I haven't personally done the math but one of the articles I read (Scientific American?) mentioned that 1 error an hour makes you basically at the break even level. Higher error rates brought the house advantage _higher_ than non-counting methods. Is this incorrect?

It probably depends on the technique. Some counts are pretty easy to manage, like simply keeping a tally of the number of 10+ cards vs. total cards. It does take practice, but it's way easier than trying to keep a tally of every card you've seen in your head.

But yeah, it's really easy for casinos to keep a track of your betting pattern and kick you out if you're playing the game too well. That's why the really successful card counters play as groups with personas. They'll station a bunch of low key guys around the casino and have them maintain a steady stream of cheap wagers while they count. When a table gets hot they'll signal a flamboyant high stakes looking guy to sit down and play a few rounds.

The Casinos will eventually catch those guys too, simply by correlating who is sitting at what tables when the high stakes guy comes to play, but it takes a lot more work and they make out with a bunch more cash. Worse, because what they're doing isn't illegal the casio can't even deny them the winnings, only kick them out and plaster their faces all over the blacklists the casinos use.

Your middle paragraph was true, but is now outdated as of the last 15 years or so. This is indeed how the MIT Blackjack Team worked, as documented in books like Bringing Down The House.

However, casinos now know of and how to foil this strategy. The countermove is to prohibit entering a table while a shoe is in progress. Either the new player must wait until the deck reaches its established end point at the cut card, or the deck gets reshuffled immmediately upon the new player entering. Either way, an incoming player can't take advantage of any count.

I didn't notice this last time I was in Vegas...then again, the blackjack tables I was at were pretty low stakes. :)