In case anyone is wondering what he is talking about (I was), I found this article this morning, but didn't have time to post it until now http://www.bankrate.com/brm/news/mortgages/20060126a1.asp
Not sure if you were referring to the OP or to my comment, but that article doesn't address the question/topic of the OP. The article is specifically about 'buying points', which means paying the bank to give you a lower interest rate on your loan. Many lenders also offer the option to 'sell points' (I don't think anyone uses that term, it's just a way to contrast to the buying option), which means that the bank pays you to take a higher interest rate on the loan.
While the concept of a calculating a breakeven point applies equally to 'selling points', the OP was asking about a specific strategy that I've found some mortgage brokers try to push. The strategy is predicated on the idea that historically, in the very short term (say 6 months) mortgage rates remain stable. So if the 'no points' rate you would get today is 4%, and you take $5000 from the bank in exchange for a 5% rate, in 6 months you would refinance back to (hopefully still available) 4%. As long as the refi costs are less than $5000, you'll have done better than just going with 4% to begin with. You'd also have to factor in paying the higher monthly mortgage payment for 6 months, but that shouldn't be a huge issue.
As you can pretty easily see, the profit/loss compared with just taking the no points option to begin with comes solely from the movement of mortgage rates during the 6 months between your original mortgage and your refi. And as I already mentioned, this is a terrible way to bet on interest rates. The only thing people should take into account when deciding which rate on a particular rate sheet they should select (positive or negative points, or no points) is the breakeven calc.