I read a sociology paper recently. It was very clear about what answers it was trying to find and whether or not each point was or was not significant. Many points were significant; many were not. Even if every point had come back as not being statistically significant, that doesn't mean it has nothing useful to say. It just says that Group X is not very different from the Control group for whatever parameter was being measured. (Although if every point came back negative, something would probably be seriously wrong with either the initial assumptions or the statistical analysis itself...)

And somehow I missed the fact that you mentioned spread-spectrum yourself. Apparently I fail.

It sounds like you are describing not broadband communication, but spread-spectrum communication. While you may or may not be technically correct in whatever field you are referring to, the term 'broadband' can mean a number of things based on context (or in this case, whether people are allowed to give you less and call it the same thing).

If the count's reliability can be accurately estimated, the data is not useless; the margin of error changes.

I'll grant that free and automatic updates to everything sounds good on paper (especially to security folks), but sometimes when they add new features, they redesign the entire interface and you have to get used to it all over again, whether you like the new version or not. And they didn't even ask or give you advance warning. Oops.

With local apps you may have the added hassle of having to update things yourself, but at least you get to decide whether an update is better than the older version or not. Added features does not necessarily mean better functionality. There's a reason http://oldversion.com/ exists.

Yes, because no one has yet invented DNS caching.
...
Oh wait.


Yes, those servers are important, but they don't handle every single DNS lookup directly. They wouldn't be able to withstand that. Taking down those servers would only inconvenience people by temporarily preventing them from contacting domains that weren't in their DNS server's cache.

A more effective target would be to attack IXPs and prevent the traffic from flowing between Internet carriers. There are quite a lot of those, though, and it would be exceptionally difficult (bordering on impossible) to pull off.

This is an excellent point. I love it when people point out what should be obvious.

The fuck? I go to UT Knoxville, and it's already suffering from budget cuts (they cut an entire department this year), so this pisses me right the hell off just for that alone. At least I'm not on their computer network any more...