The scripts run on all sites. And yes, they work. They have occasionally broken in the past, but we try to keep a sharp eye on them and folks do tend to notice and let us know when they fall apart.
The purpose of the script is to identify patterns of fraud: one account being used purely to inflate or deflate another account's reputation. It isn't intended to stop someone from gaining reputation too quickly: there are other (less dramatic) limits in place for this purpose.
What usually trips folks up is the assumption that what they're seeing on their profile is an accurate reflection of what is happening on the site. But of course, you only see the votes that affect your account - you don't see who is casting them or where they're coming from or what else the voters are doing.
I've found that I tend to collect a lot of upvotes on days when I'm particularly visible: on Stack Overflow that tends to mean a featured meta post, but on smaller sites it can be something as simple as posting a well-received answer that sits on the front page for a bit: suddenly, folks are aware that you exist and know something - and some of them will go looking to see if you know anything else of interest to them.
The exact workings of the voting fraud detection scripts are kept secret to discourage abuse. Serial voting is one of the criteria, but not the only one - and while false positive/negative results are always possible, in general the behavior is not so naive as folks commonly make it out to be. If you're concerned that someone is voting for you blindly or maliciously (up-votes or down-votes), feel free to flag one of the posts involved and ask a moderator to check up on it - but normally, you shouldn't need to be concerned about this either way.