers and people running the servers (NOT the same people putting the content on the sites that you hate) as they lock things down and do forensics.
Also, for those proud "hackers", it's pretty much the least skilled way to attack a site unless you wrote the software behind the botnet yourself. Almost all of them re-purpose some existing bot and spam links on IRC/etc until they get enough drones to do a fancier version of the Windows 95 ping of death. The client impact is similar to spamming long lines of text in your online game of choice so people can't use the chat until the GM comes and bans you. Devastating.
If you want to have fun dealing with kiddie attacks, check out IRC networks sometime. Being the places where bored teenage computer nerds hang out and play with their bot control channels, they attract more such attacks then the majority of "large" websites do. The tiny IRC network [/chat] that I run a server on would deal with several random DDoS attacks a month for several years. This is before the Escapist even existed, 3 months or so of having an IRC network with a few large channels (anime fansubbers and browser game communities in our case) engendered more DDoS attempts then the entire lifetime of our sites. Such behavior was so prevalent that the majority of data centers offering dedicated server hosting would outright ban any servers participating in an IRC network due to the amount of attacks they drew. As mentioned above though, firewalls are better at detecting bad traffic these days, so it's not nearly as common.
In conclusion. [http://underhanded.org/lookatme.gif]